Compare commits

..

2 Commits

Author SHA1 Message Date
Josh Brower affede7f0a Rename 'ScanLNK' to 'ScanLnk' in YAML config 2026-04-20 10:01:10 -04:00
Josh Brower 97366c0496 Rename 'ScanLNK' to 'ScanLnk' in defaults.yaml 2026-04-20 10:00:29 -04:00
88 changed files with 217 additions and 4588 deletions
-12
View File
@@ -1,12 +0,0 @@
# Copyright Security Onion Solutions LLC and/or licensed to Security Onion Solutions LLC under one
# or more contributor license agreements. Licensed under the Elastic License 2.0 as shown at
# https://securityonion.net/license; you may not use this file except in compliance with the
# Elastic License 2.0.
# Per-minion Telegraf Postgres credentials. so-telegraf-cred on the manager is
# the single writer; it mutates /opt/so/saltstack/local/pillar/telegraf/creds.sls
# under flock. Pillar_roots order (local before default) means the populated
# copy shadows this default on any real grid; this file exists so the pillar
# key is always defined on fresh installs and when no minions have creds yet.
telegraf:
postgres_creds: {}
-21
View File
@@ -17,7 +17,6 @@ base:
- sensoroni.adv_sensoroni
- telegraf.soc_telegraf
- telegraf.adv_telegraf
- telegraf.creds
- versionlock.soc_versionlock
- versionlock.adv_versionlock
- soc.license
@@ -39,9 +38,6 @@ base:
{% if salt['file.file_exists']('/opt/so/saltstack/local/pillar/elasticsearch/auth.sls') %}
- elasticsearch.auth
{% endif %}
{% if salt['file.file_exists']('/opt/so/saltstack/local/pillar/postgres/auth.sls') %}
- postgres.auth
{% endif %}
{% if salt['file.file_exists']('/opt/so/saltstack/local/pillar/kibana/secrets.sls') %}
- kibana.secrets
{% endif %}
@@ -64,8 +60,6 @@ base:
- redis.adv_redis
- influxdb.soc_influxdb
- influxdb.adv_influxdb
- postgres.soc_postgres
- postgres.adv_postgres
- elasticsearch.nodes
- elasticsearch.soc_elasticsearch
- elasticsearch.adv_elasticsearch
@@ -106,9 +100,6 @@ base:
{% if salt['file.file_exists']('/opt/so/saltstack/local/pillar/elasticsearch/auth.sls') %}
- elasticsearch.auth
{% endif %}
{% if salt['file.file_exists']('/opt/so/saltstack/local/pillar/postgres/auth.sls') %}
- postgres.auth
{% endif %}
{% if salt['file.file_exists']('/opt/so/saltstack/local/pillar/kibana/secrets.sls') %}
- kibana.secrets
{% endif %}
@@ -134,8 +125,6 @@ base:
- redis.adv_redis
- influxdb.soc_influxdb
- influxdb.adv_influxdb
- postgres.soc_postgres
- postgres.adv_postgres
- backup.soc_backup
- backup.adv_backup
- zeek.soc_zeek
@@ -155,9 +144,6 @@ base:
{% if salt['file.file_exists']('/opt/so/saltstack/local/pillar/elasticsearch/auth.sls') %}
- elasticsearch.auth
{% endif %}
{% if salt['file.file_exists']('/opt/so/saltstack/local/pillar/postgres/auth.sls') %}
- postgres.auth
{% endif %}
{% if salt['file.file_exists']('/opt/so/saltstack/local/pillar/kibana/secrets.sls') %}
- kibana.secrets
{% endif %}
@@ -172,8 +158,6 @@ base:
- redis.adv_redis
- influxdb.soc_influxdb
- influxdb.adv_influxdb
- postgres.soc_postgres
- postgres.adv_postgres
- elasticsearch.nodes
- elasticsearch.soc_elasticsearch
- elasticsearch.adv_elasticsearch
@@ -273,9 +257,6 @@ base:
{% if salt['file.file_exists']('/opt/so/saltstack/local/pillar/elasticsearch/auth.sls') %}
- elasticsearch.auth
{% endif %}
{% if salt['file.file_exists']('/opt/so/saltstack/local/pillar/postgres/auth.sls') %}
- postgres.auth
{% endif %}
{% if salt['file.file_exists']('/opt/so/saltstack/local/pillar/kibana/secrets.sls') %}
- kibana.secrets
{% endif %}
@@ -301,8 +282,6 @@ base:
- redis.adv_redis
- influxdb.soc_influxdb
- influxdb.adv_influxdb
- postgres.soc_postgres
- postgres.adv_postgres
- zeek.soc_zeek
- zeek.adv_zeek
- bpf.soc_bpf
+1 -5
View File
@@ -29,14 +29,10 @@
'manager',
'nginx',
'influxdb',
'postgres',
'postgres.auth',
'soc',
'kratos',
'hydra',
'elasticfleet',
'elasticfleet.manager',
'elasticsearch.cluster',
'elastic-fleet-package-registry',
'utility'
] %}
@@ -81,7 +77,7 @@
),
'so-heavynode': (
sensor_states +
['elasticagent', 'elasticsearch', 'elasticsearch.cluster', 'logstash', 'redis', 'nginx']
['elasticagent', 'elasticsearch', 'logstash', 'redis', 'nginx']
),
'so-idh': (
['idh']
-1
View File
@@ -32,4 +32,3 @@ so_config_backup:
- daymonth: '*'
- month: '*'
- dayweek: '*'
-14
View File
@@ -54,20 +54,6 @@ x509_signing_policies:
- extendedKeyUsage: serverAuth
- days_valid: 820
- copypath: /etc/pki/issued_certs/
postgres:
- minions: '*'
- signing_private_key: /etc/pki/ca.key
- signing_cert: /etc/pki/ca.crt
- C: US
- ST: Utah
- L: Salt Lake City
- basicConstraints: "critical CA:false"
- keyUsage: "critical keyEncipherment"
- subjectKeyIdentifier: hash
- authorityKeyIdentifier: keyid,issuer:always
- extendedKeyUsage: serverAuth
- days_valid: 820
- copypath: /etc/pki/issued_certs/
elasticfleet:
- minions: '*'
- signing_private_key: /etc/pki/ca.key
+4 -25
View File
@@ -31,7 +31,6 @@ container_list() {
"so-hydra"
"so-nginx"
"so-pcaptools"
"so-postgres"
"so-soc"
"so-suricata"
"so-telegraf"
@@ -56,7 +55,6 @@ container_list() {
"so-logstash"
"so-nginx"
"so-pcaptools"
"so-postgres"
"so-redis"
"so-soc"
"so-strelka-backend"
@@ -164,8 +162,8 @@ update_docker_containers() {
# Pull down the trusted docker image
run_check_net_err \
"docker pull $CONTAINER_REGISTRY/$IMAGEREPO/$image" \
"Could not pull $image, please ensure connectivity to $CONTAINER_REGISTRY" >> "$LOG_FILE" 2>&1
"Could not pull $image, please ensure connectivity to $CONTAINER_REGISTRY" >> "$LOG_FILE" 2>&1
# Get signature
run_check_net_err \
"curl --retry 5 --retry-delay 60 -A '$CURLTYPE/$CURRENTVERSION/$OS/$(uname -r)' $sig_url --output $SIGNPATH/$image.sig" \
@@ -188,27 +186,8 @@ update_docker_containers() {
if [ -z "$HOSTNAME" ]; then
HOSTNAME=$(hostname)
fi
docker tag $CONTAINER_REGISTRY/$IMAGEREPO/$image $HOSTNAME:5000/$IMAGEREPO/$image >> "$LOG_FILE" 2>&1 || {
echo "Unable to tag $image" >> "$LOG_FILE" 2>&1
exit 1
}
# Push to the embedded registry via a registry-to-registry copy. Avoids
# `docker push`, which on Docker 29.x with the containerd image store
# represents freshly-pulled images as an index whose layer content
# isn't reachable through the push path. The local `docker tag` above
# is preserved so so-image-pull's `:5000` existence check still works.
# Pin to the digest already gpg-verified above so we copy exactly the
# bytes we approved.
local VERIFIED_REF
VERIFIED_REF=$(echo "$DOCKERINSPECT" | jq -r ".[0].RepoDigests[] | select(. | contains(\"$CONTAINER_REGISTRY\"))" | head -n 1)
if [ -z "$VERIFIED_REF" ] || [ "$VERIFIED_REF" = "null" ]; then
echo "Unable to determine verified digest for $image" >> "$LOG_FILE" 2>&1
exit 1
fi
docker buildx imagetools create --tag $HOSTNAME:5000/$IMAGEREPO/$image "$VERIFIED_REF" >> "$LOG_FILE" 2>&1 || {
echo "Unable to copy $image to embedded registry" >> "$LOG_FILE" 2>&1
exit 1
}
docker tag $CONTAINER_REGISTRY/$IMAGEREPO/$image $HOSTNAME:5000/$IMAGEREPO/$image >> "$LOG_FILE" 2>&1
docker push $HOSTNAME:5000/$IMAGEREPO/$image >> "$LOG_FILE" 2>&1
fi
else
echo "There is a problem downloading the $image image. Details: " >> "$LOG_FILE" 2>&1
+1 -1
View File
@@ -227,7 +227,7 @@ if [[ $EXCLUDE_KNOWN_ERRORS == 'Y' ]]; then
EXCLUDED_ERRORS="$EXCLUDED_ERRORS|from NIC checksum offloading" # zeek reporter.log
EXCLUDED_ERRORS="$EXCLUDED_ERRORS|marked for removal" # docker container getting recycled
EXCLUDED_ERRORS="$EXCLUDED_ERRORS|tcp 127.0.0.1:6791: bind: address already in use" # so-elastic-fleet agent restarting. Seen starting w/ 8.18.8 https://github.com/elastic/kibana/issues/201459
EXCLUDED_ERRORS="$EXCLUDED_ERRORS|TransformTask\] \[logs-(tychon|aws_billing|microsoft_defender_endpoint|armis|o365_metrics|microsoft_sentinel|snyk|cyera|island_browser).*user so_kibana lacks the required permissions \[(logs|metrics)-\1" # Known issue with integrations starting transform jobs that are explicitly not allowed to start as a system user. This error should not be seen on fresh ES 9.3.3 installs or after SO 3.1.0 with soups addition of check_transform_health_and_reauthorize()
EXCLUDED_ERRORS="$EXCLUDED_ERRORS|TransformTask\] \[logs-(tychon|aws_billing|microsoft_defender_endpoint).*user so_kibana lacks the required permissions \[logs-\1" # Known issue with 3 integrations using kibana_system role vs creating unique api creds with proper permissions.
EXCLUDED_ERRORS="$EXCLUDED_ERRORS|manifest unknown" # appears in so-dockerregistry log for so-tcpreplay following docker upgrade to 29.2.1-1
fi
+2 -9
View File
@@ -9,7 +9,7 @@
. /usr/sbin/so-common
software_raid=("SOSMN" "SOSMN-DE02" "SOSSNNV" "SOSSNNV-DE02" "SOS10k-DE02" "SOS10KNV" "SOS10KNV-DE02" "SOS10KNV-DE02" "SOS2000-DE02" "SOS-GOFAST-LT-DE02" "SOS-GOFAST-MD-DE02" "SOS-GOFAST-HV-DE02" "HVGUEST")
software_raid=("SOSMN" "SOSMN-DE02" "SOSSNNV" "SOSSNNV-DE02" "SOS10k-DE02" "SOS10KNV" "SOS10KNV-DE02" "SOS10KNV-DE02" "SOS2000-DE02" "SOS-GOFAST-LT-DE02" "SOS-GOFAST-MD-DE02" "SOS-GOFAST-HV-DE02")
hardware_raid=("SOS1000" "SOS1000F" "SOSSN7200" "SOS5000" "SOS4000")
{%- if salt['grains.get']('sosmodel', '') %}
@@ -87,11 +87,6 @@ check_boss_raid() {
}
check_software_raid() {
if [[ ! -f /proc/mdstat ]]; then
SWRAID=0
return
fi
SWRC=$(grep "_" /proc/mdstat)
if [[ -n $SWRC ]]; then
# RAID is failed in some way
@@ -112,9 +107,7 @@ if [[ "$is_hwraid" == "true" ]]; then
fi
if [[ "$is_softwareraid" == "true" ]]; then
check_software_raid
if [ "$model" != "HVGUEST" ]; then
check_boss_raid
fi
check_boss_raid
fi
sum=$(($SWRAID + $BOSSRAID + $HWRAID))
-8
View File
@@ -237,11 +237,3 @@ docker:
extra_hosts: []
extra_env: []
ulimits: []
'so-postgres':
final_octet: 47
port_bindings:
- 0.0.0.0:5432:5432
custom_bind_mounts: []
extra_hosts: []
extra_env: []
ulimits: []
@@ -51,16 +51,6 @@ so-elastic-fleet-package-registry:
- {{ ULIMIT.name }}={{ ULIMIT.soft }}:{{ ULIMIT.hard }}
{% endfor %}
{% endif %}
wait_for_so-elastic-fleet-package-registry:
http.wait_for_successful_query:
- name: "http://localhost:8080/health"
- status: 200
- wait_for: 300
- request_interval: 15
- require:
- docker_container: so-elastic-fleet-package-registry
delete_so-elastic-fleet-package-registry_so-status.disabled:
file.uncomment:
- name: /opt/so/conf/so-status/so-status.conf
+103 -4
View File
@@ -17,17 +17,65 @@ include:
- logstash.ssl
- elasticfleet.config
- elasticfleet.sostatus
{%- if GLOBALS.role != "so-fleet" %}
- elasticfleet.manager
{%- endif %}
{% if GLOBALS.role != "so-fleet" %}
{% if grains.role not in ['so-fleet'] %}
# Wait for Elasticsearch to be ready - no reason to try running Elastic Fleet server if ES is not ready
wait_for_elasticsearch_elasticfleet:
cmd.run:
- name: so-elasticsearch-wait
{% endif %}
# If enabled, automatically update Fleet Logstash Outputs
{% if ELASTICFLEETMERGED.config.server.enable_auto_configuration and grains.role not in ['so-import', 'so-eval', 'so-fleet'] %}
so-elastic-fleet-auto-configure-logstash-outputs:
cmd.run:
- name: /usr/sbin/so-elastic-fleet-outputs-update
- retry:
attempts: 4
interval: 30
{# Separate from above in order to catch elasticfleet-logstash.crt changes and force update to fleet output policy #}
so-elastic-fleet-auto-configure-logstash-outputs-force:
cmd.run:
- name: /usr/sbin/so-elastic-fleet-outputs-update --certs
- retry:
attempts: 4
interval: 30
- onchanges:
- x509: etc_elasticfleet_logstash_crt
- x509: elasticfleet_kafka_crt
{% endif %}
# If enabled, automatically update Fleet Server URLs & ES Connection
{% if ELASTICFLEETMERGED.config.server.enable_auto_configuration and grains.role not in ['so-fleet'] %}
so-elastic-fleet-auto-configure-server-urls:
cmd.run:
- name: /usr/sbin/so-elastic-fleet-urls-update
- retry:
attempts: 4
interval: 30
{% endif %}
# Automatically update Fleet Server Elasticsearch URLs & Agent Artifact URLs
{% if grains.role not in ['so-fleet'] %}
so-elastic-fleet-auto-configure-elasticsearch-urls:
cmd.run:
- name: /usr/sbin/so-elastic-fleet-es-url-update
- retry:
attempts: 4
interval: 30
so-elastic-fleet-auto-configure-artifact-urls:
cmd.run:
- name: /usr/sbin/so-elastic-fleet-artifacts-url-update
- retry:
attempts: 4
interval: 30
{% endif %}
# Sync Elastic Agent artifacts to Fleet Node
{% if grains.role in ['so-fleet'] %}
elasticagent_syncartifacts:
file.recurse:
- name: /nsm/elastic-fleet/artifacts/beats
@@ -101,6 +149,57 @@ so-elastic-fleet:
- x509: etc_elasticfleet_crt
{% endif %}
{% if GLOBALS.role != "so-fleet" %}
so-elastic-fleet-package-statefile:
file.managed:
- name: /opt/so/state/elastic_fleet_packages.txt
- contents: {{ELASTICFLEETMERGED.packages}}
so-elastic-fleet-package-upgrade:
cmd.run:
- name: /usr/sbin/so-elastic-fleet-package-upgrade
- retry:
attempts: 3
interval: 10
- onchanges:
- file: /opt/so/state/elastic_fleet_packages.txt
so-elastic-fleet-integrations:
cmd.run:
- name: /usr/sbin/so-elastic-fleet-integration-policy-load
- retry:
attempts: 3
interval: 10
so-elastic-agent-grid-upgrade:
cmd.run:
- name: /usr/sbin/so-elastic-agent-grid-upgrade
- retry:
attempts: 12
interval: 5
so-elastic-fleet-integration-upgrade:
cmd.run:
- name: /usr/sbin/so-elastic-fleet-integration-upgrade
- retry:
attempts: 3
interval: 10
{# Optional integrations script doesn't need the retries like so-elastic-fleet-integration-upgrade which loads the default integrations #}
so-elastic-fleet-addon-integrations:
cmd.run:
- name: /usr/sbin/so-elastic-fleet-optional-integrations-load
{% if ELASTICFLEETMERGED.config.defend_filters.enable_auto_configuration %}
so-elastic-defend-manage-filters-file-watch:
cmd.run:
- name: python3 /sbin/so-elastic-defend-manage-filters.py -c /opt/so/conf/elasticsearch/curl.config -d /opt/so/conf/elastic-fleet/defend-exclusions/disabled-filters.yaml -i /nsm/securityonion-resources/event_filters/ -i /opt/so/conf/elastic-fleet/defend-exclusions/rulesets/custom-filters/ &>> /opt/so/log/elasticfleet/elastic-defend-manage-filters.log
- onchanges:
- file: elasticdefendcustom
- file: elasticdefenddisabled
{% endif %}
{% endif %}
delete_so-elastic-fleet_so-status.disabled:
file.uncomment:
- name: /opt/so/conf/so-status/so-status.conf
-101
View File
@@ -1,101 +0,0 @@
# Copyright Security Onion Solutions LLC and/or licensed to Security Onion Solutions LLC under one
# or more contributor license agreements. Licensed under the Elastic License 2.0 as shown at
# https://securityonion.net/license; you may not use this file except in compliance with the
# Elastic License 2.0.
{% from 'allowed_states.map.jinja' import allowed_states %}
{% if sls in allowed_states %}
{% from 'elasticfleet/map.jinja' import ELASTICFLEETMERGED %}
include:
- elasticfleet.config
# If enabled, automatically update Fleet Logstash Outputs
{% if ELASTICFLEETMERGED.config.server.enable_auto_configuration and grains.role not in ['so-import', 'so-eval'] %}
so-elastic-fleet-auto-configure-logstash-outputs:
cmd.run:
- name: /usr/sbin/so-elastic-fleet-outputs-update
- retry:
attempts: 4
interval: 30
{% endif %}
# If enabled, automatically update Fleet Server URLs & ES Connection
so-elastic-fleet-auto-configure-server-urls:
cmd.run:
- name: /usr/sbin/so-elastic-fleet-urls-update
- retry:
attempts: 4
interval: 30
# Automatically update Fleet Server Elasticsearch URLs & Agent Artifact URLs
so-elastic-fleet-auto-configure-elasticsearch-urls:
cmd.run:
- name: /usr/sbin/so-elastic-fleet-es-url-update
- retry:
attempts: 4
interval: 30
so-elastic-fleet-auto-configure-artifact-urls:
cmd.run:
- name: /usr/sbin/so-elastic-fleet-artifacts-url-update
- retry:
attempts: 4
interval: 30
so-elastic-fleet-package-statefile:
file.managed:
- name: /opt/so/state/elastic_fleet_packages.txt
- contents: {{ELASTICFLEETMERGED.packages}}
so-elastic-fleet-package-upgrade:
cmd.run:
- name: /usr/sbin/so-elastic-fleet-package-upgrade
- retry:
attempts: 3
interval: 10
- onchanges:
- file: /opt/so/state/elastic_fleet_packages.txt
so-elastic-fleet-integrations:
cmd.run:
- name: /usr/sbin/so-elastic-fleet-integration-policy-load
- retry:
attempts: 3
interval: 10
so-elastic-agent-grid-upgrade:
cmd.run:
- name: /usr/sbin/so-elastic-agent-grid-upgrade
- retry:
attempts: 12
interval: 5
so-elastic-fleet-integration-upgrade:
cmd.run:
- name: /usr/sbin/so-elastic-fleet-integration-upgrade
- retry:
attempts: 3
interval: 10
{# Optional integrations script doesn't need the retries like so-elastic-fleet-integration-upgrade which loads the default integrations #}
so-elastic-fleet-addon-integrations:
cmd.run:
- name: /usr/sbin/so-elastic-fleet-optional-integrations-load
{% if ELASTICFLEETMERGED.config.defend_filters.enable_auto_configuration %}
so-elastic-defend-manage-filters-file-watch:
cmd.run:
- name: python3 /sbin/so-elastic-defend-manage-filters.py -c /opt/so/conf/elasticsearch/curl.config -d /opt/so/conf/elastic-fleet/defend-exclusions/disabled-filters.yaml -i /nsm/securityonion-resources/event_filters/ -i /opt/so/conf/elastic-fleet/defend-exclusions/rulesets/custom-filters/ &>> /opt/so/log/elasticfleet/elastic-defend-manage-filters.log
- onchanges:
- file: elasticdefendcustom
- file: elasticdefenddisabled
{% endif %}
{% else %}
{{sls}}_state_not_allowed:
test.fail_without_changes:
- name: {{sls}}_state_not_allowed
{% endif %}
@@ -240,7 +240,7 @@ elastic_fleet_policy_create() {
--arg DESC "$DESC" \
--arg TIMEOUT $TIMEOUT \
--arg FLEETSERVER "$FLEETSERVER" \
'{"name": $NAME,"id":$NAME,"description":$DESC,"namespace":"default","monitoring_enabled":["logs"],"inactivity_timeout":$TIMEOUT,"has_fleet_server":$FLEETSERVER,"advanced_settings":{"agent_logging_level": "warning"}}'
'{"name": $NAME,"id":$NAME,"description":$DESC,"namespace":"default","monitoring_enabled":["logs"],"inactivity_timeout":$TIMEOUT,"has_fleet_server":$FLEETSERVER}'
)
# Create Fleet Policy
if ! fleet_api "agent_policies" -XPOST -H 'kbn-xsrf: true' -H 'Content-Type: application/json' -d "$JSON_STRING"; then
@@ -5,12 +5,11 @@
# this file except in compliance with the Elastic License 2.0.
. /usr/sbin/so-common
. /usr/sbin/so-elastic-fleet-common
{%- import_yaml 'elasticsearch/defaults.yaml' as ELASTICSEARCHDEFAULTS %}
{%- import_yaml 'elasticfleet/defaults.yaml' as ELASTICFLEETDEFAULTS %}
{# Optionally override Elasticsearch version for Elastic Agent patch releases #}
{%- if ELASTICFLEETDEFAULTS.elasticfleet.patch_version is defined %}
{%- do ELASTICSEARCHDEFAULTS.elasticsearch.update({'version': ELASTICFLEETDEFAULTS.elasticfleet.patch_version}) %}
{%- do ELASTICSEARCHDEFAULTS.update({'elasticsearch': {'version': ELASTICFLEETDEFAULTS.elasticfleet.patch_version}}) %}
{%- endif %}
# Only run on Managers
@@ -20,10 +19,13 @@ if ! is_manager_node; then
fi
# Get current list of Grid Node Agents that need to be upgraded
if ! RAW_JSON=$(fleet_api "agents?perPage=20&page=1&kuery=NOT%20agent.version%3A%20{{ELASTICSEARCHDEFAULTS.elasticsearch.version | urlencode }}%20AND%20policy_id%3A%20so-grid-nodes_%2A&showInactive=false&getStatusSummary=true" -H 'kbn-xsrf: true' -H 'Content-Type: application/json'); then
RAW_JSON=$(curl -K /opt/so/conf/elasticsearch/curl.config -L "http://localhost:5601/api/fleet/agents?perPage=20&page=1&kuery=NOT%20agent.version%3A%20{{ELASTICSEARCHDEFAULTS.elasticsearch.version}}%20AND%20policy_id%3A%20so-grid-nodes_%2A&showInactive=false&getStatusSummary=true" --retry 3 --retry-delay 30 --fail 2>/dev/null)
printf "Failed to query for current Grid Agents...\n"
exit 1
# Check to make sure that the server responded with good data - else, bail from script
CHECKSUM=$(jq -r '.page' <<< "$RAW_JSON")
if [ "$CHECKSUM" -ne 1 ]; then
printf "Failed to query for current Grid Agents...\n"
exit 1
fi
# Generate list of Node Agents that need updates
@@ -34,12 +36,10 @@ if [ "$OUTDATED_LIST" != '[]' ]; then
printf "Initiating upgrades for $AGENTNUMBERS Agents to Elastic {{ELASTICSEARCHDEFAULTS.elasticsearch.version}}...\n\n"
# Generate updated JSON payload
JSON_STRING=$(jq -n --arg ELASTICVERSION "{{ELASTICSEARCHDEFAULTS.elasticsearch.version}}" --argjson UPDATELIST "$OUTDATED_LIST" '{"version": $ELASTICVERSION,"agents": $UPDATELIST }')
JSON_STRING=$(jq -n --arg ELASTICVERSION {{ELASTICSEARCHDEFAULTS.elasticsearch.version}} --arg UPDATELIST $OUTDATED_LIST '{"version": $ELASTICVERSION,"agents": $UPDATELIST }')
# Update Node Agents
if ! fleet_api "agents/bulk_upgrade" -XPOST -H 'kbn-xsrf: true' -H 'Content-Type: application/json' -d "$JSON_STRING"; then
printf "Failed to initiate Agent upgrades...\n"
fi
curl -K /opt/so/conf/elasticsearch/curl.config -L -X POST "http://localhost:5601/api/fleet/agents/bulk_upgrade" -H 'kbn-xsrf: true' -H 'Content-Type: application/json' -d "$JSON_STRING"
else
printf "No Agents need updates... Exiting\n\n"
exit 0
@@ -235,16 +235,6 @@ function update_kafka_outputs() {
{% endif %}
# Compare the current Elastic Fleet certificate against what is on disk
POLICY_CERT_SHA=$(jq -r '.item.ssl.certificate' <<< $RAW_JSON | openssl x509 -noout -sha256 -fingerprint)
DISK_CERT_SHA=$(openssl x509 -in /etc/pki/elasticfleet-logstash.crt -noout -sha256 -fingerprint)
if [[ "$POLICY_CERT_SHA" != "$DISK_CERT_SHA" ]]; then
printf "Certificate on disk doesn't match certificate in policy - forcing update\n"
UPDATE_CERTS=true
FORCE_UPDATE=true
fi
# Sort & hash the new list of Logstash Outputs
NEW_LIST_JSON=$(jq --compact-output --null-input '$ARGS.positional' --args -- "${NEW_LIST[@]}")
NEW_HASH=$(sha256sum <<< "$NEW_LIST_JSON" | awk '{print $1}')
+1 -1
View File
@@ -4,7 +4,7 @@
# Elastic License 2.0.
{% from 'allowed_states.map.jinja' import allowed_states %}
{% if sls in allowed_states %}
{% if sls.split('.')[0] in allowed_states %}
{% from 'vars/globals.map.jinja' import GLOBALS %}
{% from 'elasticsearch/config.map.jinja' import ELASTICSEARCHMERGED %}
{% from 'elasticsearch/template.map.jinja' import ES_INDEX_SETTINGS, SO_MANAGED_INDICES %}
+7 -6
View File
@@ -17,7 +17,7 @@ include:
- elasticsearch.ssl
- elasticsearch.config
- elasticsearch.sostatus
{%- if GLOBALS.role != "so-searchnode" %}
{%- if GLOBALS.role != 'so-searchode' %}
- elasticsearch.cluster
{%- endif%}
@@ -102,6 +102,11 @@ so-elasticsearch:
- cmd: auth_users_roles_inode
- cmd: auth_users_inode
delete_so-elasticsearch_so-status.disabled:
file.uncomment:
- name: /opt/so/conf/so-status/so-status.conf
- regex: ^so-elasticsearch$
wait_for_so-elasticsearch:
http.wait_for_successful_query:
- name: "https://localhost:9200/"
@@ -112,14 +117,10 @@ wait_for_so-elasticsearch:
- status: 200
- wait_for: 300
- request_interval: 15
- backend: requests
- require:
- docker_container: so-elasticsearch
delete_so-elasticsearch_so-status.disabled:
file.uncomment:
- name: /opt/so/conf/so-status/so-status.conf
- regex: ^so-elasticsearch$
{% else %}
{{sls}}_state_not_allowed:
@@ -103,13 +103,11 @@ load_component_templates() {
local pattern="${ELASTICSEARCH_TEMPLATES_DIR}/component/$2"
local append_mappings="${3:-"false"}"
# current state of nullglob shell option
shopt -q nullglob && nullglob_set=1 || nullglob_set=0
shopt -s nullglob
echo -e "\nLoading $printed_name component templates...\n"
if ! compgen -G "${pattern}/*.json" > /dev/null; then
echo "No $printed_name component templates found in ${pattern}, skipping."
return
fi
for component in "$pattern"/*.json; do
tmpl_name=$(basename "${component%.json}")
@@ -123,6 +121,11 @@ load_component_templates() {
SO_LOAD_FAILURES_NAMES+=("$component")
fi
done
# restore nullglob shell option if needed
if [[ $nullglob_set -eq 1 ]]; then
shopt -u nullglob
fi
}
check_elasticsearch_responsive() {
@@ -133,32 +136,7 @@ check_elasticsearch_responsive() {
fail "Elasticsearch is not responding. Please review Elasticsearch logs /opt/so/log/elasticsearch/securityonion.log for more details. Additionally, consider running so-elasticsearch-troubleshoot."
}
index_templates_exist() {
local templates_dir="$1"
if [[ ! -d "$templates_dir" ]]; then
return 1
fi
compgen -G "${templates_dir}/*.json" > /dev/null
}
should_load_addon_templates() {
if [[ "$IS_HEAVYNODE" == "true" ]]; then
return 1
fi
# Skip statefile checks when forcing template load
if [[ "$FORCE" != "true" ]]; then
if [[ ! -f "$SO_STATEFILE_SUCCESS" || -f "$ADDON_STATEFILE_SUCCESS" ]]; then
return 1
fi
fi
index_templates_exist "$ADDON_TEMPLATES_DIR"
}
if [[ "$FORCE" == "true" || ! -f "$SO_STATEFILE_SUCCESS" ]] && index_templates_exist "$SO_TEMPLATES_DIR"; then
if [[ "$FORCE" == "true" || ! -f "$SO_STATEFILE_SUCCESS" ]]; then
check_elasticsearch_responsive
if [[ "$IS_HEAVYNODE" == "false" ]]; then
@@ -223,14 +201,13 @@ if [[ "$FORCE" == "true" || ! -f "$SO_STATEFILE_SUCCESS" ]] && index_templates_e
fail "Failed to load all Security Onion core templates successfully."
fi
fi
elif ! index_templates_exist "$SO_TEMPLATES_DIR"; then
echo "No Security Onion core index templates found in ${SO_TEMPLATES_DIR}, skipping."
elif [[ -f "$SO_STATEFILE_SUCCESS" ]]; then
else
echo "Security Onion core templates already loaded"
fi
# Start loading addon templates
if should_load_addon_templates; then
if [[ (-d "$ADDON_TEMPLATES_DIR" && -f "$SO_STATEFILE_SUCCESS" && "$IS_HEAVYNODE" == "false" && ! -f "$ADDON_STATEFILE_SUCCESS") || (-d "$ADDON_TEMPLATES_DIR" && "$IS_HEAVYNODE" == "false" && "$FORCE" == "true") ]]; then
check_elasticsearch_responsive
-3
View File
@@ -11,7 +11,6 @@
'so-kratos',
'so-hydra',
'so-nginx',
'so-postgres',
'so-redis',
'so-soc',
'so-strelka-coordinator',
@@ -35,7 +34,6 @@
'so-hydra',
'so-logstash',
'so-nginx',
'so-postgres',
'so-redis',
'so-soc',
'so-strelka-coordinator',
@@ -79,7 +77,6 @@
'so-kratos',
'so-hydra',
'so-nginx',
'so-postgres',
'so-soc'
] %}
-42
View File
@@ -98,10 +98,6 @@ firewall:
tcp:
- 8086
udp: []
postgres:
tcp:
- 5432
udp: []
kafka_controller:
tcp:
- 9093
@@ -197,7 +193,6 @@ firewall:
- kibana
- redis
- influxdb
- postgres
- elasticsearch_rest
- elasticsearch_node
- localrules
@@ -384,7 +379,6 @@ firewall:
- kibana
- redis
- influxdb
- postgres
- elasticsearch_rest
- elasticsearch_node
- docker_registry
@@ -398,7 +392,6 @@ firewall:
- elasticsearch_rest
- docker_registry
- influxdb
- postgres
- sensoroni
- yum
- beats_5044
@@ -411,7 +404,6 @@ firewall:
portgroups:
- docker_registry
- influxdb
- postgres
- sensoroni
- yum
- beats_5044
@@ -429,7 +421,6 @@ firewall:
- yum
- docker_registry
- influxdb
- postgres
- sensoroni
searchnode:
portgroups:
@@ -440,7 +431,6 @@ firewall:
- yum
- docker_registry
- influxdb
- postgres
- elastic_agent_control
- elastic_agent_data
- elastic_agent_update
@@ -454,7 +444,6 @@ firewall:
- yum
- docker_registry
- influxdb
- postgres
- elastic_agent_control
- elastic_agent_data
- elastic_agent_update
@@ -464,7 +453,6 @@ firewall:
- yum
- docker_registry
- influxdb
- postgres
- elastic_agent_control
- elastic_agent_data
- elastic_agent_update
@@ -498,7 +486,6 @@ firewall:
portgroups:
- docker_registry
- influxdb
- postgres
- sensoroni
- yum
- elastic_agent_control
@@ -509,7 +496,6 @@ firewall:
- yum
- docker_registry
- influxdb
- postgres
- elastic_agent_control
- elastic_agent_data
- elastic_agent_update
@@ -604,7 +590,6 @@ firewall:
- kibana
- redis
- influxdb
- postgres
- elasticsearch_rest
- elasticsearch_node
- docker_registry
@@ -618,7 +603,6 @@ firewall:
- elasticsearch_rest
- docker_registry
- influxdb
- postgres
- sensoroni
- yum
- beats_5044
@@ -631,7 +615,6 @@ firewall:
portgroups:
- docker_registry
- influxdb
- postgres
- sensoroni
- yum
- beats_5044
@@ -649,7 +632,6 @@ firewall:
- yum
- docker_registry
- influxdb
- postgres
- sensoroni
searchnode:
portgroups:
@@ -660,7 +642,6 @@ firewall:
- yum
- docker_registry
- influxdb
- postgres
- elastic_agent_control
- elastic_agent_data
- elastic_agent_update
@@ -674,7 +655,6 @@ firewall:
- yum
- docker_registry
- influxdb
- postgres
- elastic_agent_control
- elastic_agent_data
- elastic_agent_update
@@ -684,7 +664,6 @@ firewall:
- yum
- docker_registry
- influxdb
- postgres
- elastic_agent_control
- elastic_agent_data
- elastic_agent_update
@@ -716,7 +695,6 @@ firewall:
portgroups:
- docker_registry
- influxdb
- postgres
- sensoroni
- yum
- elastic_agent_control
@@ -727,7 +705,6 @@ firewall:
- yum
- docker_registry
- influxdb
- postgres
- elastic_agent_control
- elastic_agent_data
- elastic_agent_update
@@ -822,7 +799,6 @@ firewall:
- kibana
- redis
- influxdb
- postgres
- elasticsearch_rest
- elasticsearch_node
- docker_registry
@@ -836,7 +812,6 @@ firewall:
- elasticsearch_rest
- docker_registry
- influxdb
- postgres
- sensoroni
- yum
- beats_5044
@@ -849,7 +824,6 @@ firewall:
portgroups:
- docker_registry
- influxdb
- postgres
- sensoroni
- yum
- beats_5044
@@ -867,7 +841,6 @@ firewall:
- yum
- docker_registry
- influxdb
- postgres
- sensoroni
searchnode:
portgroups:
@@ -877,7 +850,6 @@ firewall:
- yum
- docker_registry
- influxdb
- postgres
- elastic_agent_control
- elastic_agent_data
- elastic_agent_update
@@ -890,7 +862,6 @@ firewall:
- yum
- docker_registry
- influxdb
- postgres
- elastic_agent_control
- elastic_agent_data
- elastic_agent_update
@@ -900,7 +871,6 @@ firewall:
- yum
- docker_registry
- influxdb
- postgres
- elastic_agent_control
- elastic_agent_data
- elastic_agent_update
@@ -934,7 +904,6 @@ firewall:
portgroups:
- docker_registry
- influxdb
- postgres
- sensoroni
- yum
- elastic_agent_control
@@ -945,7 +914,6 @@ firewall:
- yum
- docker_registry
- influxdb
- postgres
- elastic_agent_control
- elastic_agent_data
- elastic_agent_update
@@ -1043,7 +1011,6 @@ firewall:
- kibana
- redis
- influxdb
- postgres
- elasticsearch_rest
- elasticsearch_node
- docker_registry
@@ -1064,7 +1031,6 @@ firewall:
- elasticsearch_rest
- docker_registry
- influxdb
- postgres
- sensoroni
- yum
- beats_5044
@@ -1077,7 +1043,6 @@ firewall:
portgroups:
- docker_registry
- influxdb
- postgres
- sensoroni
- yum
- beats_5044
@@ -1089,7 +1054,6 @@ firewall:
portgroups:
- docker_registry
- influxdb
- postgres
- sensoroni
- yum
- beats_5044
@@ -1101,7 +1065,6 @@ firewall:
portgroups:
- docker_registry
- influxdb
- postgres
- sensoroni
- yum
- redis
@@ -1111,7 +1074,6 @@ firewall:
portgroups:
- docker_registry
- influxdb
- postgres
- sensoroni
- yum
- redis
@@ -1122,7 +1084,6 @@ firewall:
- yum
- docker_registry
- influxdb
- postgres
- elastic_agent_control
- elastic_agent_data
- elastic_agent_update
@@ -1159,7 +1120,6 @@ firewall:
portgroups:
- docker_registry
- influxdb
- postgres
- sensoroni
- yum
- elastic_agent_control
@@ -1170,7 +1130,6 @@ firewall:
- yum
- docker_registry
- influxdb
- postgres
- elastic_agent_control
- elastic_agent_data
- elastic_agent_update
@@ -1514,7 +1473,6 @@ firewall:
- kibana
- redis
- influxdb
- postgres
- elasticsearch_rest
- elasticsearch_node
- elastic_agent_control
+1 -1
View File
@@ -22,7 +22,7 @@ kibana:
- default
- file
migrations:
discardCorruptObjects: "9.3.3"
discardCorruptObjects: "8.18.8"
telemetry:
enabled: False
xpack:
+1 -1
View File
@@ -3,8 +3,8 @@ kratos:
description: Enables or disables the Kratos authentication system. WARNING - Disabling this process will cause the grid to malfunction. Re-enabling this setting will require manual effort via SSH.
forcedType: bool
advanced: True
readonly: True
helpLink: kratos
oidc:
enabled:
description: Set to True to enable OIDC / Single Sign-On (SSO) to SOC. Requires a valid Security Onion license key.
+1 -46
View File
@@ -273,7 +273,7 @@ function deleteMinionFiles () {
log "ERROR" "Failed to delete $PILLARFILE"
return 1
fi
rm -f $ADVPILLARFILE
if [ $? -ne 0 ]; then
log "ERROR" "Failed to delete $ADVPILLARFILE"
@@ -281,39 +281,6 @@ function deleteMinionFiles () {
fi
}
# Remove this minion's postgres Telegraf credential from the shared creds
# pillar and drop the matching role in Postgres. Always returns 0 so a dead
# or unreachable so-postgres doesn't block minion deletion — in that case we
# log a warning and leave the role behind for manual cleanup.
function remove_postgres_telegraf_from_minion() {
local MINION_SAFE
MINION_SAFE=$(echo "$MINION_ID" | tr '.-' '__' | tr '[:upper:]' '[:lower:]')
local PG_USER="so_telegraf_${MINION_SAFE}"
log "INFO" "Removing postgres telegraf cred for $MINION_ID"
so-telegraf-cred remove "$MINION_ID" >/dev/null 2>&1 || true
if docker ps --format '{{.Names}}' 2>/dev/null | grep -q '^so-postgres$'; then
if ! docker exec -i so-postgres psql -v ON_ERROR_STOP=1 -U postgres -d so_telegraf >/dev/null 2>&1 <<EOSQL
DO \$\$
BEGIN
IF EXISTS (SELECT FROM pg_catalog.pg_roles WHERE rolname = '$PG_USER') THEN
EXECUTE format('REASSIGN OWNED BY %I TO so_telegraf', '$PG_USER');
EXECUTE format('DROP OWNED BY %I', '$PG_USER');
EXECUTE format('DROP ROLE %I', '$PG_USER');
END IF;
END
\$\$;
EOSQL
then
log "WARN" "Failed to drop postgres role $PG_USER; pillar entry was removed — drop manually if the role persists"
fi
else
log "WARN" "so-postgres container is not running; skipping DB role cleanup for $PG_USER"
fi
}
# Create the minion file
function ensure_socore_ownership() {
log "INFO" "Setting socore ownership on minion files"
@@ -575,17 +542,6 @@ function add_telegraf_to_minion() {
log "ERROR" "Failed to add telegraf configuration to $PILLARFILE"
return 1
fi
# Provision the per-minion postgres Telegraf credential in the shared
# telegraf/creds.sls pillar. so-telegraf-cred is the only writer; it
# generates a password on first add and is a no-op on re-add so the cred
# is stable across repeated so-minion runs. postgres.telegraf_users on the
# manager creates/updates the DB role from the same pillar.
so-telegraf-cred add "$MINION_ID"
if [ $? -ne 0 ]; then
log "ERROR" "Failed to provision postgres telegraf cred for $MINION_ID"
return 1
fi
}
function add_influxdb_to_minion() {
@@ -1113,7 +1069,6 @@ case "$OPERATION" in
"delete")
log "INFO" "Removing minion $MINION_ID"
remove_postgres_telegraf_from_minion
deleteMinionFiles || {
log "ERROR" "Failed to delete minion files for $MINION_ID"
exit 1
-329
View File
@@ -1,329 +0,0 @@
#!/usr/bin/env python3
# Copyright Security Onion Solutions LLC and/or licensed to Security Onion Solutions LLC under one
# or more contributor license agreements. Licensed under the Elastic License 2.0 as shown at
# https://securityonion.net/license; you may not use this file except in compliance with the
# Elastic License 2.0.
"""
so-pillar-import — populate the so_pillar.* schema in so-postgres from the
on-disk Salt pillar tree.
Reads /opt/so/saltstack/local/pillar/, decomposes each .sls file into a
(scope, role|minion_id, pillar_path, data) tuple, and UPSERTs it into
so_pillar.pillar_entry. Idempotent — re-running with no SLS edits produces
no version bumps because the audit trigger only writes a row when data
actually changes.
Bootstrap and mine-driven files are skipped (see EXCLUDE_BASENAMES /
EXCLUDE_PREFIXES below). Files containing Jinja templates ({% or {{) are
also skipped — those stay disk-authoritative and ext_pillar_first: False
means they render before the PG overlay anyway.
All SQL goes through `docker exec so-postgres psql` so no separate DSN
config is required at first-install time. Designed to be called by
salt/postgres/schema_pillar.sls (initial seed) and by salt/manager/tools/
sbin/so-minion (per-minion sync on add/delete).
"""
import argparse
import json
import os
import shlex
import subprocess
import sys
from pathlib import Path
import yaml
PILLAR_LOCAL_ROOT = Path("/opt/so/saltstack/local/pillar")
PILLAR_DEFAULT_ROOT = Path("/opt/so/saltstack/default/pillar")
DOCKER_CONTAINER = "so-postgres"
PG_SUPERUSER = "postgres"
PG_DATABASE = "securityonion"
# Files that must NEVER move to Postgres. These are read by Salt before
# Postgres is reachable, or contain renderer-time computed values (mine, etc.).
EXCLUDE_BASENAMES = {
"secrets.sls",
"auth.sls", # postgres/auth.sls bootstrap
"top.sls",
}
# Filename prefixes to skip — these are renderer-time computed pillars
# (Salt mine, file_exists guards, etc.) that have to stay on disk.
EXCLUDE_PATH_FRAGMENTS = (
"/elasticsearch/nodes.sls",
"/redis/nodes.sls",
"/kafka/nodes.sls",
"/hypervisor/nodes.sls",
"/logstash/nodes.sls",
"/node_data/ips.sls",
"/postgres/auth.sls",
"/elasticsearch/auth.sls",
"/kibana/secrets.sls",
)
def log(level, msg):
print(f"[{level}] {msg}", file=sys.stderr)
def is_jinja_templated(content_bytes):
return b"{%" in content_bytes or b"{{" in content_bytes
def classify(path):
"""Return (scope, role_name, minion_id, pillar_path) for a pillar file
or None to skip it. role_name is None for now — the importer leaves role
membership to the so_pillar.minion trigger and the salt/auth reactor."""
rel_str = str(path)
if path.name in EXCLUDE_BASENAMES:
return None
for frag in EXCLUDE_PATH_FRAGMENTS:
if frag in rel_str:
return None
# /local/pillar/minions/<id>.sls or adv_<id>.sls
if path.parent.name == "minions":
stem = path.stem # filename without .sls
if stem.startswith("adv_"):
mid = stem[4:]
return ("minion", None, mid, f"minions.adv_{mid}")
return ("minion", None, stem, f"minions.{stem}")
# /local/pillar/<section>/<file>.sls
if path.parent.parent == PILLAR_LOCAL_ROOT or path.parent.parent == PILLAR_DEFAULT_ROOT:
section = path.parent.name
stem = path.stem
# Only soc_<section>.sls and adv_<section>.sls are SOC-managed pillar
# surfaces. Other files (e.g. nodes.sls, auth.sls, *.token) are
# either covered by EXCLUDE_PATH_FRAGMENTS or are bootstrap surfaces
# we leave alone for now.
if stem.startswith("soc_") or stem.startswith("adv_"):
return ("global", None, None, f"{section}.{stem}")
return None
return None
def parse_yaml_file(path):
with open(path, "rb") as f:
content = f.read()
if not content.strip():
return {}
if is_jinja_templated(content):
return None
data = yaml.safe_load(content)
if data is None:
return {}
if not isinstance(data, dict):
return {"_raw": data}
return data
def derive_node_type(minion_id):
"""Conventional Security Onion minion ids are <host>_<role>. Take the
last underscore-delimited token as the canonical role suffix."""
parts = minion_id.rsplit("_", 1)
if len(parts) == 2:
return parts[1]
return None
def docker_psql(sql, *, db=PG_DATABASE, user=PG_SUPERUSER, on_error_stop=True, capture=True):
"""Run sql via docker exec ... psql. Returns stdout as str."""
args = [
"docker", "exec", "-i", DOCKER_CONTAINER,
"psql", "-U", user, "-d", db, "-tA", "-q",
]
if on_error_stop:
args += ["-v", "ON_ERROR_STOP=1"]
proc = subprocess.run(
args, input=sql.encode(),
capture_output=capture, check=False,
)
if proc.returncode != 0:
sys.stderr.write(proc.stderr.decode(errors="replace"))
raise RuntimeError(f"docker exec psql failed (rc={proc.returncode})")
return proc.stdout.decode(errors="replace")
def upsert_minion(minion_id, node_type):
sql = (
"INSERT INTO so_pillar.minion (minion_id, node_type) "
f"VALUES ({pg_str(minion_id)}, {pg_str(node_type) if node_type else 'NULL'}) "
"ON CONFLICT (minion_id) DO UPDATE SET node_type = EXCLUDED.node_type;"
)
docker_psql(sql)
def delete_minion(minion_id):
"""CASCADE removes pillar_entry + role_member rows."""
sql = f"DELETE FROM so_pillar.minion WHERE minion_id = {pg_str(minion_id)};"
docker_psql(sql)
def upsert_pillar_entry(scope, role_name, minion_id, pillar_path, data, reason):
"""Insert or update the row keyed by the partial unique index that
matches scope. Audit trigger handles history; versioning trigger bumps
version only when data changes."""
data_json = json.dumps(data)
role_sql = pg_str(role_name) if role_name else "NULL"
minion_sql = pg_str(minion_id) if minion_id else "NULL"
reason_sql = pg_str(reason)
if scope == "global":
conflict = "(pillar_path) WHERE scope='global'"
elif scope == "role":
conflict = "(role_name, pillar_path) WHERE scope='role'"
elif scope == "minion":
conflict = "(minion_id, pillar_path) WHERE scope='minion'"
else:
raise ValueError(f"unknown scope {scope!r}")
sql = (
"BEGIN;\n"
f"SELECT set_config('so_pillar.change_reason', {reason_sql}, true);\n"
f"INSERT INTO so_pillar.pillar_entry "
f"(scope, role_name, minion_id, pillar_path, data, change_reason) "
f"VALUES ({pg_str(scope)}, {role_sql}, {minion_sql}, {pg_str(pillar_path)}, {pg_jsonb(data_json)}, {reason_sql}) "
f"ON CONFLICT {conflict} DO UPDATE "
f"SET data = EXCLUDED.data, change_reason = EXCLUDED.change_reason;\n"
"COMMIT;\n"
)
docker_psql(sql)
def pg_str(s):
"""Escape a Python str for inclusion in literal SQL. Pillar content has
already been validated as YAML; we just need standard SQL escaping."""
if s is None:
return "NULL"
return "'" + str(s).replace("'", "''") + "'"
def pg_jsonb(json_str):
return pg_str(json_str) + "::jsonb"
def walk_pillar_root(root, paths):
if not root.is_dir():
return
for path in root.rglob("*.sls"):
if path.is_file():
paths.append(path)
def import_minion(minion_id, node_type, dry_run, reason):
"""Re-import every pillar file for a single minion."""
if not minion_id:
raise ValueError("minion_id required for --scope minion")
upsert_minion(minion_id, node_type)
log("INFO", f"Upserted minion row {minion_id} (node_type={node_type})")
targets = [
PILLAR_LOCAL_ROOT / "minions" / f"{minion_id}.sls",
PILLAR_LOCAL_ROOT / "minions" / f"adv_{minion_id}.sls",
]
for path in targets:
if not path.exists():
log("INFO", f" (no file at {path})")
continue
klass = classify(path)
if not klass:
log("INFO", f" skip {path} (excluded)")
continue
scope, role, mid, pillar_path = klass
data = parse_yaml_file(path)
if data is None:
log("WARN", f" skip {path} (Jinja-templated; stays disk-only)")
continue
if dry_run:
log("DRY", f" would upsert {scope}/{pillar_path} = {len(json.dumps(data))} bytes")
continue
upsert_pillar_entry(scope, role, mid, pillar_path, data, reason)
log("INFO", f" imported {scope}/{pillar_path}")
def import_all(dry_run, reason):
"""Walk the entire local pillar tree and import every eligible file."""
paths = []
walk_pillar_root(PILLAR_LOCAL_ROOT, paths)
imported = 0
skipped = 0
minions_seen = set()
for path in sorted(paths):
klass = classify(path)
if not klass:
skipped += 1
continue
scope, role, minion_id, pillar_path = klass
data = parse_yaml_file(path)
if data is None:
log("WARN", f"skip {path} (Jinja-templated; stays disk-only)")
skipped += 1
continue
if scope == "minion" and minion_id not in minions_seen:
node_type = derive_node_type(minion_id)
if not dry_run:
upsert_minion(minion_id, node_type)
minions_seen.add(minion_id)
if dry_run:
log("DRY", f"would upsert {scope}/{pillar_path} ({len(json.dumps(data))} bytes)")
else:
upsert_pillar_entry(scope, role, minion_id, pillar_path, data, reason)
log("INFO", f"imported {scope}/{pillar_path}")
imported += 1
log("INFO", f"done: {imported} imported, {skipped} skipped")
def main():
ap = argparse.ArgumentParser(description=__doc__)
ap.add_argument("--scope", choices=("global", "role", "minion", "all"), default="all")
ap.add_argument("--minion-id")
ap.add_argument("--node-type", help="override node_type for --scope minion (default: derived from minion_id)")
ap.add_argument("--delete", action="store_true",
help="With --scope minion, remove the minion row (and its pillar rows via CASCADE)")
ap.add_argument("--dry-run", action="store_true")
ap.add_argument("--diff", action="store_true",
help="(reserved) print structural diffs vs current DB content")
ap.add_argument("--yes", action="store_true",
help="Skip confirmation prompts (currently unused; reserved)")
ap.add_argument("--reason", default="so-pillar-import",
help="change_reason recorded in pillar_entry_history")
args = ap.parse_args()
try:
if args.scope == "minion":
if not args.minion_id:
ap.error("--minion-id required when --scope minion")
if args.delete:
if args.dry_run:
log("DRY", f"would delete {args.minion_id}")
else:
delete_minion(args.minion_id)
log("INFO", f"deleted {args.minion_id}")
else:
node_type = args.node_type or derive_node_type(args.minion_id)
import_minion(args.minion_id, node_type, args.dry_run, args.reason)
elif args.scope == "all":
import_all(args.dry_run, args.reason)
else:
log("ERROR", f"--scope {args.scope} not yet implemented; use --scope all or --scope minion")
return 2
except Exception as e:
log("ERROR", str(e))
return 1
return 0
if __name__ == "__main__":
sys.exit(main())
-54
View File
@@ -1,54 +0,0 @@
#!/bin/bash
# Copyright Security Onion Solutions LLC and/or licensed to Security Onion Solutions LLC under one
# or more contributor license agreements. Licensed under the Elastic License 2.0 as shown at
# https://securityonion.net/license; you may not use this file except in compliance with the
# Elastic License 2.0.
# Single writer for the Telegraf Postgres credentials pillar. Thin wrapper
# around so-yaml.py that generates a password on first add and no-ops on
# re-add so the cred is stable across repeated so-minion runs.
#
# Note: so-yaml.py splits keys on '.' with no escape. SO minion ids are
# dot-free by construction (setup/so-functions:1884 takes the short_name
# before the first '.'), so using the raw minion id as the key is safe.
CREDS=/opt/so/saltstack/local/pillar/telegraf/creds.sls
usage() {
echo "Usage: $0 <add|remove> <minion_id>" >&2
exit 2
}
seed_creds_file() {
mkdir -p "$(dirname "$CREDS")" || return 1
if [[ ! -f "$CREDS" ]]; then
(umask 027 && printf 'telegraf:\n postgres_creds: {}\n' > "$CREDS") || return 1
chown socore:socore "$CREDS" 2>/dev/null || true
chmod 640 "$CREDS" || return 1
fi
}
OP=$1
MID=$2
[[ -z "$OP" || -z "$MID" ]] && usage
case "$OP" in
add)
SAFE=$(echo "$MID" | tr '.-' '__' | tr '[:upper:]' '[:lower:]')
seed_creds_file || exit 1
if so-yaml.py get -r "$CREDS" "telegraf.postgres_creds.${MID}.user" >/dev/null 2>&1; then
exit 0
fi
PASS=$(tr -dc 'A-Za-z0-9~!@#^&*()_=+[]|;:,.<>?-' < /dev/urandom | head -c 72)
so-yaml.py replace "$CREDS" "telegraf.postgres_creds.${MID}.user" "so_telegraf_${SAFE}" >/dev/null
so-yaml.py replace "$CREDS" "telegraf.postgres_creds.${MID}.pass" "$PASS" >/dev/null
;;
remove)
[[ -f "$CREDS" ]] || exit 0
so-yaml.py remove "$CREDS" "telegraf.postgres_creds.${MID}" >/dev/null 2>&1 || true
;;
*)
usage
;;
esac
+5 -197
View File
@@ -13,64 +13,6 @@ import json
lockFile = "/tmp/so-yaml.lock"
# postsalt: so-yaml supports three backend modes for PG-managed pillar paths:
#
# dual — write disk + mirror to so_pillar.*. Reads from disk.
# Used during the migration transition when disk is still
# canonical and PG runs as a shadow.
# postgres — write to so_pillar.* only. Reads from so_pillar.*. No disk
# file is touched. The end state once cutover is complete.
# disk — disk only, no PG. Emergency rollback escape hatch.
#
# Bootstrap and mine-driven files (secrets.sls, ca/init.sls, */nodes.sls,
# top.sls, etc.) are always handled on disk regardless of mode — those paths
# are explicitly excluded by so_yaml_postgres.locate() raising SkipPath.
#
# Mode resolution: SO_YAML_BACKEND env var, then /opt/so/conf/so-yaml/mode,
# then default 'dual' (safe upgrade behavior — flipping to 'postgres' is
# done by schema_pillar.sls after the schema is in place and the importer
# has run at least once).
MODE_FILE = "/opt/so/conf/so-yaml/mode"
VALID_MODES = ("dual", "postgres", "disk")
DEFAULT_MODE = "dual"
try:
sys.path.insert(0, os.path.dirname(os.path.abspath(__file__)))
import so_yaml_postgres
_SO_YAML_PG_AVAILABLE = True
except Exception as _exc:
_SO_YAML_PG_AVAILABLE = False
def _resolveBackendMode():
env = os.environ.get("SO_YAML_BACKEND")
if env and env in VALID_MODES:
return env
try:
with open(MODE_FILE, "r") as fh:
value = fh.read().strip()
if value in VALID_MODES:
return value
except (IOError, OSError):
pass
return DEFAULT_MODE
_BACKEND_MODE = _resolveBackendMode()
def _isPgManaged(filename):
"""True when so-yaml should route this file's reads/writes through
so_pillar.*. False for bootstrap/mine-driven files that always live on
disk, and for arbitrary YAML paths outside the pillar tree."""
if not _SO_YAML_PG_AVAILABLE:
return False
try:
return so_yaml_postgres.is_pg_managed(filename)
except Exception:
return False
def showUsage(args):
print('Usage: {} <COMMAND> <YAML_FILE> [ARGS...]'.format(sys.argv[0]), file=sys.stderr)
@@ -83,14 +25,8 @@ def showUsage(args):
print(' get [-r] - Displays (to stdout) the value stored in the given key. Requires KEY arg. Use -r for raw output without YAML formatting.', file=sys.stderr)
print(' remove - Removes a yaml key, if it exists. Requires KEY arg.', file=sys.stderr)
print(' replace - Replaces (or adds) a new key and set its value. Requires KEY and VALUE args.', file=sys.stderr)
print(' purge - Delete the YAML file from disk and remove its rows from so_pillar.* (no KEY arg).', file=sys.stderr)
print(' help - Prints this usage information.', file=sys.stderr)
print('', file=sys.stderr)
print(' Backend mode:', file=sys.stderr)
print(' Resolved from $SO_YAML_BACKEND, then /opt/so/conf/so-yaml/mode, default "dual".', file=sys.stderr)
print(' Valid values: dual | postgres | disk. Bootstrap pillar files (secrets, ca, *.nodes.sls)', file=sys.stderr)
print(' are always handled on disk regardless of mode.', file=sys.stderr)
print('', file=sys.stderr)
print(' Where:', file=sys.stderr)
print(' YAML_FILE - Path to the file that will be modified. Ex: /opt/so/conf/service/conf.yaml', file=sys.stderr)
print(' KEY - YAML key, does not support \' or " characters at this time. Ex: level1.level2', file=sys.stderr)
@@ -103,128 +39,14 @@ def showUsage(args):
def loadYaml(filename):
"""Load a YAML file's content as a dict.
PG-canonical mode (`postgres`): for PG-managed paths, read from
so_pillar.pillar_entry. A missing row is treated as an empty dict so
that `replace`/`add` on a fresh path can populate it from scratch.
Other modes / non-PG-managed paths: read from disk as today.
"""
if _BACKEND_MODE == "postgres" and _isPgManaged(filename):
try:
data = so_yaml_postgres.read_yaml(filename)
except so_yaml_postgres.SkipPath:
data = None
except Exception as e:
print(f"so-yaml: pg read failed for {filename}: {e}", file=sys.stderr)
sys.exit(1)
return data if data is not None else {}
try:
with open(filename, "r") as file:
content = file.read()
return yaml.safe_load(content)
except FileNotFoundError:
print(f"File not found: {filename}", file=sys.stderr)
sys.exit(1)
except Exception as e:
print(f"Error reading file {filename}: {e}", file=sys.stderr)
sys.exit(1)
file = open(filename, "r")
content = file.read()
return yaml.safe_load(content)
def writeYaml(filename, content):
"""Persist `content` for `filename`.
PG-canonical mode + PG-managed path: write only to so_pillar.*. A PG
failure is fatal (no disk fallback) — caller must retry.
Dual mode: write disk, then mirror to PG (failures are warnings).
Disk mode or non-PG-managed path: write disk only.
"""
if _BACKEND_MODE == "postgres" and _isPgManaged(filename):
if not _SO_YAML_PG_AVAILABLE:
print("so-yaml: PG-canonical mode requires so_yaml_postgres module", file=sys.stderr)
sys.exit(1)
ok, msg = so_yaml_postgres.write_yaml(
filename, content,
reason="so-yaml " + " ".join(sys.argv[1:2]))
if not ok:
print(f"so-yaml: pg write failed for {filename}: {msg}", file=sys.stderr)
sys.exit(1)
return None
file = open(filename, "w")
result = yaml.safe_dump(content, file)
file.close()
if _BACKEND_MODE == "dual":
_mirrorToPostgres(filename, content)
return result
def _mirrorToPostgres(filename, content):
"""Best-effort dual-write of a YAML mutation into so_pillar.*. Skips
files outside the PG-managed pillar surface (secrets.sls,
elasticsearch/nodes.sls, etc.) and silently degrades when so-postgres
is unreachable. Disk write is canonical in dual mode; this never
raises.
Only real PG failures (`pg write failed: ...`) are logged so the
common cases (skipped path, postgres not running) don't pollute
stderr."""
if not _SO_YAML_PG_AVAILABLE:
return
try:
ok, msg = so_yaml_postgres.write_yaml(filename, content,
reason="so-yaml " + " ".join(sys.argv[1:2]))
if not ok and msg.startswith("pg write failed"):
print(f"so-yaml: {msg}", file=sys.stderr)
except Exception as e: # pragma: no cover — defensive: never break disk write
print(f"so-yaml: pg mirror exception: {e}", file=sys.stderr)
def purgeFile(filename):
"""Delete a YAML file from disk and remove the matching rows from
so_pillar.*. Idempotent — missing file/row counts as success.
PG-canonical mode + PG-managed path: PG delete is canonical. If a stale
disk file from the dual-write era happens to still exist, it's removed
too as a cleanup courtesy. PG failure is fatal in this mode.
Dual / disk modes: remove disk first; PG cleanup is best-effort."""
if _BACKEND_MODE == "postgres" and _isPgManaged(filename):
if not _SO_YAML_PG_AVAILABLE:
print("so-yaml: PG-canonical mode requires so_yaml_postgres module", file=sys.stderr)
return 1
ok, msg = so_yaml_postgres.purge_yaml(filename, reason="so-yaml purge")
if not ok:
print(f"so-yaml: pg purge failed for {filename}: {msg}", file=sys.stderr)
return 1
if os.path.exists(filename):
try:
os.remove(filename)
except Exception as e:
print(f"so-yaml: warn — could not remove stale disk file {filename}: {e}", file=sys.stderr)
return 0
if os.path.exists(filename):
try:
os.remove(filename)
except Exception as e:
print(f"Failed to remove {filename}: {e}", file=sys.stderr)
return 1
if _BACKEND_MODE == "dual" and _SO_YAML_PG_AVAILABLE:
try:
ok, msg = so_yaml_postgres.purge_yaml(filename,
reason="so-yaml purge")
if not ok and msg.startswith("pg purge failed"):
print(f"so-yaml: {msg}", file=sys.stderr)
except Exception as e:
print(f"so-yaml: pg purge exception: {e}", file=sys.stderr)
return 0
return yaml.safe_dump(content, file)
def appendItem(content, key, listItem):
@@ -463,8 +285,7 @@ def add(args):
def removeKey(content, key):
pieces = key.split(".", 1)
if len(pieces) > 1:
if pieces[0] in content:
removeKey(content[pieces[0]], pieces[1])
removeKey(content[pieces[0]], pieces[1])
else:
content.pop(key, None)
@@ -542,18 +363,6 @@ def get(args):
return 0
def purge(args):
"""purge YAML_FILE — delete the file from disk and remove the matching
rows from so_pillar.* in so-postgres. Used by so-minion's delete path
(in place of `rm -f`) so the audit log captures the deletion and
role_member rows get cleaned up via FK CASCADE on so_pillar.minion."""
if len(args) != 1:
print('Missing filename arg', file=sys.stderr)
showUsage(None)
return 1
return purgeFile(args[0])
def main():
args = sys.argv[1:]
@@ -571,7 +380,6 @@ def main():
"get": get,
"remove": remove,
"replace": replace,
"purge": purge,
}
code = 1
-344
View File
@@ -973,347 +973,3 @@ class TestReplaceListObject(unittest.TestCase):
expected = "key1:\n- id: '1'\n status: updated\n- id: '2'\n status: inactive\n"
self.assertEqual(actual, expected)
class TestLoadYaml(unittest.TestCase):
def test_load_yaml_missing_file(self):
with patch('sys.exit', new=MagicMock()) as sysmock:
with patch('sys.stderr', new=StringIO()) as mock_stderr:
soyaml.loadYaml("/tmp/so-yaml_test-does-not-exist.yaml")
sysmock.assert_called_with(1)
self.assertIn("File not found:", mock_stderr.getvalue())
def test_load_yaml_read_error(self):
with patch('sys.exit', new=MagicMock()) as sysmock:
with patch('sys.stderr', new=StringIO()) as mock_stderr:
with patch('builtins.open', side_effect=PermissionError("denied")):
soyaml.loadYaml("/tmp/so-yaml_test-unreadable.yaml")
sysmock.assert_called_with(1)
self.assertIn("Error reading file", mock_stderr.getvalue())
class TestPurge(unittest.TestCase):
def test_purge_missing_arg(self):
# showUsage calls sys.exit(1); patch it like the other tests do.
with patch('sys.exit', new=MagicMock()):
with patch('sys.stderr', new=StringIO()) as mock_stderr:
rc = soyaml.purge([])
self.assertEqual(rc, 1)
self.assertIn("Missing filename", mock_stderr.getvalue())
def test_purge_existing_file(self):
filename = "/tmp/so-yaml_test_purge.yaml"
with open(filename, "w") as f:
f.write("key: value\n")
# Disable PG mirror so the test doesn't shell out to docker.
with patch.object(soyaml, '_SO_YAML_PG_AVAILABLE', False):
rc = soyaml.purge([filename])
self.assertEqual(rc, 0)
import os as _os
self.assertFalse(_os.path.exists(filename))
def test_purge_missing_file_idempotent(self):
filename = "/tmp/so-yaml_test_purge_missing.yaml"
import os as _os
if _os.path.exists(filename):
_os.remove(filename)
with patch.object(soyaml, '_SO_YAML_PG_AVAILABLE', False):
rc = soyaml.purge([filename])
self.assertEqual(rc, 0)
class TestSoYamlPostgres(unittest.TestCase):
"""Tests the path-locator and write/purge contract of the dual-write
backend module without actually contacting Postgres."""
def setUp(self):
import importlib
self.mod = importlib.import_module("so_yaml_postgres")
def test_locate_global_soc(self):
scope, role, mid, path = self.mod.locate(
"/opt/so/saltstack/local/pillar/soc/soc_soc.sls")
self.assertEqual(scope, "global")
self.assertIsNone(role)
self.assertIsNone(mid)
self.assertEqual(path, "soc.soc_soc")
def test_locate_global_advanced(self):
scope, role, mid, path = self.mod.locate(
"/opt/so/saltstack/local/pillar/soc/adv_soc.sls")
self.assertEqual(scope, "global")
self.assertEqual(path, "soc.adv_soc")
def test_locate_minion(self):
scope, role, mid, path = self.mod.locate(
"/opt/so/saltstack/local/pillar/minions/h1_sensor.sls")
self.assertEqual(scope, "minion")
self.assertEqual(mid, "h1_sensor")
self.assertEqual(path, "minions.h1_sensor")
def test_locate_minion_advanced(self):
scope, role, mid, path = self.mod.locate(
"/opt/so/saltstack/local/pillar/minions/adv_h1_sensor.sls")
self.assertEqual(scope, "minion")
self.assertEqual(mid, "h1_sensor")
self.assertEqual(path, "minions.adv_h1_sensor")
def test_locate_skip_secrets(self):
with self.assertRaises(self.mod.SkipPath):
self.mod.locate("/opt/so/saltstack/local/pillar/secrets.sls")
def test_locate_skip_postgres_auth(self):
with self.assertRaises(self.mod.SkipPath):
self.mod.locate("/opt/so/saltstack/local/pillar/postgres/auth.sls")
def test_locate_skip_mine_driven(self):
with self.assertRaises(self.mod.SkipPath):
self.mod.locate("/opt/so/saltstack/local/pillar/elasticsearch/nodes.sls")
def test_locate_skip_top(self):
with self.assertRaises(self.mod.SkipPath):
self.mod.locate("/opt/so/saltstack/local/pillar/top.sls")
def test_locate_skip_unrelated(self):
with self.assertRaises(self.mod.SkipPath):
self.mod.locate("/etc/hostname")
def test_pg_str_escapes(self):
self.assertEqual(self.mod._pg_str("a'b"), "'a''b'")
self.assertEqual(self.mod._pg_str(None), "NULL")
def test_conflict_target(self):
self.assertIn("scope='global'", self.mod._conflict_target("global"))
self.assertIn("scope='role'", self.mod._conflict_target("role"))
self.assertIn("scope='minion'", self.mod._conflict_target("minion"))
with self.assertRaises(ValueError):
self.mod._conflict_target("bogus")
def test_write_yaml_skips_disk_only_path(self):
with patch.object(self.mod, '_is_enabled', return_value=True):
ok, msg = self.mod.write_yaml(
"/opt/so/saltstack/local/pillar/secrets.sls",
{"secrets": {"foo": "bar"}})
self.assertFalse(ok)
self.assertIn("disk-only", msg)
def test_write_yaml_unreachable(self):
with patch.object(self.mod, '_is_enabled', return_value=False):
ok, msg = self.mod.write_yaml(
"/opt/so/saltstack/local/pillar/soc/soc_soc.sls",
{"soc": {"foo": "bar"}})
self.assertFalse(ok)
self.assertEqual(msg, "postgres unreachable")
def test_is_pg_managed_true(self):
self.assertTrue(self.mod.is_pg_managed(
"/opt/so/saltstack/local/pillar/minions/h1_sensor.sls"))
self.assertTrue(self.mod.is_pg_managed(
"/opt/so/saltstack/local/pillar/soc/soc_soc.sls"))
def test_is_pg_managed_false_for_bootstrap(self):
self.assertFalse(self.mod.is_pg_managed(
"/opt/so/saltstack/local/pillar/secrets.sls"))
self.assertFalse(self.mod.is_pg_managed(
"/opt/so/saltstack/local/pillar/postgres/auth.sls"))
self.assertFalse(self.mod.is_pg_managed(
"/opt/so/saltstack/local/pillar/elasticsearch/nodes.sls"))
def test_read_yaml_unreachable(self):
with patch.object(self.mod, '_is_enabled', return_value=False):
self.assertIsNone(self.mod.read_yaml(
"/opt/so/saltstack/local/pillar/soc/soc_soc.sls"))
def test_read_yaml_skips_disk_only(self):
with patch.object(self.mod, '_is_enabled', return_value=True):
with self.assertRaises(self.mod.SkipPath):
self.mod.read_yaml(
"/opt/so/saltstack/local/pillar/secrets.sls")
def test_read_yaml_returns_data(self):
with patch.object(self.mod, '_is_enabled', return_value=True):
with patch.object(self.mod, '_docker_psql',
return_value='{"soc": {"foo": "bar"}}\n'):
data = self.mod.read_yaml(
"/opt/so/saltstack/local/pillar/soc/soc_soc.sls")
self.assertEqual(data, {"soc": {"foo": "bar"}})
def test_read_yaml_returns_none_when_no_row(self):
with patch.object(self.mod, '_is_enabled', return_value=True):
with patch.object(self.mod, '_docker_psql', return_value=''):
data = self.mod.read_yaml(
"/opt/so/saltstack/local/pillar/soc/soc_soc.sls")
self.assertIsNone(data)
def test_read_yaml_minion_query_shape(self):
captured = {}
def fake_psql(sql):
captured['sql'] = sql
return '{"host": {"mainip": "10.0.0.1"}}'
with patch.object(self.mod, '_is_enabled', return_value=True):
with patch.object(self.mod, '_docker_psql', side_effect=fake_psql):
data = self.mod.read_yaml(
"/opt/so/saltstack/local/pillar/minions/h1_sensor.sls")
self.assertEqual(data, {"host": {"mainip": "10.0.0.1"}})
self.assertIn("scope='minion'", captured['sql'])
self.assertIn("'h1_sensor'", captured['sql'])
self.assertIn("'minions.h1_sensor'", captured['sql'])
def test_is_enabled_public_alias(self):
with patch.object(self.mod, '_is_enabled', return_value=True):
self.assertTrue(self.mod.is_enabled())
with patch.object(self.mod, '_is_enabled', return_value=False):
self.assertFalse(self.mod.is_enabled())
class TestSoYamlBackendMode(unittest.TestCase):
"""Tests so-yaml's backend-mode resolution and PG-canonical routing
for read/write/purge. The PG calls themselves are stubbed; what we're
asserting is that the right backend is chosen for each (mode, path)
combination."""
def test_resolve_mode_env_overrides_file(self):
with patch.dict('os.environ', {'SO_YAML_BACKEND': 'postgres'}):
self.assertEqual(soyaml._resolveBackendMode(), 'postgres')
with patch.dict('os.environ', {'SO_YAML_BACKEND': 'disk'}):
self.assertEqual(soyaml._resolveBackendMode(), 'disk')
def test_resolve_mode_invalid_env_falls_back(self):
with patch.dict('os.environ', {'SO_YAML_BACKEND': 'garbage'}, clear=False):
with patch('builtins.open', side_effect=IOError):
self.assertEqual(soyaml._resolveBackendMode(), 'dual')
def test_resolve_mode_default_dual(self):
env = {k: v for k, v in __import__('os').environ.items()
if k != 'SO_YAML_BACKEND'}
with patch.dict('os.environ', env, clear=True):
with patch('builtins.open', side_effect=IOError):
self.assertEqual(soyaml._resolveBackendMode(), 'dual')
def test_is_pg_managed_proxies(self):
with patch.object(soyaml, '_SO_YAML_PG_AVAILABLE', True):
self.assertTrue(soyaml._isPgManaged(
"/opt/so/saltstack/local/pillar/minions/h1_sensor.sls"))
self.assertFalse(soyaml._isPgManaged(
"/opt/so/saltstack/local/pillar/secrets.sls"))
def test_is_pg_managed_false_when_module_unavailable(self):
with patch.object(soyaml, '_SO_YAML_PG_AVAILABLE', False):
self.assertFalse(soyaml._isPgManaged(
"/opt/so/saltstack/local/pillar/minions/h1_sensor.sls"))
def test_load_yaml_postgres_mode_reads_pg(self):
with patch.object(soyaml, '_BACKEND_MODE', 'postgres'):
with patch.object(soyaml, '_SO_YAML_PG_AVAILABLE', True):
with patch.object(soyaml.so_yaml_postgres, 'is_pg_managed',
return_value=True):
with patch.object(soyaml.so_yaml_postgres, 'read_yaml',
return_value={"a": 1}):
result = soyaml.loadYaml(
"/opt/so/saltstack/local/pillar/soc/soc_soc.sls")
self.assertEqual(result, {"a": 1})
def test_load_yaml_postgres_mode_returns_empty_when_no_row(self):
with patch.object(soyaml, '_BACKEND_MODE', 'postgres'):
with patch.object(soyaml, '_SO_YAML_PG_AVAILABLE', True):
with patch.object(soyaml.so_yaml_postgres, 'is_pg_managed',
return_value=True):
with patch.object(soyaml.so_yaml_postgres, 'read_yaml',
return_value=None):
result = soyaml.loadYaml(
"/opt/so/saltstack/local/pillar/soc/soc_soc.sls")
self.assertEqual(result, {})
def test_load_yaml_postgres_mode_reads_disk_for_bootstrap(self):
import tempfile, os as _os
with tempfile.NamedTemporaryFile(mode='w', suffix='.yaml', delete=False) as f:
f.write("foo: bar\n")
tmp = f.name
try:
with patch.object(soyaml, '_BACKEND_MODE', 'postgres'):
with patch.object(soyaml, '_SO_YAML_PG_AVAILABLE', True):
with patch.object(soyaml.so_yaml_postgres,
'is_pg_managed', return_value=False):
result = soyaml.loadYaml(tmp)
self.assertEqual(result, {"foo": "bar"})
finally:
_os.unlink(tmp)
def test_write_yaml_postgres_mode_skips_disk(self):
import tempfile, os as _os
with tempfile.NamedTemporaryFile(mode='w', suffix='.yaml', delete=False) as f:
tmp = f.name
_os.unlink(tmp)
try:
with patch.object(soyaml, '_BACKEND_MODE', 'postgres'):
with patch.object(soyaml, '_SO_YAML_PG_AVAILABLE', True):
with patch.object(soyaml.so_yaml_postgres, 'is_pg_managed',
return_value=True):
with patch.object(soyaml.so_yaml_postgres, 'write_yaml',
return_value=(True, 'ok')) as mock_w:
soyaml.writeYaml(tmp, {"x": 1})
self.assertFalse(_os.path.exists(tmp))
mock_w.assert_called_once()
finally:
if _os.path.exists(tmp):
_os.unlink(tmp)
def test_write_yaml_postgres_mode_failure_is_fatal(self):
with patch.object(soyaml, '_BACKEND_MODE', 'postgres'):
with patch.object(soyaml, '_SO_YAML_PG_AVAILABLE', True):
with patch.object(soyaml.so_yaml_postgres, 'is_pg_managed',
return_value=True):
with patch.object(soyaml.so_yaml_postgres, 'write_yaml',
return_value=(False, 'pg write failed: connection refused')):
with patch('sys.exit', new=MagicMock()) as sysmock:
with patch('sys.stderr', new=StringIO()) as mock_err:
soyaml.writeYaml(
"/opt/so/saltstack/local/pillar/soc/soc_soc.sls",
{"x": 1})
sysmock.assert_called_with(1)
def test_write_yaml_disk_mode_skips_pg(self):
import tempfile, os as _os
with tempfile.NamedTemporaryFile(mode='w', suffix='.yaml', delete=False) as f:
tmp = f.name
try:
with patch.object(soyaml, '_BACKEND_MODE', 'disk'):
with patch.object(soyaml, '_SO_YAML_PG_AVAILABLE', True):
with patch.object(soyaml.so_yaml_postgres, 'write_yaml') as mock_w:
soyaml.writeYaml(tmp, {"x": 1})
mock_w.assert_not_called()
with open(tmp) as f:
self.assertIn('x: 1', f.read())
finally:
_os.unlink(tmp)
def test_purge_postgres_mode_calls_pg_only(self):
import tempfile, os as _os
with tempfile.NamedTemporaryFile(mode='w', suffix='.yaml', delete=False) as f:
tmp = f.name
_os.unlink(tmp)
with patch.object(soyaml, '_BACKEND_MODE', 'postgres'):
with patch.object(soyaml, '_SO_YAML_PG_AVAILABLE', True):
with patch.object(soyaml.so_yaml_postgres, 'is_pg_managed',
return_value=True):
with patch.object(soyaml.so_yaml_postgres, 'purge_yaml',
return_value=(True, 'ok')) as mock_p:
rc = soyaml.purgeFile(tmp)
self.assertEqual(rc, 0)
mock_p.assert_called_once()
def test_purge_postgres_mode_failure_returns_nonzero(self):
with patch.object(soyaml, '_BACKEND_MODE', 'postgres'):
with patch.object(soyaml, '_SO_YAML_PG_AVAILABLE', True):
with patch.object(soyaml.so_yaml_postgres, 'is_pg_managed',
return_value=True):
with patch.object(soyaml.so_yaml_postgres, 'purge_yaml',
return_value=(False, 'pg purge failed: x')):
with patch('sys.stderr', new=StringIO()):
rc = soyaml.purgeFile(
"/opt/so/saltstack/local/pillar/minions/h1_sensor.sls")
self.assertEqual(rc, 1)
-320
View File
@@ -1,320 +0,0 @@
# Copyright Security Onion Solutions LLC and/or licensed to Security Onion Solutions LLC under one
# or more contributor license agreements. Licensed under the Elastic License 2.0 as shown at
# https://securityonion.net/license; you may not use this file except in compliance with the
# Elastic License 2.0.
"""
so_yaml_postgres Postgres-backed dual-write helpers for so-yaml.py.
so-yaml.py writes YAML pillar files on disk; this module mirrors those
writes into so_pillar.* in so-postgres so ext_pillar and the SOC
PostgresConfigstore see the same data. During the postsalt transition
disk is canonical; PG writes are best-effort and never fail the disk
operation.
Connection: shells out to `docker exec so-postgres psql -U postgres -d
securityonion`. Same pattern so-pillar-import uses; avoids needing a
separate DSN config at install time. Performance is fine because so-yaml
is invoked from infrequent code paths (setup scripts, so-minion,
so-firewall); SOC's hot path uses the in-process pgxpool in
PostgresConfigstore, not so-yaml.
Path-to-row mapping mirrors PostgresConfigstore.locateSetting in
securityonion-soc:
/opt/so/saltstack/local/pillar/<section>/soc_<section>.sls
-> scope=global, pillar_path=<section>.soc_<section>
/opt/so/saltstack/local/pillar/<section>/adv_<section>.sls
-> scope=global, pillar_path=<section>.adv_<section>
/opt/so/saltstack/local/pillar/minions/<id>.sls
-> scope=minion, minion_id=<id>, pillar_path=minions.<id>
/opt/so/saltstack/local/pillar/minions/adv_<id>.sls
-> scope=minion, minion_id=<id>, pillar_path=minions.adv_<id>
Files outside that mapping (notably secrets.sls, postgres/auth.sls,
elasticsearch/nodes.sls, etc.) are skipped they stay disk-only forever
or render dynamically and don't belong in PG.
"""
import json
import os
import shlex
import subprocess
import sys
DOCKER_CONTAINER = os.environ.get("SO_PILLAR_PG_CONTAINER", "so-postgres")
PG_DATABASE = os.environ.get("SO_PILLAR_PG_DATABASE", "securityonion")
PG_USER = os.environ.get("SO_PILLAR_PG_USER", "postgres")
# File paths whose mutations stay disk-only forever. Mirrors EXCLUDE_*
# in so-pillar-import.
DISK_ONLY_PATHS = (
"/opt/so/saltstack/local/pillar/secrets.sls",
"/opt/so/saltstack/local/pillar/postgres/auth.sls",
"/opt/so/saltstack/local/pillar/elasticsearch/auth.sls",
"/opt/so/saltstack/local/pillar/kibana/secrets.sls",
)
DISK_ONLY_FRAGMENTS = (
"/elasticsearch/nodes.sls",
"/redis/nodes.sls",
"/kafka/nodes.sls",
"/hypervisor/nodes.sls",
"/logstash/nodes.sls",
"/node_data/ips.sls",
"/top.sls",
)
class SkipPath(Exception):
"""Raised when a file path is intentionally not mirrored to PG."""
def is_enabled():
"""Public alias for callers that want to probe PG reachability without
relying on a leading-underscore private name."""
return _is_enabled()
def _is_enabled():
"""PG dual-write only fires if so-postgres is reachable. Cheap probe.
Returns True when docker exec succeeds, False otherwise. We never
want a PG hiccup to fail a disk write on a manager whose Postgres is
momentarily unreachable."""
try:
proc = subprocess.run(
["docker", "exec", DOCKER_CONTAINER,
"pg_isready", "-h", "127.0.0.1", "-U", PG_USER, "-q"],
capture_output=True, timeout=5, check=False,
)
return proc.returncode == 0
except (FileNotFoundError, subprocess.TimeoutExpired, OSError):
return False
def locate(path):
"""Translate a so-yaml file path to (scope, role_name, minion_id, pillar_path).
Raises SkipPath when the file is not part of the PG-managed surface."""
norm = os.path.normpath(path)
if norm in DISK_ONLY_PATHS:
raise SkipPath(f"{path}: explicit disk-only allowlist")
for frag in DISK_ONLY_FRAGMENTS:
if frag in norm:
raise SkipPath(f"{path}: matches disk-only fragment {frag}")
parent = os.path.basename(os.path.dirname(norm))
grandparent = os.path.basename(os.path.dirname(os.path.dirname(norm)))
name = os.path.basename(norm)
if not name.endswith(".sls"):
raise SkipPath(f"{path}: not a .sls file")
stem = name[:-4]
if parent == "minions":
if stem.startswith("adv_"):
mid = stem[4:]
return ("minion", None, mid, f"minions.adv_{mid}")
return ("minion", None, stem, f"minions.{stem}")
# /local/pillar/<section>/<file>.sls
if grandparent == "pillar" and parent and parent != "":
if stem.startswith("soc_") or stem.startswith("adv_"):
return ("global", None, None, f"{parent}.{stem}")
raise SkipPath(f"{path}: <section>/{stem}.sls is not a soc_/adv_ file")
raise SkipPath(f"{path}: unrecognised pillar layout")
def _pg_str(s):
if s is None:
return "NULL"
return "'" + str(s).replace("'", "''") + "'"
def _docker_psql(sql):
"""Run sql via docker exec ... psql. Returns stdout. Caller catches
exceptions and downgrades to a warning."""
proc = subprocess.run(
["docker", "exec", "-i", DOCKER_CONTAINER,
"psql", "-U", PG_USER, "-d", PG_DATABASE,
"-tA", "-q", "-v", "ON_ERROR_STOP=1"],
input=sql.encode(), capture_output=True, check=False, timeout=30,
)
if proc.returncode != 0:
raise RuntimeError(proc.stderr.decode(errors="replace") or
f"docker exec psql exit {proc.returncode}")
return proc.stdout.decode(errors="replace")
def _conflict_target(scope):
if scope == "global":
return "(pillar_path) WHERE scope='global'"
if scope == "role":
return "(role_name, pillar_path) WHERE scope='role'"
if scope == "minion":
return "(minion_id, pillar_path) WHERE scope='minion'"
raise ValueError(f"unknown scope {scope!r}")
def is_pg_managed(path):
"""True if this path maps to a so_pillar.* row (locate() succeeds).
Bootstrap and mine-driven files return False they always live on
disk regardless of so-yaml's backend mode."""
try:
locate(path)
return True
except SkipPath:
return False
def read_yaml(path):
"""Return the content dict stored in so_pillar.pillar_entry for `path`,
or None when no row exists. Raises SkipPath when `path` is not part of
the PG-managed surface (caller should read disk in that case).
Used by so-yaml.py PG-canonical mode so `replace`, `get`, etc. resolve
against the database rather than a stale (or absent) disk file."""
if not _is_enabled():
return None
scope, role, minion_id, pillar_path = locate(path)
if scope == "minion":
sql = ("SELECT data FROM so_pillar.pillar_entry "
"WHERE scope='minion' "
f"AND minion_id={_pg_str(minion_id)} "
f"AND pillar_path={_pg_str(pillar_path)}")
elif scope == "role":
sql = ("SELECT data FROM so_pillar.pillar_entry "
"WHERE scope='role' "
f"AND role_name={_pg_str(role)} "
f"AND pillar_path={_pg_str(pillar_path)}")
else:
sql = ("SELECT data FROM so_pillar.pillar_entry "
"WHERE scope='global' "
f"AND pillar_path={_pg_str(pillar_path)}")
try:
out = _docker_psql(sql).strip()
except Exception:
return None
if not out:
return None
try:
return json.loads(out)
except (ValueError, TypeError):
return None
def write_yaml(path, content_dict, *, reason="so-yaml dual-write"):
"""Mirror the disk write at `path` (whose content was just rendered as
`content_dict`) into so_pillar.pillar_entry. Best-effort: any failure
is swallowed so the caller (so-yaml.py) does not see it as a fatal."""
if not _is_enabled():
return False, "postgres unreachable"
try:
scope, role, minion_id, pillar_path = locate(path)
except SkipPath as e:
return False, str(e)
data_json = json.dumps(content_dict if content_dict is not None else {})
role_sql = _pg_str(role)
minion_sql = _pg_str(minion_id)
reason_sql = _pg_str(reason)
conflict = _conflict_target(scope)
sql_parts = []
if scope == "minion":
# FK requires the minion row before pillar_entry can reference it.
sql_parts.append(
f"INSERT INTO so_pillar.minion (minion_id) VALUES ({minion_sql}) "
"ON CONFLICT (minion_id) DO NOTHING;"
)
sql_parts.append(
"BEGIN;\n"
f"SELECT set_config('so_pillar.change_reason', {reason_sql}, true);\n"
"INSERT INTO so_pillar.pillar_entry "
"(scope, role_name, minion_id, pillar_path, data, change_reason) "
f"VALUES ({_pg_str(scope)}, {role_sql}, {minion_sql}, "
f"{_pg_str(pillar_path)}, {_pg_str(data_json)}::jsonb, {reason_sql}) "
f"ON CONFLICT {conflict} DO UPDATE "
"SET data = EXCLUDED.data, change_reason = EXCLUDED.change_reason;\n"
"COMMIT;\n"
)
try:
_docker_psql("\n".join(sql_parts))
except Exception as e:
return False, f"pg write failed: {e}"
return True, "ok"
def purge_yaml(path, *, reason="so-yaml purge"):
"""Mirror the disk file deletion at `path` by deleting the matching
pillar_entry rows. For minion files also deletes the so_pillar.minion
row (CASCADE removes pillar_entry + role_member rows)."""
if not _is_enabled():
return False, "postgres unreachable"
try:
scope, role, minion_id, pillar_path = locate(path)
except SkipPath as e:
return False, str(e)
reason_sql = _pg_str(reason)
parts = ["BEGIN;",
f"SELECT set_config('so_pillar.change_reason', {reason_sql}, true);"]
if scope == "minion":
# If both <id>.sls and adv_<id>.sls are gone the trigger / CASCADE
# cleans up role_member; otherwise we just remove this one row.
parts.append(
f"DELETE FROM so_pillar.pillar_entry "
f"WHERE scope='minion' AND minion_id={_pg_str(minion_id)} "
f"AND pillar_path={_pg_str(pillar_path)};"
)
parts.append(
f"DELETE FROM so_pillar.minion WHERE minion_id={_pg_str(minion_id)} "
"AND NOT EXISTS (SELECT 1 FROM so_pillar.pillar_entry "
f"WHERE minion_id={_pg_str(minion_id)});"
)
else:
parts.append(
f"DELETE FROM so_pillar.pillar_entry "
f"WHERE scope={_pg_str(scope)} AND pillar_path={_pg_str(pillar_path)};"
)
parts.append("COMMIT;")
try:
_docker_psql("\n".join(parts))
except Exception as e:
return False, f"pg purge failed: {e}"
return True, "ok"
# CLI for diagnostics. Not exercised by so-yaml.py itself.
def _main(argv):
import argparse
ap = argparse.ArgumentParser()
ap.add_argument("op", choices=("locate", "ping"))
ap.add_argument("path", nargs="?")
args = ap.parse_args(argv)
if args.op == "ping":
ok = _is_enabled()
print("ok" if ok else "unreachable")
return 0 if ok else 1
if args.op == "locate":
if not args.path:
ap.error("locate requires PATH")
try:
scope, role, minion_id, pillar_path = locate(args.path)
print(f"scope={scope} role={role} minion_id={minion_id} pillar_path={pillar_path}")
return 0
except SkipPath as e:
print(f"SKIP: {e}", file=sys.stderr)
return 2
return 1
if __name__ == "__main__":
sys.exit(_main(sys.argv[1:]))
+9 -193
View File
@@ -24,14 +24,6 @@ BACKUPTOPFILE=/opt/so/saltstack/default/salt/top.sls.backup
SALTUPGRADED=false
SALT_CLOUD_INSTALLED=false
SALT_CLOUD_CONFIGURED=false
# Check if salt-cloud is installed
if rpm -q salt-cloud &>/dev/null; then
SALT_CLOUD_INSTALLED=true
fi
# Check if salt-cloud is configured
if [[ -f /etc/salt/cloud.profiles.d/socloud.conf ]]; then
SALT_CLOUD_CONFIGURED=true
fi
# used to display messages to the user at the end of soup
declare -a FINAL_MESSAGE_QUEUE=()
@@ -485,168 +477,7 @@ elasticsearch_backup_index_templates() {
tar -czf /nsm/backup/3.0.0_elasticsearch_index_templates.tar.gz -C /opt/so/conf/elasticsearch/templates/index/ .
}
elasticfleet_set_agent_logging_level_warn() {
. /usr/sbin/so-elastic-fleet-common
local current_agent_policies
if ! current_agent_policies=$(fleet_api "agent_policies?perPage=1000"); then
echo "Warning: unable to retrieve Fleet agent policies"
return 0
fi
# Only updating policies that are within Security Onion defaults and do not already have any user configured advanced_settings.
local policies_to_update
policies_to_update=$(jq -c '
.items[]
| select(has("advanced_settings") | not)
| select(
.id == "so-grid-nodes_general"
or .id == "so-grid-nodes_heavy"
or .id == "endpoints-initial"
or (.id | startswith("FleetServer_"))
)
' <<< "$current_agent_policies")
if [[ -z "$policies_to_update" ]]; then
return 0
fi
while IFS= read -r policy; do
[[ -z "$policy" ]] && continue
local policy_id policy_name policy_namespace
policy_id=$(jq -r '.id' <<< "$policy")
policy_name=$(jq -r '.name' <<< "$policy")
policy_namespace=$(jq -r '.namespace' <<< "$policy")
local update_logging
update_logging=$(jq -n \
--arg name "$policy_name" \
--arg namespace "$policy_namespace" \
'{name: $name, namespace: $namespace, advanced_settings: {agent_logging_level: "warning"}}'
)
echo "Setting elastic agent_logging_level to warning on policy '$policy_name' ($policy_id)."
if ! fleet_api "agent_policies/$policy_id" -XPUT -H 'kbn-xsrf: true' -H 'Content-Type: application/json' -d "$update_logging" >/dev/null; then
echo " warning: failed to update agent policy '$policy_name' ($policy_id)" >&2
fi
done <<< "$policies_to_update"
}
check_transform_health_and_reauthorize() {
. /usr/sbin/so-elastic-fleet-common
echo "Checking integration transform jobs for unhealthy / unauthorized status..."
local transforms_doc stats_doc installed_doc
if ! transforms_doc=$(so-elasticsearch-query "_transform/_all?size=1000" --fail --retry 3 --retry-delay 5 2>/dev/null); then
echo "Unable to query for transform jobs, skipping reauthorization."
return 0
fi
if ! stats_doc=$(so-elasticsearch-query "_transform/_all/_stats?size=1000" --fail --retry 3 --retry-delay 5 2>/dev/null); then
echo "Unable to query for transform job stats, skipping reauthorization."
return 0
fi
if ! installed_doc=$(fleet_api "epm/packages/installed?perPage=500"); then
echo "Unable to list installed Fleet packages, skipping reauthorization."
return 0
fi
# Get all transforms that meet the following
# - unhealthy (any non-green health status)
# - metadata has run_as_kibana_system: false (this fix is specific to transforms started prior to Kibana 9.3.3)
# - are not orphaned (integration is not somehow missing/corrupt/uninstalled)
local unhealthy_transforms
unhealthy_transforms=$(jq -c -n \
--argjson t "$transforms_doc" \
--argjson s "$stats_doc" \
--argjson i "$installed_doc" '
($i.items | map({key: .name, value: .version}) | from_entries) as $pkg_ver
| ($s.transforms | map({key: .id, value: .health.status}) | from_entries) as $health
| [ $t.transforms[]
| select(._meta.run_as_kibana_system == false)
| select(($health[.id] // "unknown") != "green")
| {id, pkg: ._meta.package.name, ver: ($pkg_ver[._meta.package.name])}
]
| if length == 0 then empty else . end
| (map(select(.ver == null)) | map({orphan: .id})[]),
(map(select(.ver != null))
| group_by(.pkg)
| map({pkg: .[0].pkg, ver: .[0].ver, transformIds: map(.id)})[])
')
if [[ -z "$unhealthy_transforms" ]]; then
return 0
fi
local unhealthy_count
unhealthy_count=$(jq -s '[.[].transformIds? // empty | .[]] | length' <<< "$unhealthy_transforms")
echo "Found $unhealthy_count transform(s) needing reauthorization."
local total_failures=0
while IFS= read -r transform; do
[[ -z "$transform" ]] && continue
if jq -e 'has("orphan")' <<< "$transform" >/dev/null 2>&1; then
echo "Skipping transform not owned by any installed Fleet package: $(jq -r '.orphan' <<< "$transform")"
continue
fi
local pkg ver body resp
pkg=$(jq -r '.pkg' <<< "$transform")
ver=$(jq -r '.ver' <<< "$transform")
body=$(jq -c '{transforms: (.transformIds | map({transformId: .}))}' <<< "$transform")
echo "Reauthorizing transform(s) for ${pkg}-${ver}..."
resp=$(fleet_api "epm/packages/${pkg}/${ver}/transforms/authorize" \
-XPOST -H 'kbn-xsrf: true' -H 'Content-Type: application/json' \
-d "$body") || { echo "Could not reauthorize transform(s) for ${pkg}-${ver}"; continue; }
(( total_failures += $(jq 'map(select(.success != true)) | length' <<< "$resp" 2>/dev/null) ))
done <<< "$unhealthy_transforms"
if [[ "$total_failures" -gt 0 ]]; then
echo "Some transform(s) failed to reauthorize."
fi
}
ensure_postgres_local_pillar() {
# Postgres was added as a service after 3.0.0, so the new pillar/top.sls
# references postgres.soc_postgres / postgres.adv_postgres unconditionally.
# Managers upgrading from 3.0.0 have no /opt/so/saltstack/local/pillar/postgres/
# (make_some_dirs only runs at install time), so the stubs must be created
# here before salt-master restarts against the new top.sls.
echo "Ensuring postgres local pillar stubs exist."
local dir=/opt/so/saltstack/local/pillar/postgres
mkdir -p "$dir"
[[ -f "$dir/soc_postgres.sls" ]] || touch "$dir/soc_postgres.sls"
[[ -f "$dir/adv_postgres.sls" ]] || touch "$dir/adv_postgres.sls"
chown -R socore:socore "$dir"
}
ensure_postgres_secret() {
# On a fresh install, generate_passwords + secrets_pillar seed
# secrets:postgres_pass in /opt/so/saltstack/local/pillar/secrets.sls. That
# code path is skipped on upgrade (secrets.sls already exists from 3.0.0
# with import_pass/influx_pass but no postgres_pass), so the postgres
# container's POSTGRES_PASSWORD_FILE and SOC's PG_ADMIN_PASS would be empty
# after highstate. Generate one now if missing.
local secrets_file=/opt/so/saltstack/local/pillar/secrets.sls
if [[ ! -f "$secrets_file" ]]; then
echo "WARNING: $secrets_file missing; skipping postgres_pass backfill."
return 0
fi
if so-yaml.py get -r "$secrets_file" secrets.postgres_pass >/dev/null 2>&1; then
echo "secrets.postgres_pass already set; leaving as-is."
return 0
fi
echo "Seeding secrets.postgres_pass in $secrets_file."
so-yaml.py add "$secrets_file" secrets.postgres_pass "$(get_random_value)"
chown socore:socore "$secrets_file"
}
up_to_3.1.0() {
ensure_postgres_local_pillar
ensure_postgres_secret
determine_elastic_agent_upgrade
elasticsearch_backup_index_templates
# Clear existing component template state file.
@@ -658,30 +489,6 @@ up_to_3.1.0() {
post_to_3.1.0() {
/usr/sbin/so-kibana-space-defaults
# ensure manager has new version of socloud.conf
if [[ $SALT_CLOUD_CONFIGURED == true ]]; then
salt-call state.apply salt.cloud.config concurrent=True
fi
# Backfill the Telegraf creds pillar for every accepted minion. so-telegraf-cred
# add is idempotent — it no-ops when an entry already exists — so this is safe
# to run on every soup. The subsequent state.apply creates/updates the matching
# Postgres roles from the reconciled pillar.
echo "Reconciling Telegraf Postgres creds for accepted minions."
for mid in $(salt-key --out=json --list=accepted 2>/dev/null | jq -r '.minions[]?' 2>/dev/null); do
[[ -n "$mid" ]] || continue
/usr/sbin/so-telegraf-cred add "$mid" || echo " warning: so-telegraf-cred add $mid failed" >&2
done
# Run through the master (not --local) so state compilation uses the
# master's configured file_roots; the manager's /etc/salt/minion has no
# file_roots of its own and --local would fail with "No matching sls found".
salt-call state.apply postgres.telegraf_users queue=True || true
# Update default agent policies to use logging level warn.
elasticfleet_set_agent_logging_level_warn || true
# Check for unhealthy / unauthorized integration transform jobs and attempt reauthorizations
check_transform_health_and_reauthorize || true
POSTVERSION=3.1.0
}
@@ -856,6 +663,15 @@ upgrade_check_salt() {
upgrade_salt() {
echo "Performing upgrade of Salt from $INSTALLEDSALTVERSION to $NEWSALTVERSION."
echo ""
# Check if salt-cloud is installed
if rpm -q salt-cloud &>/dev/null; then
SALT_CLOUD_INSTALLED=true
fi
# Check if salt-cloud is configured
if [[ -f /etc/salt/cloud.profiles.d/socloud.conf ]]; then
SALT_CLOUD_CONFIGURED=true
fi
echo "Removing yum versionlock for Salt."
echo ""
yum versionlock delete "salt"
-25
View File
@@ -25,33 +25,8 @@ manager_run_es_soc:
- salt: {{NEWNODE}}_update_mine
{% endif %}
# so-minion has already added the new minion's entry to telegraf/creds.sls
# via so-telegraf-cred before this orch fires. Reconcile the Postgres role
# on the manager so the new minion can authenticate on its first highstate,
# then refresh the minion's pillar so its telegraf.conf renders with the
# freshly-written cred.
manager_create_postgres_telegraf_role:
salt.state:
- tgt: {{ MANAGER }}
- sls:
- postgres.telegraf_users
- queue: True
- require:
- salt: {{NEWNODE}}_update_mine
{{NEWNODE}}_refresh_pillar:
salt.function:
- name: saltutil.refresh_pillar
- tgt: {{ NEWNODE }}
- kwarg:
wait: True
- require:
- salt: manager_create_postgres_telegraf_role
{{NEWNODE}}_run_highstate:
salt.state:
- tgt: {{ NEWNODE }}
- highstate: True
- queue: True
- require:
- salt: {{NEWNODE}}_refresh_pillar
-112
View File
@@ -1,112 +0,0 @@
# Copyright Security Onion Solutions LLC and/or licensed to Security Onion Solutions LLC under one
# or more contributor license agreements. Licensed under the Elastic License 2.0 as shown at
# https://securityonion.net/license; you may not use this file except in compliance with the
# Elastic License 2.0.
# Driven by the so_pillar_changed reactor. Translates a so_pillar.pillar_entry
# change into (cache.clear_pillar -> saltutil.refresh_pillar -> state.apply)
# on the appropriate target.
#
# Routing rules live in the DISPATCH map below — one entry per
# (pillar_path prefix) -> (state sls, role grain). Add new services here
# rather than wiring more reactors.
#
# Idempotent: state.apply is idempotent; if the pillar value didn't actually
# change anything observable, the affected state runs a no-op. Bulk imports
# and replays are safe.
{% set change = salt['pillar.get']('so_pillar_change', {}) %}
{% set scope = change.get('scope') %}
{% set role = change.get('role_name') %}
{% set minion = change.get('minion_id') %}
{% set changes = change.get('changes', []) %}
{# (pillar_path prefix) -> {sls: <state to apply>, role: <role grain that runs it>}
role is a grain value (e.g. 'so-sensor'), used to compute compound targets
when the change is global or role-scoped. #}
{% set DISPATCH = {
'suricata.': {'sls': 'suricata.config', 'roles': ['so-sensor', 'so-heavynode', 'so-standalone']},
'sensor.': {'sls': 'suricata.config', 'roles': ['so-sensor', 'so-heavynode', 'so-standalone']},
'zeek.': {'sls': 'zeek.config', 'roles': ['so-sensor', 'so-heavynode', 'so-standalone']},
'stenographer.': {'sls': 'stenographer.config', 'roles': ['so-sensor', 'so-heavynode', 'so-standalone']},
'pcap.': {'sls': 'pcap.config', 'roles': ['so-sensor', 'so-heavynode', 'so-standalone']},
'logstash.': {'sls': 'logstash.config', 'roles': ['so-manager', 'so-managersearch', 'so-managerhype', 'so-receiver']},
'redis.': {'sls': 'redis.config', 'roles': ['so-manager', 'so-managersearch', 'so-managerhype', 'so-standalone']},
'kafka.': {'sls': 'kafka.config', 'roles': ['so-manager', 'so-managersearch', 'so-managerhype', 'so-receiver', 'so-searchnode']},
'elasticsearch.': {'sls': 'elasticsearch.config','roles': ['so-manager', 'so-managersearch', 'so-managerhype', 'so-searchnode', 'so-heavynode', 'so-standalone']},
'kibana.': {'sls': 'kibana.config', 'roles': ['so-manager', 'so-managersearch', 'so-managerhype', 'so-standalone']},
'soc.': {'sls': 'soc.config', 'roles': ['so-manager', 'so-managersearch', 'so-managerhype', 'so-standalone']},
'telegraf.': {'sls': 'telegraf.config', 'roles': ['*']},
'fleet.': {'sls': 'fleet.config', 'roles': ['so-fleet']},
'strelka.': {'sls': 'strelka.config', 'roles': ['so-sensor', 'so-heavynode', 'so-standalone']},
} %}
{# Collect a deduplicated set of (sls, target_kind) actions. target_kind is
either 'minion:<id>' (scope=minion) or 'roles:so-x,so-y' (scope=role/global). #}
{% set actions = {} %}
{% for c in changes %}
{% set path = c.get('pillar_path', '') %}
{% for prefix, action in DISPATCH.items() %}
{% if path.startswith(prefix) %}
{% set sls = action['sls'] %}
{% if scope == 'minion' and minion %}
{% set key = sls ~ '|minion|' ~ minion %}
{% set _ = actions.update({key: {'sls': sls, 'tgt': minion, 'tgt_type': 'glob'}}) %}
{% else %}
{% set role_targets = action['roles'] %}
{% if '*' in role_targets %}
{% set tgt = '*' %}
{% set tgt_type = 'glob' %}
{% else %}
{% set tgt = ('I@role:' ~ role_targets|join(' or I@role:')) %}
{% set tgt_type = 'compound' %}
{% endif %}
{% set key = sls ~ '|' ~ tgt %}
{% set _ = actions.update({key: {'sls': sls, 'tgt': tgt, 'tgt_type': tgt_type}}) %}
{% endif %}
{% endif %}
{% endfor %}
{% endfor %}
{% if actions %}
{% for key, action in actions.items() %}
{% set safe_id = loop.index0 | string %}
so_pillar_reload_clear_cache_{{ safe_id }}:
salt.runner:
- name: cache.clear_pillar
- tgt: '{{ action.tgt }}'
- tgt_type: '{{ action.tgt_type }}'
so_pillar_reload_refresh_pillar_{{ safe_id }}:
salt.function:
- name: saltutil.refresh_pillar
- tgt: '{{ action.tgt }}'
- tgt_type: '{{ action.tgt_type }}'
- kwarg:
wait: True
- require:
- salt: so_pillar_reload_clear_cache_{{ safe_id }}
so_pillar_reload_apply_state_{{ safe_id }}:
salt.state:
- tgt: '{{ action.tgt }}'
- tgt_type: '{{ action.tgt_type }}'
- sls:
- {{ action.sls }}
- queue: True
- require:
- salt: so_pillar_reload_refresh_pillar_{{ safe_id }}
{% endfor %}
{% else %}
{# No DISPATCH entry matched. Pillar still gets refreshed so any other states
read fresh values, but no service-specific reload is invoked. #}
so_pillar_reload_unmapped_path_noop:
test.nop
{% do salt.log.info('orch.so_pillar_reload: no dispatch match for %s' % changes) %}
{% endif %}
-37
View File
@@ -1,37 +0,0 @@
# Copyright Security Onion Solutions LLC and/or licensed to Security Onion Solutions LLC under one
# or more contributor license agreements. Licensed under the Elastic License 2.0 as shown at
# https://securityonion.net/license; you may not use this file except in compliance with the
# Elastic License 2.0.
{% from 'allowed_states.map.jinja' import allowed_states %}
{% if sls in allowed_states %}
{% set DIGITS = "1234567890" %}
{% set LOWERCASE = "qwertyuiopasdfghjklzxcvbnm" %}
{% set UPPERCASE = "QWERTYUIOPASDFGHJKLZXCVBNM" %}
{% set SYMBOLS = "~!@#^&*()-_=+[]|;:,.<>?" %}
{% set CHARS = DIGITS~LOWERCASE~UPPERCASE~SYMBOLS %}
{% set so_postgres_user_pass = salt['pillar.get']('postgres:auth:users:so_postgres_user:pass', salt['random.get_str'](72, chars=CHARS)) %}
# Admin cred only. Per-minion Telegraf creds live in telegraf/creds.sls,
# managed by /usr/sbin/so-telegraf-cred (called from so-minion).
postgres_auth_pillar:
file.managed:
- name: /opt/so/saltstack/local/pillar/postgres/auth.sls
- mode: 640
- reload_pillar: True
- contents: |
postgres:
auth:
users:
so_postgres_user:
user: so_postgres
pass: "{{ so_postgres_user_pass }}"
- show_changes: False
{% else %}
{{sls}}_state_not_allowed:
test.fail_without_changes:
- name: {{sls}}_state_not_allowed
{% endif %}
-111
View File
@@ -1,111 +0,0 @@
# Copyright Security Onion Solutions LLC and/or licensed to Security Onion Solutions LLC under one
# or more contributor license agreements. Licensed under the Elastic License 2.0 as shown at
# https://securityonion.net/license; you may not use this file except in compliance with the
# Elastic License 2.0.
{% from 'allowed_states.map.jinja' import allowed_states %}
{% if sls.split('.')[0] in allowed_states %}
{% from 'postgres/map.jinja' import PGMERGED %}
# Postgres Setup
postgresconfdir:
file.directory:
- name: /opt/so/conf/postgres
- user: 939
- group: 939
- makedirs: True
postgressecretsdir:
file.directory:
- name: /opt/so/conf/postgres/secrets
- user: 939
- group: 939
- mode: 700
- require:
- file: postgresconfdir
postgresdatadir:
file.directory:
- name: /nsm/postgres
- user: 939
- group: 939
- makedirs: True
postgreslogdir:
file.directory:
- name: /opt/so/log/postgres
- user: 939
- group: 939
- makedirs: True
postgresinitdir:
file.directory:
- name: /opt/so/conf/postgres/init
- user: 939
- group: 939
- require:
- file: postgresconfdir
postgresinitusers:
file.managed:
- name: /opt/so/conf/postgres/init/init-users.sh
- source: salt://postgres/files/init-users.sh
- user: 939
- group: 939
- mode: 755
postgresconf:
file.managed:
- name: /opt/so/conf/postgres/postgresql.conf
- source: salt://postgres/files/postgresql.conf.jinja
- user: 939
- group: 939
- template: jinja
- defaults:
PGMERGED: {{ PGMERGED }}
postgreshba:
file.managed:
- name: /opt/so/conf/postgres/pg_hba.conf
- source: salt://postgres/files/pg_hba.conf
- user: 939
- group: 939
- mode: 640
postgres_super_secret:
file.managed:
- name: /opt/so/conf/postgres/secrets/postgres_password
- user: 939
- group: 939
- mode: 600
- contents_pillar: 'secrets:postgres_pass'
- show_changes: False
- require:
- file: postgressecretsdir
postgres_app_secret:
file.managed:
- name: /opt/so/conf/postgres/secrets/so_postgres_pass
- user: 939
- group: 939
- mode: 600
- contents_pillar: 'postgres:auth:users:so_postgres_user:pass'
- show_changes: False
- require:
- file: postgressecretsdir
postgres_sbin:
file.recurse:
- name: /usr/sbin
- source: salt://postgres/tools/sbin
- user: root
- group: root
- file_mode: 755
{% else %}
{{sls}}_state_not_allowed:
test.fail_without_changes:
- name: {{sls}}_state_not_allowed
{% endif %}
-19
View File
@@ -1,19 +0,0 @@
postgres:
enabled: True
telegraf:
retention_days: 14
config:
listen_addresses: '*'
port: 5432
max_connections: 100
shared_buffers: 256MB
ssl: 'on'
ssl_cert_file: '/conf/postgres.crt'
ssl_key_file: '/conf/postgres.key'
ssl_ca_file: '/conf/ca.crt'
hba_file: '/conf/pg_hba.conf'
log_destination: 'stderr'
logging_collector: 'off'
log_min_messages: 'warning'
shared_preload_libraries: pg_cron
cron.database_name: so_telegraf
-33
View File
@@ -1,33 +0,0 @@
# Copyright Security Onion Solutions LLC and/or licensed to Security Onion Solutions LLC under one
# or more contributor license agreements. Licensed under the Elastic License 2.0 as shown at
# https://securityonion.net/license; you may not use this file except in compliance with the
# Elastic License 2.0.
{% from 'allowed_states.map.jinja' import allowed_states %}
{% if sls.split('.')[0] in allowed_states %}
include:
- postgres.sostatus
so-postgres:
docker_container.absent:
- force: True
so-postgres_so-status.disabled:
file.comment:
- name: /opt/so/conf/so-status/so-status.conf
- regex: ^so-postgres$
so_postgres_backup:
cron.absent:
- name: /usr/sbin/so-postgres-backup > /dev/null 2>&1
- identifier: so_postgres_backup
- user: root
{% else %}
{{sls}}_state_not_allowed:
test.fail_without_changes:
- name: {{sls}}_state_not_allowed
{% endif %}
-109
View File
@@ -1,109 +0,0 @@
# Copyright Security Onion Solutions LLC and/or licensed to Security Onion Solutions LLC under one
# or more contributor license agreements. Licensed under the Elastic License 2.0 as shown at
# https://securityonion.net/license; you may not use this file except in compliance with the
# Elastic License 2.0.
{% from 'allowed_states.map.jinja' import allowed_states %}
{% if sls.split('.')[0] in allowed_states %}
{% from 'vars/globals.map.jinja' import GLOBALS %}
{% from 'docker/docker.map.jinja' import DOCKERMERGED %}
{% set SO_POSTGRES_USER = salt['pillar.get']('postgres:auth:users:so_postgres_user:user', 'so_postgres') %}
include:
- postgres.auth
- postgres.ssl
- postgres.config
- postgres.sostatus
- postgres.telegraf_users
so-postgres:
docker_container.running:
- image: {{ GLOBALS.registry_host }}:5000/{{ GLOBALS.image_repo }}/so-postgres:{{ GLOBALS.so_version }}
- hostname: so-postgres
- networks:
- sobridge:
- ipv4_address: {{ DOCKERMERGED.containers['so-postgres'].ip }}
- port_bindings:
{% for BINDING in DOCKERMERGED.containers['so-postgres'].port_bindings %}
- {{ BINDING }}
{% endfor %}
- environment:
- POSTGRES_DB=securityonion
# Passwords are delivered via mounted 0600 secret files, not plaintext env vars.
# The upstream postgres image resolves POSTGRES_PASSWORD_FILE; entrypoint.sh and
# init-users.sh resolve SO_POSTGRES_PASS_FILE the same way.
- POSTGRES_PASSWORD_FILE=/run/secrets/postgres_password
- SO_POSTGRES_USER={{ SO_POSTGRES_USER }}
- SO_POSTGRES_PASS_FILE=/run/secrets/so_postgres_pass
{% if DOCKERMERGED.containers['so-postgres'].extra_env %}
{% for XTRAENV in DOCKERMERGED.containers['so-postgres'].extra_env %}
- {{ XTRAENV }}
{% endfor %}
{% endif %}
- binds:
- /opt/so/log/postgres/:/log:rw
- /nsm/postgres:/var/lib/postgresql/data:rw
- /opt/so/conf/postgres/postgresql.conf:/conf/postgresql.conf:ro
- /opt/so/conf/postgres/pg_hba.conf:/conf/pg_hba.conf:ro
- /opt/so/conf/postgres/secrets:/run/secrets:ro
- /opt/so/conf/postgres/init/init-users.sh:/docker-entrypoint-initdb.d/init-users.sh:ro
- /etc/pki/postgres.crt:/conf/postgres.crt:ro
- /etc/pki/postgres.key:/conf/postgres.key:ro
- /etc/pki/tls/certs/intca.crt:/conf/ca.crt:ro
{% if DOCKERMERGED.containers['so-postgres'].custom_bind_mounts %}
{% for BIND in DOCKERMERGED.containers['so-postgres'].custom_bind_mounts %}
- {{ BIND }}
{% endfor %}
{% endif %}
{% if DOCKERMERGED.containers['so-postgres'].extra_hosts %}
- extra_hosts:
{% for XTRAHOST in DOCKERMERGED.containers['so-postgres'].extra_hosts %}
- {{ XTRAHOST }}
{% endfor %}
{% endif %}
{% if DOCKERMERGED.containers['so-postgres'].ulimits %}
- ulimits:
{% for ULIMIT in DOCKERMERGED.containers['so-postgres'].ulimits %}
- {{ ULIMIT.name }}={{ ULIMIT.soft }}:{{ ULIMIT.hard }}
{% endfor %}
{% endif %}
- watch:
- file: postgresconf
- file: postgreshba
- file: postgresinitusers
- file: postgres_super_secret
- file: postgres_app_secret
- x509: postgres_crt
- x509: postgres_key
- require:
- file: postgresconf
- file: postgreshba
- file: postgresinitusers
- file: postgres_super_secret
- file: postgres_app_secret
- x509: postgres_crt
- x509: postgres_key
delete_so-postgres_so-status.disabled:
file.uncomment:
- name: /opt/so/conf/so-status/so-status.conf
- regex: ^so-postgres$
so_postgres_backup:
cron.present:
- name: /usr/sbin/so-postgres-backup > /dev/null 2>&1
- identifier: so_postgres_backup
- user: root
- minute: '5'
- hour: '0'
- daymonth: '*'
- month: '*'
- dayweek: '*'
{% else %}
{{sls}}_state_not_allowed:
test.fail_without_changes:
- name: {{sls}}_state_not_allowed
{% endif %}
-34
View File
@@ -1,34 +0,0 @@
#!/bin/bash
set -e
# Create or update application user for SOC platform access
# This script runs on first database initialization via docker-entrypoint-initdb.d
# The password is properly escaped to handle special characters
if [ -z "${SO_POSTGRES_PASS:-}" ] && [ -n "${SO_POSTGRES_PASS_FILE:-}" ] && [ -r "$SO_POSTGRES_PASS_FILE" ]; then
SO_POSTGRES_PASS="$(< "$SO_POSTGRES_PASS_FILE")"
fi
psql -v ON_ERROR_STOP=1 --username "$POSTGRES_USER" --dbname "$POSTGRES_DB" <<-EOSQL
DO \$\$
BEGIN
IF NOT EXISTS (SELECT FROM pg_catalog.pg_roles WHERE rolname = '${SO_POSTGRES_USER}') THEN
EXECUTE format('CREATE ROLE %I WITH LOGIN PASSWORD %L', '${SO_POSTGRES_USER}', '${SO_POSTGRES_PASS}');
ELSE
EXECUTE format('ALTER ROLE %I WITH PASSWORD %L', '${SO_POSTGRES_USER}', '${SO_POSTGRES_PASS}');
END IF;
END
\$\$;
GRANT ALL PRIVILEGES ON DATABASE "$POSTGRES_DB" TO "$SO_POSTGRES_USER";
-- Lock the SOC database down at the connect layer; PUBLIC gets CONNECT
-- by default, which would let per-minion telegraf roles open sessions
-- here. They have no schema/table grants inside so reads fail, but
-- revoking CONNECT closes the soft edge entirely.
REVOKE CONNECT ON DATABASE "$POSTGRES_DB" FROM PUBLIC;
GRANT CONNECT ON DATABASE "$POSTGRES_DB" TO "$SO_POSTGRES_USER";
EOSQL
# Bootstrap the Telegraf metrics database. Per-minion roles + schemas are
# reconciled on every state.apply by postgres/telegraf_users.sls; this block
# only ensures the shared database exists on first initialization.
if ! psql -U "$POSTGRES_USER" -tAc "SELECT 1 FROM pg_database WHERE datname='so_telegraf'" | grep -q 1; then
psql -v ON_ERROR_STOP=1 -U "$POSTGRES_USER" -c "CREATE DATABASE so_telegraf"
fi
-16
View File
@@ -1,16 +0,0 @@
# Copyright Security Onion Solutions LLC and/or licensed to Security Onion Solutions LLC under one
# or more contributor license agreements. Licensed under the Elastic License 2.0 as shown at
# https://securityonion.net/license; you may not use this file except in compliance with the
# Elastic License 2.0.
#
# Managed by Salt — do not edit by hand.
# Client authentication config: only local (Unix socket) connections and TLS-wrapped TCP
# connections are accepted. Plain-text `host ...` lines are intentionally omitted so a
# misconfigured client with sslmode=disable cannot negotiate a cleartext session.
# Local connections (Unix socket, container-internal) use peer/trust.
local all all trust
# TCP connections MUST use TLS (hostssl) and authenticate with SCRAM.
hostssl all all 0.0.0.0/0 scram-sha-256
hostssl all all ::/0 scram-sha-256
@@ -1,8 +0,0 @@
{# Copyright Security Onion Solutions LLC and/or licensed to Security Onion Solutions LLC under one
or more contributor license agreements. Licensed under the Elastic License 2.0 as shown at
https://securityonion.net/license; you may not use this file except in compliance with the
Elastic License 2.0. #}
{% for key, value in PGMERGED.config.items() %}
{{ key }} = '{{ value | string | replace("'", "''") }}'
{% endfor %}
@@ -1,124 +0,0 @@
-- so_pillar schema: queryable, versioned, audited pillar config store.
-- Replaces flat-file Salt pillar consumed via salt.pillar.postgres ext_pillar.
-- Idempotent. Run via salt/postgres/schema_pillar.sls inside the so-postgres container.
CREATE SCHEMA IF NOT EXISTS so_pillar;
CREATE TABLE IF NOT EXISTS so_pillar.scope (
scope_kind text PRIMARY KEY,
precedence int NOT NULL,
description text
);
INSERT INTO so_pillar.scope(scope_kind, precedence, description) VALUES
('global', 100, 'Applies to every minion'),
('role', 200, 'Applies to minions whose minion_id matches a top.sls compound role match'),
('minion', 300, 'Applies only to a single minion (per-minion overlay)')
ON CONFLICT (scope_kind) DO NOTHING;
CREATE TABLE IF NOT EXISTS so_pillar.role (
role_name text PRIMARY KEY,
match_kind text NOT NULL CHECK (match_kind IN ('compound','grain','glob','list')),
match_expr text NOT NULL,
description text
);
CREATE TABLE IF NOT EXISTS so_pillar.minion (
minion_id text PRIMARY KEY,
node_type text,
hostname text,
extra_roles text[] NOT NULL DEFAULT '{}',
created_at timestamptz NOT NULL DEFAULT now(),
updated_at timestamptz NOT NULL DEFAULT now()
);
CREATE TABLE IF NOT EXISTS so_pillar.role_member (
role_name text NOT NULL REFERENCES so_pillar.role(role_name) ON DELETE CASCADE,
minion_id text NOT NULL REFERENCES so_pillar.minion(minion_id) ON DELETE CASCADE,
source text NOT NULL DEFAULT 'computed' CHECK (source IN ('computed','manual','imported')),
PRIMARY KEY (role_name, minion_id)
);
CREATE INDEX IF NOT EXISTS ix_role_member_minion ON so_pillar.role_member(minion_id);
-- pillar_entry holds the actual data. as_json=True ext_pillar reads `data` directly.
CREATE TABLE IF NOT EXISTS so_pillar.pillar_entry (
id bigserial PRIMARY KEY,
scope text NOT NULL REFERENCES so_pillar.scope(scope_kind),
role_name text REFERENCES so_pillar.role(role_name) ON DELETE CASCADE,
minion_id text REFERENCES so_pillar.minion(minion_id) ON DELETE CASCADE,
pillar_path text NOT NULL,
data jsonb NOT NULL,
is_secret boolean NOT NULL DEFAULT false,
sort_key int NOT NULL DEFAULT 0,
version int NOT NULL DEFAULT 1,
updated_at timestamptz NOT NULL DEFAULT now(),
updated_by text NOT NULL DEFAULT current_user,
change_reason text,
CONSTRAINT pillar_entry_scope_target CHECK (
(scope='global' AND role_name IS NULL AND minion_id IS NULL)
OR (scope='role' AND role_name IS NOT NULL AND minion_id IS NULL)
OR (scope='minion' AND role_name IS NULL AND minion_id IS NOT NULL)
),
-- Reserved namespaces that MUST stay rendered from SLS (mine-driven). Nothing
-- under these prefixes is allowed in the database; the merge logic relies on
-- ext_pillar leaving these subtrees alone.
CONSTRAINT pillar_entry_reserved_paths CHECK (
pillar_path NOT LIKE 'elasticsearch.nodes%'
AND pillar_path NOT LIKE 'redis.nodes%'
AND pillar_path NOT LIKE 'kafka.nodes%'
AND pillar_path NOT LIKE 'hypervisor.nodes%'
AND pillar_path NOT LIKE 'logstash.nodes%'
AND pillar_path NOT LIKE 'node_data.ips%'
)
);
CREATE UNIQUE INDEX IF NOT EXISTS ux_pillar_entry_global ON so_pillar.pillar_entry(pillar_path)
WHERE scope = 'global';
CREATE UNIQUE INDEX IF NOT EXISTS ux_pillar_entry_role ON so_pillar.pillar_entry(role_name, pillar_path)
WHERE scope = 'role';
CREATE UNIQUE INDEX IF NOT EXISTS ux_pillar_entry_minion ON so_pillar.pillar_entry(minion_id, pillar_path)
WHERE scope = 'minion';
CREATE INDEX IF NOT EXISTS ix_pillar_entry_minion_hot ON so_pillar.pillar_entry(minion_id)
WHERE scope = 'minion';
CREATE INDEX IF NOT EXISTS ix_pillar_entry_role_hot ON so_pillar.pillar_entry(role_name)
WHERE scope = 'role';
-- Append-only audit log for every change to pillar_entry. No FK to entry so DELETE
-- history survives the row removal.
CREATE TABLE IF NOT EXISTS so_pillar.pillar_entry_history (
history_id bigserial PRIMARY KEY,
entry_id bigint,
op text NOT NULL CHECK (op IN ('INSERT','UPDATE','DELETE')),
scope text NOT NULL,
role_name text,
minion_id text,
pillar_path text NOT NULL,
old_data jsonb,
new_data jsonb,
is_secret boolean,
version int,
changed_at timestamptz NOT NULL DEFAULT now(),
changed_by text NOT NULL DEFAULT current_user,
change_reason text
);
CREATE INDEX IF NOT EXISTS ix_pillar_history_entry ON so_pillar.pillar_entry_history(entry_id, changed_at DESC);
CREATE INDEX IF NOT EXISTS ix_pillar_history_minion ON so_pillar.pillar_entry_history(minion_id, changed_at DESC);
CREATE INDEX IF NOT EXISTS ix_pillar_history_role ON so_pillar.pillar_entry_history(role_name, changed_at DESC);
-- Drift watch — populated by a pg_cron job that re-renders the on-disk SLS files
-- and compares them to pillar_entry. Cleared once cutover completes.
CREATE TABLE IF NOT EXISTS so_pillar.drift_log (
id bigserial PRIMARY KEY,
scope text NOT NULL,
role_name text,
minion_id text,
pillar_path text NOT NULL,
disk_data jsonb,
db_data jsonb,
detected_at timestamptz NOT NULL DEFAULT now()
);
CREATE INDEX IF NOT EXISTS ix_drift_log_detected ON so_pillar.drift_log(detected_at DESC);
@@ -1,49 +0,0 @@
-- Views consumed by the Salt master's salt.pillar.postgres ext_pillar with
-- as_json=True. Each view exposes data ordered by (sort_key, pillar_path) so
-- the deep-merge in ext_pillar resolves precedence deterministically.
--
-- ext_pillar always binds exactly one parameter to the query: (minion_id,).
-- Master-config queries reference these views and add WHERE clauses, e.g.:
-- SELECT data FROM so_pillar.v_pillar_role WHERE minion_id = %s
-- SELECT data FROM so_pillar.v_pillar_minion WHERE minion_id = %s
-- For v_pillar_global the binding is satisfied with `WHERE %s IS NOT NULL`.
CREATE OR REPLACE VIEW so_pillar.v_pillar_global AS
SELECT pillar_path, sort_key, data
FROM so_pillar.pillar_entry
WHERE scope = 'global'
AND is_secret = false
ORDER BY sort_key, pillar_path;
-- Role view exposes minion_id so the master-config WHERE clause can filter to
-- the rows that apply to the requesting minion. JOIN to role_member fans out
-- one row per (role assignment, pillar entry) tuple.
CREATE OR REPLACE VIEW so_pillar.v_pillar_role AS
SELECT rm.minion_id,
pe.role_name,
pe.pillar_path,
pe.sort_key,
pe.data
FROM so_pillar.pillar_entry pe
JOIN so_pillar.role_member rm ON rm.role_name = pe.role_name
WHERE pe.scope = 'role'
AND pe.is_secret = false;
CREATE OR REPLACE VIEW so_pillar.v_pillar_minion AS
SELECT minion_id,
pillar_path,
sort_key,
data
FROM so_pillar.pillar_entry
WHERE scope = 'minion'
AND is_secret = false;
-- v_pillar_secrets is filled in by 004_secrets.sql once pgcrypto is available;
-- placeholder here returns no rows so initial schema deploy succeeds even on a
-- container that has not yet loaded pgcrypto.
CREATE OR REPLACE VIEW so_pillar.v_pillar_secrets AS
SELECT NULL::text AS minion_id,
NULL::text AS pillar_path,
NULL::int AS sort_key,
'{}'::jsonb AS data
WHERE false;
@@ -1,120 +0,0 @@
-- Audit trigger: every INSERT/UPDATE/DELETE on so_pillar.pillar_entry writes a
-- row to pillar_entry_history. Captures the actor (current_user), reason
-- (passed via SET LOCAL so_pillar.change_reason), and full before/after data.
CREATE OR REPLACE FUNCTION so_pillar.fn_pillar_entry_audit() RETURNS trigger
LANGUAGE plpgsql AS $fn$
DECLARE
v_reason text := current_setting('so_pillar.change_reason', true);
BEGIN
IF (TG_OP = 'INSERT') THEN
INSERT INTO so_pillar.pillar_entry_history(
entry_id, op, scope, role_name, minion_id, pillar_path,
old_data, new_data, is_secret, version, changed_by, change_reason)
VALUES (NEW.id, 'INSERT', NEW.scope, NEW.role_name, NEW.minion_id, NEW.pillar_path,
NULL, NEW.data, NEW.is_secret, NEW.version, NEW.updated_by, v_reason);
RETURN NEW;
ELSIF (TG_OP = 'UPDATE') THEN
IF OLD.data IS DISTINCT FROM NEW.data
OR OLD.is_secret IS DISTINCT FROM NEW.is_secret THEN
INSERT INTO so_pillar.pillar_entry_history(
entry_id, op, scope, role_name, minion_id, pillar_path,
old_data, new_data, is_secret, version, changed_by, change_reason)
VALUES (NEW.id, 'UPDATE', NEW.scope, NEW.role_name, NEW.minion_id, NEW.pillar_path,
OLD.data, NEW.data, NEW.is_secret, NEW.version, NEW.updated_by, v_reason);
END IF;
RETURN NEW;
ELSIF (TG_OP = 'DELETE') THEN
INSERT INTO so_pillar.pillar_entry_history(
entry_id, op, scope, role_name, minion_id, pillar_path,
old_data, new_data, is_secret, version, changed_by, change_reason)
VALUES (OLD.id, 'DELETE', OLD.scope, OLD.role_name, OLD.minion_id, OLD.pillar_path,
OLD.data, NULL, OLD.is_secret, OLD.version, current_user, v_reason);
RETURN OLD;
END IF;
RETURN NULL;
END
$fn$;
DROP TRIGGER IF EXISTS pillar_entry_audit ON so_pillar.pillar_entry;
CREATE TRIGGER pillar_entry_audit
AFTER INSERT OR UPDATE OR DELETE ON so_pillar.pillar_entry
FOR EACH ROW EXECUTE FUNCTION so_pillar.fn_pillar_entry_audit();
-- updated_at + version maintenance: bump version on every UPDATE that changes data.
CREATE OR REPLACE FUNCTION so_pillar.fn_pillar_entry_versioning() RETURNS trigger
LANGUAGE plpgsql AS $fn$
BEGIN
IF (TG_OP = 'UPDATE') THEN
IF OLD.data IS DISTINCT FROM NEW.data
OR OLD.is_secret IS DISTINCT FROM NEW.is_secret THEN
NEW.version := OLD.version + 1;
NEW.updated_at := now();
ELSE
NEW.version := OLD.version;
NEW.updated_at := OLD.updated_at;
END IF;
END IF;
RETURN NEW;
END
$fn$;
DROP TRIGGER IF EXISTS pillar_entry_versioning ON so_pillar.pillar_entry;
CREATE TRIGGER pillar_entry_versioning
BEFORE UPDATE ON so_pillar.pillar_entry
FOR EACH ROW EXECUTE FUNCTION so_pillar.fn_pillar_entry_versioning();
-- Recompute role_member rows for a minion based on node_type.
-- Compound matchers in pillar/top.sls are pure suffix patterns of the form
-- '*_<rolename>' plus the special multi-role 'manager/managersearch/managerhype'
-- bucket. node_type is split on common dashes/underscores; any token that
-- matches a known role_name produces a role_member row.
CREATE OR REPLACE FUNCTION so_pillar.fn_recompute_role_members(p_minion_id text)
RETURNS void LANGUAGE plpgsql AS $fn$
DECLARE
v_node_type text;
v_extra text[];
v_role text;
BEGIN
SELECT node_type, extra_roles INTO v_node_type, v_extra
FROM so_pillar.minion WHERE minion_id = p_minion_id;
IF v_node_type IS NULL THEN
RETURN;
END IF;
DELETE FROM so_pillar.role_member
WHERE minion_id = p_minion_id AND source = 'computed';
-- Main role from node_type.
IF EXISTS (SELECT 1 FROM so_pillar.role WHERE role_name = lower(v_node_type)) THEN
INSERT INTO so_pillar.role_member(role_name, minion_id, source)
VALUES (lower(v_node_type), p_minion_id, 'computed')
ON CONFLICT DO NOTHING;
END IF;
-- Extra roles supplied by the importer / reactor for compound matchers
-- that need to apply multiple buckets (e.g. managersearch also gets the
-- 'manager' bucket per top.sls line 36 grouping).
FOREACH v_role IN ARRAY COALESCE(v_extra, '{}'::text[]) LOOP
IF EXISTS (SELECT 1 FROM so_pillar.role WHERE role_name = v_role) THEN
INSERT INTO so_pillar.role_member(role_name, minion_id, source)
VALUES (v_role, p_minion_id, 'computed')
ON CONFLICT DO NOTHING;
END IF;
END LOOP;
END
$fn$;
CREATE OR REPLACE FUNCTION so_pillar.fn_minion_after_change() RETURNS trigger
LANGUAGE plpgsql AS $fn$
BEGIN
PERFORM so_pillar.fn_recompute_role_members(COALESCE(NEW.minion_id, OLD.minion_id));
RETURN COALESCE(NEW, OLD);
END
$fn$;
DROP TRIGGER IF EXISTS minion_role_sync ON so_pillar.minion;
CREATE TRIGGER minion_role_sync
AFTER INSERT OR UPDATE OF node_type, extra_roles ON so_pillar.minion
FOR EACH ROW EXECUTE FUNCTION so_pillar.fn_minion_after_change();
@@ -1,130 +0,0 @@
-- pgcrypto-backed secret storage for pillar_entry rows where is_secret = true.
-- The plaintext value is encrypted with a symmetric key held in a server-side
-- GUC (so_pillar.master_key) which is set per-role via ALTER ROLE so the key
-- never touches a flat file readable by Salt itself.
CREATE EXTENSION IF NOT EXISTS pgcrypto WITH SCHEMA public;
-- Encrypt a JSONB value using the configured master key. Stored as a JSONB
-- envelope {"_enc": "<armored ciphertext>"} so the same column type is reused.
CREATE OR REPLACE FUNCTION so_pillar.fn_encrypt_jsonb(p_value jsonb)
RETURNS jsonb LANGUAGE plpgsql AS $fn$
DECLARE
v_key text := current_setting('so_pillar.master_key', true);
BEGIN
IF v_key IS NULL OR v_key = '' THEN
RAISE EXCEPTION 'so_pillar.master_key GUC not configured';
END IF;
RETURN jsonb_build_object(
'_enc',
encode(pgp_sym_encrypt(p_value::text, v_key), 'base64')
);
END
$fn$;
-- Decrypt the envelope produced by fn_encrypt_jsonb. SECURITY DEFINER so callers
-- with no direct access to pgcrypto/master_key can still pull plaintext via the
-- v_pillar_secrets view.
CREATE OR REPLACE FUNCTION so_pillar.fn_decrypt_jsonb(p_envelope jsonb)
RETURNS jsonb LANGUAGE plpgsql SECURITY DEFINER AS $fn$
DECLARE
v_key text := current_setting('so_pillar.master_key', true);
v_ct text;
BEGIN
IF v_key IS NULL OR v_key = '' THEN
RAISE EXCEPTION 'so_pillar.master_key GUC not configured';
END IF;
v_ct := p_envelope->>'_enc';
IF v_ct IS NULL THEN
RETURN p_envelope; -- not encrypted; pass through
END IF;
RETURN pgp_sym_decrypt(decode(v_ct, 'base64'), v_key)::jsonb;
END
$fn$;
REVOKE ALL ON FUNCTION so_pillar.fn_decrypt_jsonb(jsonb) FROM PUBLIC;
-- Secrets view consumed by ext_pillar. Decrypts at the boundary so Salt sees
-- plaintext JSONB. Filters the rows to those that apply to the requesting
-- minion via current_setting, since views can't take parameters and ext_pillar
-- can only bind one parameter per query.
--
-- Master-config query: SELECT data FROM so_pillar.v_pillar_secrets WHERE %s IS NOT NULL
-- The %s satisfies the bound parameter; the view itself reads the minion_id
-- from a session GUC set by a small wrapper function (see fn_pillar_secrets).
CREATE OR REPLACE FUNCTION so_pillar.fn_pillar_secrets(p_minion_id text)
RETURNS TABLE(data jsonb)
LANGUAGE sql STABLE SECURITY DEFINER AS $fn$
SELECT so_pillar.fn_decrypt_jsonb(pe.data)
FROM so_pillar.pillar_entry pe
WHERE pe.is_secret = true
AND ( pe.scope = 'global'
OR (pe.scope = 'role'
AND pe.role_name IN (
SELECT role_name FROM so_pillar.role_member
WHERE minion_id = p_minion_id))
OR (pe.scope = 'minion' AND pe.minion_id = p_minion_id))
ORDER BY pe.sort_key, pe.pillar_path;
$fn$;
-- Replace the placeholder view from 002 with a parameterised version. Master
-- config query becomes:
-- SELECT data FROM so_pillar.fn_pillar_secrets(%s) AS s
DROP VIEW IF EXISTS so_pillar.v_pillar_secrets;
CREATE OR REPLACE VIEW so_pillar.v_pillar_secrets AS
SELECT NULL::text AS minion_id,
NULL::text AS pillar_path,
NULL::int AS sort_key,
'{}'::jsonb AS data
WHERE false;
COMMENT ON VIEW so_pillar.v_pillar_secrets IS
'Deprecated placeholder; use SELECT data FROM so_pillar.fn_pillar_secrets(minion_id) instead';
-- Convenience helper for so-yaml.py and the importer to set a secret without
-- ever exposing the master_key to the caller. SECURITY DEFINER means the
-- caller does not need read access to so_pillar.master_key.
CREATE OR REPLACE FUNCTION so_pillar.fn_set_secret(
p_scope text,
p_role_name text,
p_minion_id text,
p_pillar_path text,
p_value jsonb,
p_change_reason text DEFAULT NULL
) RETURNS bigint LANGUAGE plpgsql SECURITY DEFINER AS $fn$
DECLARE
v_envelope jsonb := so_pillar.fn_encrypt_jsonb(p_value);
v_id bigint;
BEGIN
PERFORM set_config('so_pillar.change_reason',
COALESCE(p_change_reason, 'fn_set_secret'),
true);
INSERT INTO so_pillar.pillar_entry(
scope, role_name, minion_id, pillar_path, data, is_secret, change_reason)
VALUES (p_scope, p_role_name, p_minion_id, p_pillar_path, v_envelope, true, p_change_reason)
ON CONFLICT (pillar_path) WHERE scope='global' DO UPDATE
SET data = EXCLUDED.data, is_secret = true, change_reason = EXCLUDED.change_reason
RETURNING id INTO v_id;
IF v_id IS NULL THEN
UPDATE so_pillar.pillar_entry
SET data = v_envelope, is_secret = true, change_reason = p_change_reason
WHERE scope = p_scope
AND COALESCE(role_name,'') = COALESCE(p_role_name,'')
AND COALESCE(minion_id,'') = COALESCE(p_minion_id,'')
AND pillar_path = p_pillar_path
RETURNING id INTO v_id;
IF v_id IS NULL THEN
INSERT INTO so_pillar.pillar_entry(
scope, role_name, minion_id, pillar_path, data, is_secret, change_reason)
VALUES (p_scope, p_role_name, p_minion_id, p_pillar_path, v_envelope, true, p_change_reason)
RETURNING id INTO v_id;
END IF;
END IF;
RETURN v_id;
END
$fn$;
REVOKE ALL ON FUNCTION so_pillar.fn_set_secret(text,text,text,text,jsonb,text) FROM PUBLIC;
@@ -1,39 +0,0 @@
-- Seed the so_pillar.role table with the role buckets defined in pillar/top.sls.
-- The match_expr column preserves the original Salt compound expression purely
-- as documentation; PG-side membership is materialised in role_member.
-- Idempotent: ON CONFLICT lets re-application leave existing rows untouched.
INSERT INTO so_pillar.role(role_name, match_kind, match_expr, description) VALUES
('manager', 'compound', '*_manager or *_managersearch or *_managerhype',
'Manager-class node. Includes managersearch and managerhype subtypes.'),
('managersearch', 'compound', '*_managersearch',
'Combined manager + searchnode role.'),
('managerhype', 'compound', '*_managerhype',
'Combined manager + hypervisor role.'),
('sensor', 'compound', '*_sensor',
'Sensor node running zeek/suricata/strelka.'),
('eval', 'compound', '*_eval',
'Single-node evaluation install (manager + sensor + storage on one host).'),
('standalone', 'compound', '*_standalone',
'Single-node production install (no distributed cluster).'),
('heavynode', 'compound', '*_heavynode',
'Distributed manager node carrying logstash + ES.'),
('idh', 'compound', '*_idh',
'Intrusion-detection-honeypot node.'),
('searchnode', 'compound', '*_searchnode',
'Distributed Elasticsearch search node.'),
('receiver', 'compound', '*_receiver',
'Kafka receiver node.'),
('import', 'compound', '*_import',
'Single-node import-only install.'),
('fleet', 'compound', '*_fleet',
'Elastic Fleet server node.'),
('hypervisor', 'compound', '*_hypervisor',
'Hypervisor host (libvirt). Hosts VM minions.'),
('desktop', 'compound', '*_desktop',
'Desktop minion (no firewall/nginx pillars apply).'),
('not_desktop', 'compound', '* and not *_desktop',
'Pseudo-role; matches every minion that is not a desktop. Used for global firewall/nginx.'),
('libvirt', 'grain', 'salt-cloud:driver:libvirt',
'Pseudo-role; matches any minion with grain salt-cloud.driver = libvirt.')
ON CONFLICT (role_name) DO NOTHING;
@@ -1,107 +0,0 @@
-- Roles + Row-Level Security policies for the so_pillar schema.
-- Three roles:
-- so_pillar_master — connected by salt-master ext_pillar. Read-only.
-- RLS forces it to skip is_secret rows; reads
-- encrypted secrets only via fn_pillar_secrets().
-- so_pillar_writer — connected by so-yaml dual-write and the SOC
-- PostgresConfigstore. Read+write on pillar_entry,
-- minion, role_member.
-- so_pillar_secret_owner — owns the master encryption key GUC; sole role
-- allowed to call fn_set_secret directly. Other
-- writers reach this function only via grants.
--
-- The existing app role so_postgres_user (created by init-users.sh) is granted
-- INTO so_pillar_writer so SOC keeps using its existing connection but inherits
-- pillar-write capability.
DO $$
BEGIN
IF NOT EXISTS (SELECT 1 FROM pg_roles WHERE rolname = 'so_pillar_master') THEN
CREATE ROLE so_pillar_master NOLOGIN;
END IF;
IF NOT EXISTS (SELECT 1 FROM pg_roles WHERE rolname = 'so_pillar_writer') THEN
CREATE ROLE so_pillar_writer NOLOGIN;
END IF;
IF NOT EXISTS (SELECT 1 FROM pg_roles WHERE rolname = 'so_pillar_secret_owner') THEN
CREATE ROLE so_pillar_secret_owner NOLOGIN;
END IF;
END
$$;
-- USAGE on the schema is the bare minimum needed to reference its tables.
-- CONNECT on the database is needed before the role can establish a session
-- at all (default privileges on a new DB grant CONNECT to PUBLIC, but if the
-- securityonion database is restricted that grant has to be explicit).
-- Password + LOGIN privileges are set later in schema_pillar.sls because
-- the password lives in pillar (secrets:pillar_master_pass) and plain SQL
-- can't substitute pillar values.
GRANT CONNECT ON DATABASE securityonion TO so_pillar_master, so_pillar_writer, so_pillar_secret_owner;
GRANT USAGE ON SCHEMA so_pillar TO so_pillar_master, so_pillar_writer, so_pillar_secret_owner;
-- Read access for ext_pillar through the views only.
GRANT SELECT ON so_pillar.v_pillar_global,
so_pillar.v_pillar_role,
so_pillar.v_pillar_minion
TO so_pillar_master;
GRANT EXECUTE ON FUNCTION so_pillar.fn_pillar_secrets(text) TO so_pillar_master;
-- (change_queue grants live in 008_change_notify.sql alongside the table itself,
-- since the table doesn't exist until 008 runs.)
-- Writer needs CRUD on pillar_entry/minion/role_member plus access to seed tables.
GRANT SELECT, INSERT, UPDATE, DELETE
ON so_pillar.pillar_entry,
so_pillar.minion,
so_pillar.role_member
TO so_pillar_writer;
GRANT SELECT ON so_pillar.role, so_pillar.scope TO so_pillar_writer;
GRANT SELECT, INSERT, UPDATE, DELETE ON so_pillar.drift_log TO so_pillar_writer;
GRANT USAGE, SELECT ON ALL SEQUENCES IN SCHEMA so_pillar TO so_pillar_writer;
GRANT SELECT ON so_pillar.pillar_entry_history TO so_pillar_writer;
-- Secret owner can call fn_set_secret directly; writer goes through it via the
-- function's SECURITY DEFINER attribute, which executes as the function owner.
GRANT EXECUTE ON FUNCTION so_pillar.fn_set_secret(text,text,text,text,jsonb,text)
TO so_pillar_writer, so_pillar_secret_owner;
-- so_postgres_user (SOC's existing app user, created by init-users.sh) inherits
-- writer privilege so the PostgresConfigstore in SOC can mutate pillars without
-- a second connection pool. Inheritance is per-PG default (NOINHERIT must be
-- explicit), so this just works.
DO $$
BEGIN
IF EXISTS (SELECT 1 FROM pg_roles WHERE rolname = current_setting('so_pillar.app_role', true))
THEN
EXECUTE format('GRANT so_pillar_writer TO %I',
current_setting('so_pillar.app_role', true));
ELSIF EXISTS (SELECT 1 FROM pg_roles WHERE rolname = 'so_postgres_user') THEN
GRANT so_pillar_writer TO so_postgres_user;
END IF;
END
$$;
-- RLS on pillar_entry: master sees only non-secret rows. Writer sees all
-- (it must, to UPDATE secret rows when so-yaml replaces them). Secret rows
-- still require fn_decrypt_jsonb to read plaintext.
ALTER TABLE so_pillar.pillar_entry ENABLE ROW LEVEL SECURITY;
ALTER TABLE so_pillar.pillar_entry FORCE ROW LEVEL SECURITY;
DROP POLICY IF EXISTS pillar_entry_master_read ON so_pillar.pillar_entry;
DROP POLICY IF EXISTS pillar_entry_writer_all ON so_pillar.pillar_entry;
DROP POLICY IF EXISTS pillar_entry_owner_all ON so_pillar.pillar_entry;
CREATE POLICY pillar_entry_master_read ON so_pillar.pillar_entry
FOR SELECT TO so_pillar_master
USING (NOT is_secret);
CREATE POLICY pillar_entry_writer_all ON so_pillar.pillar_entry
FOR ALL TO so_pillar_writer
USING (true)
WITH CHECK (true);
CREATE POLICY pillar_entry_owner_all ON so_pillar.pillar_entry
FOR ALL TO so_pillar_secret_owner
USING (true)
WITH CHECK (true);
-- minion / role_member do not need RLS — they hold no secrets.
@@ -1,43 +0,0 @@
-- Drift detection + retention via pg_cron. Optional — the schema_pillar.sls
-- state guards this file behind the postgres:so_pillar:drift_check_enabled
-- pillar flag because pg_cron may not be loaded on every install.
CREATE EXTENSION IF NOT EXISTS pg_cron;
-- Retention: trim pillar_entry_history older than a year. Adjustable via the
-- so_pillar.history_retention_days GUC (default 365 if unset).
CREATE OR REPLACE FUNCTION so_pillar.fn_history_retain()
RETURNS void LANGUAGE plpgsql AS $fn$
DECLARE
v_days int := COALESCE(current_setting('so_pillar.history_retention_days', true)::int, 365);
BEGIN
DELETE FROM so_pillar.pillar_entry_history
WHERE changed_at < (now() - (v_days::text || ' days')::interval);
END
$fn$;
-- Drift retention: keep two weeks of drift_log.
CREATE OR REPLACE FUNCTION so_pillar.fn_drift_retain()
RETURNS void LANGUAGE plpgsql AS $fn$
BEGIN
DELETE FROM so_pillar.drift_log
WHERE detected_at < (now() - interval '14 days');
END
$fn$;
-- pg_cron schedules (idempotent — unschedule any existing same-named job first).
DO $$
DECLARE
v_jobid bigint;
BEGIN
SELECT jobid INTO v_jobid FROM cron.job WHERE jobname = 'so_pillar_history_retain';
IF v_jobid IS NOT NULL THEN PERFORM cron.unschedule(v_jobid); END IF;
PERFORM cron.schedule('so_pillar_history_retain', '15 3 * * *',
'SELECT so_pillar.fn_history_retain();');
SELECT jobid INTO v_jobid FROM cron.job WHERE jobname = 'so_pillar_drift_retain';
IF v_jobid IS NOT NULL THEN PERFORM cron.unschedule(v_jobid); END IF;
PERFORM cron.schedule('so_pillar_drift_retain', '20 3 * * *',
'SELECT so_pillar.fn_drift_retain();');
END
$$;
@@ -1,89 +0,0 @@
-- pg_notify-driven change fan-out for so_pillar.pillar_entry.
--
-- Two layers:
-- 1. so_pillar.change_queue — durable, drained by the salt-master
-- engine. Survives engine downtime,
-- de-duplicated by id, processed once.
-- 2. pg_notify('so_pillar_change') — wakeup signal. Payload is the
-- change_queue row id and locator
-- (no secret data — channels are
-- snoopable by anyone with LISTEN).
--
-- The salt-master engine LISTENs on the channel for low-latency wakeup,
-- then SELECTs unprocessed change_queue rows so a missed notification
-- (engine restart, network blip) self-heals on the next event.
CREATE TABLE IF NOT EXISTS so_pillar.change_queue (
id bigserial PRIMARY KEY,
scope text NOT NULL,
role_name text,
minion_id text,
pillar_path text NOT NULL,
op text NOT NULL CHECK (op IN ('INSERT','UPDATE','DELETE')),
enqueued_at timestamptz NOT NULL DEFAULT now(),
processed_at timestamptz
);
-- Hot index for the engine's drain query.
CREATE INDEX IF NOT EXISTS ix_change_queue_unprocessed
ON so_pillar.change_queue (id)
WHERE processed_at IS NULL;
-- Retention index: pg_cron job in 007 sweeps processed rows older than 7d.
CREATE INDEX IF NOT EXISTS ix_change_queue_processed_at
ON so_pillar.change_queue (processed_at)
WHERE processed_at IS NOT NULL;
CREATE OR REPLACE FUNCTION so_pillar.fn_pillar_entry_notify()
RETURNS trigger
LANGUAGE plpgsql
AS $$
DECLARE
v_row record;
v_id bigint;
BEGIN
IF TG_OP = 'DELETE' THEN
v_row := OLD;
ELSE
v_row := NEW;
END IF;
INSERT INTO so_pillar.change_queue
(scope, role_name, minion_id, pillar_path, op)
VALUES
(v_row.scope, v_row.role_name, v_row.minion_id, v_row.pillar_path, TG_OP)
RETURNING id INTO v_id;
-- Payload is the queue id + locator only. Engine joins back to
-- pillar_entry if it needs the data — keeps secrets off the wire.
PERFORM pg_notify('so_pillar_change', json_build_object(
'queue_id', v_id,
'scope', v_row.scope,
'role_name', v_row.role_name,
'minion_id', v_row.minion_id,
'pillar_path', v_row.pillar_path,
'op', TG_OP
)::text);
RETURN NULL;
END;
$$;
DROP TRIGGER IF EXISTS tg_pillar_entry_notify ON so_pillar.pillar_entry;
CREATE TRIGGER tg_pillar_entry_notify
AFTER INSERT OR UPDATE OR DELETE
ON so_pillar.pillar_entry
FOR EACH ROW
EXECUTE FUNCTION so_pillar.fn_pillar_entry_notify();
-- Role grants on the change_queue table. Lived in 006_rls.sql historically but
-- moved here so the GRANT references resolve — 006 runs before this file does.
-- Engine reads + drains the change queue from the salt-master process. It
-- needs SELECT to find unprocessed rows and UPDATE to mark them processed.
-- The queue contains only locator metadata (no pillar data), so the master
-- role's existing privilege footprint is unchanged in practice.
GRANT SELECT, UPDATE ON so_pillar.change_queue TO so_pillar_master;
GRANT USAGE ON SEQUENCE so_pillar.change_queue_id_seq TO so_pillar_master;
-- Writer needs INSERT (the trigger runs as table owner, so this is just for
-- direct testing / manual replays from psql).
GRANT INSERT ON so_pillar.change_queue TO so_pillar_writer;
-14
View File
@@ -1,14 +0,0 @@
# Copyright Security Onion Solutions LLC and/or licensed to Security Onion Solutions LLC under one
# or more contributor license agreements. Licensed under the Elastic License 2.0 as shown at
# https://securityonion.net/license; you may not use this file except in compliance with the
# Elastic License 2.0.
{% from 'postgres/map.jinja' import PGMERGED %}
include:
{% if PGMERGED.enabled %}
- postgres.enabled
- postgres.schema_pillar
{% else %}
- postgres.disabled
{% endif %}
-7
View File
@@ -1,7 +0,0 @@
{# Copyright Security Onion Solutions LLC and/or licensed to Security Onion Solutions LLC under one
or more contributor license agreements. Licensed under the Elastic License 2.0 as shown at
https://securityonion.net/license; you may not use this file except in compliance with the
Elastic License 2.0. #}
{% import_yaml 'postgres/defaults.yaml' as PGDEFAULTS %}
{% set PGMERGED = salt['pillar.get']('postgres', PGDEFAULTS.postgres, merge=True) %}
-187
View File
@@ -1,187 +0,0 @@
# Copyright Security Onion Solutions LLC and/or licensed to Security Onion Solutions LLC under one
# or more contributor license agreements. Licensed under the Elastic License 2.0 as shown at
# https://securityonion.net/license; you may not use this file except in compliance with the
# Elastic License 2.0.
{% from 'allowed_states.map.jinja' import allowed_states %}
{% if 'postgres' in allowed_states %}
{% from 'vars/globals.map.jinja' import GLOBALS %}
# Deploys the so_pillar schema (tables, views, audit triggers, secrets,
# RLS, pg_cron retention) inside the so-postgres container. Idempotent —
# every CREATE / GRANT is wrapped in IF NOT EXISTS / ON CONFLICT or DO
# blocks so re-running the state is a no-op when the schema is current.
#
# Gated on the postgres:so_pillar:enabled feature flag (default false).
# Flip to true once the postsalt branch is ready to bring ext_pillar live.
include:
- postgres.enabled
{% set so_pillar_enabled = salt['pillar.get']('postgres:so_pillar:enabled', False) %}
{% if so_pillar_enabled %}
{% set drift_enabled = salt['pillar.get']('postgres:so_pillar:drift_check_enabled', False) %}
{% set schema_dir = '/opt/so/saltstack/default/salt/postgres/files/schema/pillar' %}
# Wait for postgres to actually accept TCP connections. Same idiom as
# telegraf_users.sls. The docker_container.running state returns earlier than
# the database is ready on first init.
so_pillar_postgres_wait_ready:
cmd.run:
- name: |
for i in $(seq 1 60); do
if docker exec so-postgres pg_isready -h 127.0.0.1 -U postgres -q 2>/dev/null; then
exit 0
fi
sleep 2
done
echo "so-postgres did not accept TCP connections within 120s" >&2
exit 1
- require:
- docker_container: so-postgres
{% set sql_files = [
'001_schema.sql',
'002_views.sql',
'003_history_trigger.sql',
'004_secrets.sql',
'005_seed_roles.sql',
'006_rls.sql',
] %}
{% if drift_enabled %}
{% do sql_files.append('007_drift_pgcron.sql') %}
{% endif %}
# 008 always applies — pg_notify-driven change fan-out is what the salt-master
# pg_notify_pillar engine consumes. Without it reactor wiring sees no events.
{% do sql_files.append('008_change_notify.sql') %}
{% for sql_file in sql_files %}
so_pillar_apply_{{ sql_file | replace('.', '_') }}:
cmd.run:
- name: |
docker exec -i so-postgres psql -v ON_ERROR_STOP=1 -U postgres -d securityonion \
< {{ schema_dir }}/{{ sql_file }}
- require:
- cmd: so_pillar_postgres_wait_ready
{% if not loop.first %}
- cmd: so_pillar_apply_{{ sql_files[loop.index0 - 1] | replace('.', '_') }}
{% endif %}
{% endfor %}
# Set the master encryption key GUC on the secret-owner role. The key itself
# is generated by setup/so-functions::secrets_pillar() (extended for postsalt)
# and lives in /opt/so/conf/postgres/so_pillar.key (mode 0400) — never read by
# Salt itself; the value flows into PG via ALTER ROLE so it sits only in the
# server's role catalog.
so_pillar_master_key_configure:
cmd.run:
- name: |
if [ -r /opt/so/conf/postgres/so_pillar.key ]; then
KEY="$(< /opt/so/conf/postgres/so_pillar.key)"
docker exec -i so-postgres psql -v ON_ERROR_STOP=1 -U postgres -d securityonion <<EOSQL
ALTER ROLE so_pillar_secret_owner SET so_pillar.master_key = '$KEY';
ALTER ROLE so_pillar_master SET so_pillar.master_key = '$KEY';
ALTER ROLE so_pillar_writer SET so_pillar.master_key = '$KEY';
EOSQL
else
echo "so_pillar.key not present yet; setup/so-functions must generate it before schema_pillar.sls" >&2
exit 1
fi
- require:
- cmd: so_pillar_apply_{{ sql_files[-1] | replace('.', '_') }}
# Set login passwords on the so_pillar_* roles. 006_rls.sql creates the roles
# as NOLOGIN with no password (plain SQL can't substitute pillar values), so
# the salt-master ext_pillar and the pg_notify_pillar engine — both of which
# connect as so_pillar_master via TCP — would fail with "password
# authentication failed" without this step. The password lives in pillar
# under secrets:pillar_master_pass (generated by setup/so-functions::secrets_pillar)
# and is the same one rendered into ext_pillar_postgres.conf.jinja and the
# engines.conf pg_notify_pillar block, so all three sides agree.
so_pillar_role_login_passwords:
cmd.run:
- name: |
docker exec -i so-postgres psql -v ON_ERROR_STOP=1 -U postgres -d securityonion <<EOSQL
ALTER ROLE so_pillar_master WITH LOGIN PASSWORD '{{ pillar['secrets']['pillar_master_pass'] }}';
ALTER ROLE so_pillar_writer WITH LOGIN PASSWORD '{{ pillar['secrets']['pillar_master_pass'] }}';
ALTER ROLE so_pillar_secret_owner WITH LOGIN PASSWORD '{{ pillar['secrets']['pillar_master_pass'] }}';
EOSQL
- require:
- cmd: so_pillar_master_key_configure
# Install psycopg2 into salt-master's bundled python so the pg_notify_pillar
# engine module can `import psycopg2`. Without this the engine's import fails
# silently in salt's loader and the engine just never starts. salt's bundled
# python at /opt/saltstack/salt/bin/python3 doesn't ship psycopg by default.
#
# Uses cmd.run with an `unless` import-test rather than pip.installed because
# pip exits non-zero if patchelf isn't on PATH (it tries to rewrite the
# psycopg2 wheel's RPATH after extraction), even though the wheel is fully
# installed and importable. salt's pip.installed surfaces the non-zero exit
# as a state failure and the cascade kills schema_pillar's downstream work.
# `import psycopg2` succeeds either way, so that's the actual readiness gate.
#
# Pip's stdout/stderr is redirected to /opt/so/log/so_pillar/psycopg2_install.log
# so the literal "ERROR: ... patchelf" line doesn't get hoovered up into
# /root/sosetup.log and then into /root/errors.log by verify_setup's
# substring-grep for "ERROR". The redirect target is preserved for
# triage if `import psycopg2` ever does fail.
so_pillar_psycopg2_in_salt_python:
cmd.run:
- name: |
mkdir -p /opt/so/log/so_pillar
/opt/saltstack/salt/bin/pip3 install --quiet psycopg2-binary \
>/opt/so/log/so_pillar/psycopg2_install.log 2>&1 \
|| true
- unless: /opt/saltstack/salt/bin/python3 -c "import psycopg2"
- require:
- cmd: so_pillar_role_login_passwords
# Run the importer once after the schema is in place. Idempotent — re-runs
# with no SLS edits produce zero row changes.
so_pillar_initial_import:
cmd.run:
- name: /usr/sbin/so-pillar-import --yes --reason 'schema_pillar.sls initial import'
- require:
- cmd: so_pillar_psycopg2_in_salt_python
# Flip so-yaml from dual-write to PG-canonical for managed paths now that
# the schema and importer are both in place. Bootstrap files (secrets.sls,
# postgres/auth.sls, ca/init.sls, *.nodes.sls, top.sls, ...) remain on disk
# regardless because so_yaml_postgres.locate() raises SkipPath for them.
so_pillar_so_yaml_mode_dir:
file.directory:
- name: /opt/so/conf/so-yaml
- user: socore
- group: socore
- mode: '0755'
- makedirs: True
so_pillar_so_yaml_mode_postgres:
file.managed:
- name: /opt/so/conf/so-yaml/mode
- contents: postgres
- user: socore
- group: socore
- mode: '0644'
- require:
- file: so_pillar_so_yaml_mode_dir
- cmd: so_pillar_initial_import
{% else %}
so_pillar_disabled_noop:
test.nop
{% endif %}
{% else %}
{{sls}}_state_not_allowed:
test.fail_without_changes:
- name: {{sls}}_state_not_allowed
{% endif %}
-89
View File
@@ -1,89 +0,0 @@
postgres:
enabled:
description: Whether the PostgreSQL database container is enabled on this grid. Backs the assistant store and the Telegraf metrics database.
forcedType: bool
readonly: True
helpLink: influxdb
telegraf:
retention_days:
description: Number of days of Telegraf metrics to keep in the so_telegraf database. Older partitions are dropped hourly by pg_partman.
forcedType: int
helpLink: postgres
config:
max_connections:
description: Maximum number of concurrent PostgreSQL connections.
forcedType: int
global: True
helpLink: postgres
shared_buffers:
description: Amount of memory PostgreSQL uses for shared buffers (e.g. 256MB, 1GB). Raising this improves read cache hit rate at the cost of system RAM.
global: True
helpLink: postgres
log_min_messages:
description: Minimum severity of server messages written to the PostgreSQL log.
options:
- debug1
- info
- notice
- warning
- error
- log
- fatal
global: True
helpLink: postgres
listen_addresses:
description: Interfaces PostgreSQL listens on. Must remain '*' so clients on the docker bridge network can connect.
global: True
advanced: True
helpLink: postgres
port:
description: TCP port PostgreSQL listens on inside the container. Firewall rules and container port mapping assume 5432.
forcedType: int
global: True
advanced: True
helpLink: postgres
ssl:
description: Whether PostgreSQL accepts TLS connections. Must remain 'on' — pg_hba.conf requires hostssl for TCP.
global: True
advanced: True
helpLink: postgres
ssl_cert_file:
description: Path (inside the container) to the TLS server certificate. Salt-managed.
global: True
advanced: True
helpLink: postgres
ssl_key_file:
description: Path (inside the container) to the TLS server private key. Salt-managed.
global: True
advanced: True
helpLink: postgres
ssl_ca_file:
description: Path (inside the container) to the CA bundle PostgreSQL uses to verify client certificates. Salt-managed.
global: True
advanced: True
helpLink: postgres
hba_file:
description: Path (inside the container) to the pg_hba.conf authentication file. Salt-managed — edit salt/postgres/files/pg_hba.conf.
global: True
advanced: True
helpLink: postgres
log_destination:
description: Where PostgreSQL writes its server log. 'stderr' routes to the container log stream.
global: True
advanced: True
helpLink: postgres
logging_collector:
description: Whether to run a separate logging collector process. Disabled because the docker log stream already captures stderr.
global: True
advanced: True
helpLink: postgres
shared_preload_libraries:
description: Comma-separated list of extensions loaded at server start. Required for pg_cron which drives pg_partman maintenance — do not remove.
global: True
advanced: True
helpLink: postgres
cron.database_name:
description: Database pg_cron schedules jobs in. Must be so_telegraf so partman maintenance runs in the right database context.
global: True
advanced: True
helpLink: postgres
-21
View File
@@ -1,21 +0,0 @@
# Copyright Security Onion Solutions LLC and/or licensed to Security Onion Solutions LLC under one
# or more contributor license agreements. Licensed under the Elastic License 2.0 as shown at
# https://securityonion.net/license; you may not use this file except in compliance with the
# Elastic License 2.0.
{% from 'allowed_states.map.jinja' import allowed_states %}
{% if sls.split('.')[0] in allowed_states %}
append_so-postgres_so-status.conf:
file.append:
- name: /opt/so/conf/so-status/so-status.conf
- text: so-postgres
- unless: grep -q so-postgres /opt/so/conf/so-status/so-status.conf
{% else %}
{{sls}}_state_not_allowed:
test.fail_without_changes:
- name: {{sls}}_state_not_allowed
{% endif %}
-55
View File
@@ -1,55 +0,0 @@
# Copyright Security Onion Solutions LLC and/or licensed to Security Onion Solutions LLC under one
# or more contributor license agreements. Licensed under the Elastic License 2.0 as shown at
# https://securityonion.net/license; you may not use this file except in compliance with the
# Elastic License 2.0.
{% from 'allowed_states.map.jinja' import allowed_states %}
{% if sls.split('.')[0] in allowed_states %}
{% from 'vars/globals.map.jinja' import GLOBALS %}
{% from 'ca/map.jinja' import CA %}
postgres_key:
x509.private_key_managed:
- name: /etc/pki/postgres.key
- keysize: 4096
- backup: True
- new: True
{% if salt['file.file_exists']('/etc/pki/postgres.key') -%}
- prereq:
- x509: /etc/pki/postgres.crt
{%- endif %}
- retry:
attempts: 5
interval: 30
postgres_crt:
x509.certificate_managed:
- name: /etc/pki/postgres.crt
- ca_server: {{ CA.server }}
- subjectAltName: DNS:{{ GLOBALS.hostname }}, IP:{{ GLOBALS.node_ip }}
- signing_policy: postgres
- private_key: /etc/pki/postgres.key
- CN: {{ GLOBALS.hostname }}
- days_remaining: 7
- days_valid: 820
- backup: True
- timeout: 30
- retry:
attempts: 5
interval: 30
postgresKeyperms:
file.managed:
- replace: False
- name: /etc/pki/postgres.key
- mode: 400
- user: 939
- group: 939
{% else %}
{{sls}}_state_not_allowed:
test.fail_without_changes:
- name: {{sls}}_state_not_allowed
{% endif %}
-157
View File
@@ -1,157 +0,0 @@
# Copyright Security Onion Solutions LLC and/or licensed to Security Onion Solutions LLC under one
# or more contributor license agreements. Licensed under the Elastic License 2.0 as shown at
# https://securityonion.net/license; you may not use this file except in compliance with the
# Elastic License 2.0.
{% from 'allowed_states.map.jinja' import allowed_states %}
{% if sls.split('.')[0] in allowed_states %}
{% from 'vars/globals.map.jinja' import GLOBALS %}
{% from 'telegraf/map.jinja' import TELEGRAFMERGED %}
{# postgres_wait_ready below requires `docker_container: so-postgres`, which is
declared in postgres.enabled. Include it here so state.apply postgres.telegraf_users
on its own (e.g. from orch.deploy_newnode) still has that ID in scope. Salt
de-duplicates the circular include. #}
include:
- postgres.enabled
{% set TG_OUT = TELEGRAFMERGED.output | upper %}
{% if TG_OUT in ['POSTGRES', 'BOTH'] %}
# docker_container.running returns as soon as the container starts, but on
# first-init docker-entrypoint.sh starts a temporary postgres with
# `listen_addresses=''` to run /docker-entrypoint-initdb.d scripts, then
# shuts it down before exec'ing the real CMD. A default pg_isready check
# (Unix socket) passes during that ephemeral phase and races the shutdown
# with "the database system is shutting down". Checking TCP readiness on
# 127.0.0.1 only succeeds after the final postgres binds the port.
postgres_wait_ready:
cmd.run:
- name: |
for i in $(seq 1 60); do
if docker exec so-postgres pg_isready -h 127.0.0.1 -U postgres -q 2>/dev/null; then
exit 0
fi
sleep 2
done
echo "so-postgres did not accept TCP connections within 120s" >&2
exit 1
- require:
- docker_container: so-postgres
# Ensure the shared Telegraf database exists. init-users.sh only runs on a
# fresh data dir, so hosts upgraded onto an existing /nsm/postgres volume
# would otherwise never get so_telegraf.
postgres_create_telegraf_db:
cmd.run:
- name: |
if ! docker exec so-postgres psql -U postgres -tAc "SELECT 1 FROM pg_database WHERE datname='so_telegraf'" | grep -q 1; then
docker exec so-postgres psql -v ON_ERROR_STOP=1 -U postgres -c "CREATE DATABASE so_telegraf"
fi
- require:
- cmd: postgres_wait_ready
# Provision the shared group role and schema once. Every per-minion role is a
# member of so_telegraf, and each Telegraf connection does SET ROLE so_telegraf
# (via options='-c role=so_telegraf' in the connection string) so tables created
# on first write are owned by the group role and every member can INSERT/SELECT.
postgres_telegraf_group_role:
cmd.run:
- name: |
docker exec -i so-postgres psql -v ON_ERROR_STOP=1 -U postgres -d so_telegraf <<'EOSQL'
DO $$
BEGIN
IF NOT EXISTS (SELECT FROM pg_catalog.pg_roles WHERE rolname = 'so_telegraf') THEN
CREATE ROLE so_telegraf NOLOGIN;
END IF;
END
$$;
GRANT CONNECT ON DATABASE so_telegraf TO so_telegraf;
CREATE SCHEMA IF NOT EXISTS telegraf AUTHORIZATION so_telegraf;
GRANT USAGE, CREATE ON SCHEMA telegraf TO so_telegraf;
CREATE SCHEMA IF NOT EXISTS partman;
CREATE EXTENSION IF NOT EXISTS pg_partman SCHEMA partman;
CREATE EXTENSION IF NOT EXISTS pg_cron;
-- Telegraf (running as so_telegraf) calls partman.create_parent()
-- on first write of each metric, which needs USAGE on the partman
-- schema, EXECUTE on its functions/procedures, and write access to
-- partman.part_config so it can register new partitioned parents.
GRANT USAGE, CREATE ON SCHEMA partman TO so_telegraf;
GRANT SELECT, INSERT, UPDATE, DELETE ON ALL TABLES IN SCHEMA partman TO so_telegraf;
GRANT EXECUTE ON ALL FUNCTIONS IN SCHEMA partman TO so_telegraf;
GRANT EXECUTE ON ALL PROCEDURES IN SCHEMA partman TO so_telegraf;
-- partman creates per-parent template tables (partman.template_*) at
-- runtime; default privileges extend DML/sequence access to them.
ALTER DEFAULT PRIVILEGES IN SCHEMA partman
GRANT SELECT, INSERT, UPDATE, DELETE ON TABLES TO so_telegraf;
ALTER DEFAULT PRIVILEGES IN SCHEMA partman
GRANT USAGE, SELECT, UPDATE ON SEQUENCES TO so_telegraf;
-- Hourly partman maintenance. cron.schedule is idempotent by jobname.
SELECT cron.schedule(
'telegraf-partman-maintenance',
'17 * * * *',
'CALL partman.run_maintenance_proc()'
);
EOSQL
- require:
- cmd: postgres_create_telegraf_db
{% set creds = salt['pillar.get']('telegraf:postgres_creds', {}) %}
{% for mid, entry in creds.items() %}
{% if entry.get('user') and entry.get('pass') %}
{% set u = entry.user %}
{% set p = entry.pass | replace("'", "''") %}
postgres_telegraf_role_{{ u }}:
cmd.run:
- name: |
docker exec -i so-postgres psql -v ON_ERROR_STOP=1 -U postgres -d so_telegraf <<'EOSQL'
DO $$
BEGIN
IF NOT EXISTS (SELECT FROM pg_catalog.pg_roles WHERE rolname = '{{ u }}') THEN
EXECUTE format('CREATE ROLE %I WITH LOGIN PASSWORD %L', '{{ u }}', '{{ p }}');
ELSE
EXECUTE format('ALTER ROLE %I WITH PASSWORD %L', '{{ u }}', '{{ p }}');
END IF;
END
$$;
GRANT CONNECT ON DATABASE so_telegraf TO "{{ u }}";
GRANT so_telegraf TO "{{ u }}";
EOSQL
- require:
- cmd: postgres_telegraf_group_role
{% endif %}
{% endfor %}
# Reconcile partman retention from pillar. Runs after role/schema setup so
# any partitioned parents Telegraf has already created get their retention
# refreshed whenever postgres.telegraf.retention_days changes.
{% set retention = salt['pillar.get']('postgres:telegraf:retention_days', 14) | int %}
postgres_telegraf_retention_reconcile:
cmd.run:
- name: |
docker exec -i so-postgres psql -v ON_ERROR_STOP=1 -U postgres -d so_telegraf <<'EOSQL'
DO $$
BEGIN
IF EXISTS (SELECT 1 FROM pg_catalog.pg_extension WHERE extname = 'pg_partman') THEN
UPDATE partman.part_config
SET retention = '{{ retention }} days',
retention_keep_table = false
WHERE parent_table LIKE 'telegraf.%';
END IF;
END
$$;
EOSQL
- require:
- cmd: postgres_telegraf_group_role
{% endif %}
{% else %}
{{sls}}_state_not_allowed:
test.fail_without_changes:
- name: {{sls}}_state_not_allowed
{% endif %}
@@ -1,39 +0,0 @@
#!/bin/bash
#
# Copyright Security Onion Solutions LLC and/or licensed to Security Onion Solutions LLC under one
# or more contributor license agreements. Licensed under the Elastic License 2.0 as shown at
# https://securityonion.net/license; you may not use this file except in compliance with the
# Elastic License 2.0.
. /usr/sbin/so-common
# Backups contain role password hashes and full chat data; keep them 0600.
umask 0077
TODAY=$(date '+%Y_%m_%d')
BACKUPDIR=/nsm/backup
BACKUPFILE="$BACKUPDIR/so-postgres-backup-$TODAY.sql.gz"
MAXBACKUPS=7
mkdir -p $BACKUPDIR
# Skip if already backed up today
if [ -f "$BACKUPFILE" ]; then
exit 0
fi
# Skip if container isn't running
if ! docker ps --format '{{.Names}}' | grep -q '^so-postgres$'; then
exit 0
fi
# Dump all databases and roles, compress
docker exec so-postgres pg_dumpall -U postgres | gzip > "$BACKUPFILE"
# Retention cleanup
NUMBACKUPS=$(find $BACKUPDIR -type f -name "so-postgres-backup*" | wc -l)
while [ "$NUMBACKUPS" -gt "$MAXBACKUPS" ]; do
OLDEST=$(find $BACKUPDIR -type f -name "so-postgres-backup*" -printf '%T+ %p\n' | sort | head -n 1 | awk -F" " '{print $2}')
rm -f "$OLDEST"
NUMBACKUPS=$(find $BACKUPDIR -type f -name "so-postgres-backup*" | wc -l)
done
@@ -1,80 +0,0 @@
#!/bin/bash
# Copyright Security Onion Solutions LLC and/or licensed to Security Onion Solutions LLC under one
# or more contributor license agreements. Licensed under the Elastic License 2.0 as shown at
# https://securityonion.net/license; you may not use this file except in compliance with the
# Elastic License 2.0.
. /usr/sbin/so-common
usage() {
echo "Usage: $0 <operation> [args]"
echo ""
echo "Supported Operations:"
echo " sql Execute a SQL command, requires: <sql>"
echo " sqlfile Execute a SQL file, requires: <path>"
echo " shell Open an interactive psql shell"
echo " dblist List databases"
echo " userlist List database roles"
echo ""
exit 1
}
if [ $# -lt 1 ]; then
usage
fi
# Check for prerequisites
if [ "$(id -u)" -ne 0 ]; then
echo "This script must be run using sudo!"
exit 1
fi
COMMAND=$(basename $0)
OP=$1
shift
set -eo pipefail
log() {
echo -e "$(date) | $COMMAND | $@" >&2
}
so_psql() {
docker exec so-postgres psql -U postgres -d securityonion "$@"
}
case "$OP" in
sql)
[ $# -lt 1 ] && usage
so_psql -c "$1"
;;
sqlfile)
[ $# -ne 1 ] && usage
if [ ! -f "$1" ]; then
log "File not found: $1"
exit 1
fi
docker cp "$1" so-postgres:/tmp/sqlfile.sql
docker exec so-postgres psql -U postgres -d securityonion -f /tmp/sqlfile.sql
docker exec so-postgres rm -f /tmp/sqlfile.sql
;;
shell)
docker exec -it so-postgres psql -U postgres -d securityonion
;;
dblist)
so_psql -c "\l"
;;
userlist)
so_psql -c "\du"
;;
*)
usage
;;
esac
@@ -1,10 +0,0 @@
#!/bin/bash
# Copyright Security Onion Solutions LLC and/or licensed to Security Onion Solutions LLC under one
# or more contributor license agreements. Licensed under the Elastic License 2.0 as shown at
# https://securityonion.net/license; you may not use this file except in compliance with the
# Elastic License 2.0.
. /usr/sbin/so-common
/usr/sbin/so-restart postgres $1
@@ -1,10 +0,0 @@
#!/bin/bash
# Copyright Security Onion Solutions LLC and/or licensed to Security Onion Solutions LLC under one
# or more contributor license agreements. Licensed under the Elastic License 2.0 as shown at
# https://securityonion.net/license; you may not use this file except in compliance with the
# Elastic License 2.0.
. /usr/sbin/so-common
/usr/sbin/so-start postgres $1
-10
View File
@@ -1,10 +0,0 @@
#!/bin/bash
# Copyright Security Onion Solutions LLC and/or licensed to Security Onion Solutions LLC under one
# or more contributor license agreements. Licensed under the Elastic License 2.0 as shown at
# https://securityonion.net/license; you may not use this file except in compliance with the
# Elastic License 2.0.
. /usr/sbin/so-common
/usr/sbin/so-stop postgres $1
-157
View File
@@ -1,157 +0,0 @@
#!/bin/bash
# Copyright Security Onion Solutions LLC and/or licensed to Security Onion Solutions LLC under one
# or more contributor license agreements. Licensed under the Elastic License 2.0 as shown at
# https://securityonion.net/license; you may not use this file except in compliance with the
# Elastic License 2.0.
# Point-in-time host metrics from the Telegraf Postgres backend.
# Sanity-check tool for verifying metrics are landing before the grid
# dashboards consume them.
#
# Assumes Telegraf's postgresql output is configured with
# tags_as_foreign_keys = true, tags_as_jsonb = true, fields_as_jsonb = true,
# so metric tables are (time, tag_id, fields jsonb) and tag tables are
# (tag_id, tags jsonb).
. /usr/sbin/so-common
usage() {
cat <<EOF
Usage: $0 [host]
Shows the most recent CPU, memory, disk, and load metrics for each host
from the so_telegraf Postgres database. Without an argument, reports on
every host that has data. With a host, limits output to that one.
Requires: sudo, so-postgres running, telegraf.output set to
POSTGRES or BOTH.
EOF
exit 1
}
if [ "$(id -u)" -ne 0 ]; then
echo "This script must be run using sudo!"
exit 1
fi
case "${1:-}" in
-h|--help) usage ;;
esac
FILTER_HOST="${1:-}"
SCHEMA="telegraf"
# Host values are interpolated into SQL below. Hostnames are [A-Za-z0-9._-];
# any other character in a tag value or CLI arg is rejected to prevent a
# stored-tag (or CLI) → SQL injection via a compromised Telegraf writer.
HOST_RE='^[A-Za-z0-9._-]+$'
if [ -n "$FILTER_HOST" ] && ! [[ "$FILTER_HOST" =~ $HOST_RE ]]; then
echo "Invalid host filter: $FILTER_HOST" >&2
exit 1
fi
so_psql() {
docker exec so-postgres psql -U postgres -d so_telegraf -At -F $'\t' "$@"
}
if ! docker exec so-postgres psql -U postgres -lqt 2>/dev/null | cut -d\| -f1 | grep -qw so_telegraf; then
echo "Database so_telegraf not found. Is telegraf.output set to POSTGRES or BOTH?"
exit 2
fi
table_exists() {
local table="$1"
[ -n "$(so_psql -c "SELECT 1 FROM information_schema.tables WHERE table_schema='${SCHEMA}' AND table_name='${table}' LIMIT 1;")" ]
}
# Discover hosts from cpu_tag (every minion reports cpu).
if ! table_exists "cpu_tag"; then
echo "${SCHEMA}.cpu_tag not found. Has Telegraf written any rows yet?"
exit 0
fi
HOSTS=$(so_psql -c "
SELECT DISTINCT tags->>'host'
FROM \"${SCHEMA}\".cpu_tag
WHERE tags ? 'host'
ORDER BY 1;")
if [ -z "$HOSTS" ]; then
echo "No hosts found in ${SCHEMA}. Is Telegraf configured to write to Postgres?"
exit 0
fi
print_metric() {
so_psql -c "$1"
}
for host in $HOSTS; do
if ! [[ "$host" =~ $HOST_RE ]]; then
echo "Skipping host with invalid characters in tag value: $host" >&2
continue
fi
if [ -n "$FILTER_HOST" ] && [ "$host" != "$FILTER_HOST" ]; then
continue
fi
echo "===================================================================="
echo " Host: $host"
echo "===================================================================="
if table_exists "cpu"; then
print_metric "
SELECT 'cpu ' AS metric,
to_char(c.time, 'YYYY-MM-DD HH24:MI:SS') AS ts,
round((100 - (c.fields->>'usage_idle')::numeric), 1) || '% used'
FROM \"${SCHEMA}\".cpu c
JOIN \"${SCHEMA}\".cpu_tag t USING (tag_id)
WHERE t.tags->>'host' = '${host}' AND t.tags->>'cpu' = 'cpu-total'
ORDER BY c.time DESC LIMIT 1;"
fi
if table_exists "mem"; then
print_metric "
SELECT 'memory ' AS metric,
to_char(m.time, 'YYYY-MM-DD HH24:MI:SS') AS ts,
round((m.fields->>'used_percent')::numeric, 1) || '% used (' ||
pg_size_pretty((m.fields->>'used')::bigint) || ' of ' ||
pg_size_pretty((m.fields->>'total')::bigint) || ')'
FROM \"${SCHEMA}\".mem m
JOIN \"${SCHEMA}\".mem_tag t USING (tag_id)
WHERE t.tags->>'host' = '${host}'
ORDER BY m.time DESC LIMIT 1;"
fi
if table_exists "disk"; then
print_metric "
SELECT 'disk ' || rpad(t.tags->>'path', 12) AS metric,
to_char(d.time, 'YYYY-MM-DD HH24:MI:SS') AS ts,
round((d.fields->>'used_percent')::numeric, 1) || '% used (' ||
pg_size_pretty((d.fields->>'used')::bigint) || ' of ' ||
pg_size_pretty((d.fields->>'total')::bigint) || ')'
FROM \"${SCHEMA}\".disk d
JOIN \"${SCHEMA}\".disk_tag t USING (tag_id)
WHERE t.tags->>'host' = '${host}'
AND d.time = (SELECT max(d2.time)
FROM \"${SCHEMA}\".disk d2
JOIN \"${SCHEMA}\".disk_tag t2 USING (tag_id)
WHERE t2.tags->>'host' = '${host}')
ORDER BY t.tags->>'path';"
fi
if table_exists "system"; then
print_metric "
SELECT 'load ' AS metric,
to_char(s.time, 'YYYY-MM-DD HH24:MI:SS') AS ts,
(s.fields->>'load1') || ' / ' ||
(s.fields->>'load5') || ' / ' ||
(s.fields->>'load15') || ' (1/5/15m)'
FROM \"${SCHEMA}\".system s
JOIN \"${SCHEMA}\".system_tag t USING (tag_id)
WHERE t.tags->>'host' = '${host}'
ORDER BY s.time DESC LIMIT 1;"
fi
echo ""
done
-27
View File
@@ -1,27 +0,0 @@
# Copyright Security Onion Solutions LLC and/or licensed to Security Onion Solutions LLC under one
# or more contributor license agreements. Licensed under the Elastic License 2.0 as shown at
# https://securityonion.net/license; you may not use this file except in compliance with the
# Elastic License 2.0.
# Fires for every event tagged 'so/pillar/changed'. Source of those events
# is the pg_notify_pillar engine on the salt-master, which in turn drains
# so_pillar.change_queue (populated by the AFTER trigger on
# so_pillar.pillar_entry — see 008_change_notify.sql).
#
# All routing logic — which pillar paths reload which services on which
# targets — lives in orch.so_pillar_reload so it stays editable as one
# YAML table without touching reactor wiring.
{% set payload = data.get('data', {}) %}
{% do salt.log.info('so_pillar_changed reactor: %s' % payload) %}
so_pillar_dispatch_reload:
runner.state.orchestrate:
- args:
- mods: orch.so_pillar_reload
- pillar:
so_pillar_change:
scope: {{ payload.get('scope') | json }}
role_name: {{ payload.get('role_name') | json }}
minion_id: {{ payload.get('minion_id') | json }}
changes: {{ payload.get('changes', []) | json }}
+18 -53
View File
@@ -6,74 +6,39 @@
# Elastic License 2.0.
import logging
import os
import re
import shlex
import subprocess
from subprocess import call
import yaml
log = logging.getLogger(__name__)
SO_MINION = '/usr/sbin/so-minion'
_NODETYPE_RE = re.compile(r'^[A-Z][A-Z0-9_]{0,31}$')
_MINIONID_RE = re.compile(r'^[A-Za-z0-9._-]{1,253}$')
_HOSTPART_RE = re.compile(r'^[A-Za-z0-9._-]{1,253}$')
_IPV4_RE = re.compile(
r'^(?:(?:25[0-5]|2[0-4]\d|[01]?\d?\d)\.){3}'
r'(?:25[0-5]|2[0-4]\d|[01]?\d?\d)$'
)
_HEAP_RE = re.compile(r'^\d{1,6}[kKmMgG]?$')
def _check(name, value, pattern):
s = str(value)
if not pattern.match(s):
raise ValueError("sominion_setup_reactor: refusing unsafe %s=%r" % (name, value))
return s
def run():
log.info('sominion_setup_reactor: Running')
minionid = data['id']
DATA = data['data']
hv_name = DATA['HYPERVISOR_HOST']
log.info('sominion_setup_reactor: DATA: %s' % DATA)
nodetype = _check('NODETYPE', DATA['NODETYPE'], _NODETYPE_RE)
argv = [
SO_MINION,
'-o=addVM',
'-m=' + _check('minionid', minionid, _MINIONID_RE),
'-n=' + _check('MNIC', DATA['MNIC'], _HOSTPART_RE),
'-i=' + _check('MAINIP', DATA['MAINIP'], _IPV4_RE),
'-c=' + str(int(DATA['CPUCORES'])),
'-d=' + str(DATA['NODE_DESCRIPTION']),
]
# Build the base command
cmd = "NODETYPE=" + DATA['NODETYPE'] + " /usr/sbin/so-minion -o=addVM -m=" + minionid + " -n=" + DATA['MNIC'] + " -i=" + DATA['MAINIP'] + " -c=" + str(DATA['CPUCORES']) + " -d='" + DATA['NODE_DESCRIPTION'] + "'"
# Add optional arguments only if they exist in DATA
if 'CORECOUNT' in DATA:
argv.append('-C=' + str(int(DATA['CORECOUNT'])))
cmd += " -C=" + str(DATA['CORECOUNT'])
if 'INTERFACE' in DATA:
argv.append('-a=' + _check('INTERFACE', DATA['INTERFACE'], _HOSTPART_RE))
cmd += " -a=" + DATA['INTERFACE']
if 'ES_HEAP_SIZE' in DATA:
argv.append('-e=' + _check('ES_HEAP_SIZE', DATA['ES_HEAP_SIZE'], _HEAP_RE))
cmd += " -e=" + DATA['ES_HEAP_SIZE']
if 'LS_HEAP_SIZE' in DATA:
argv.append('-l=' + _check('LS_HEAP_SIZE', DATA['LS_HEAP_SIZE'], _HEAP_RE))
cmd += " -l=" + DATA['LS_HEAP_SIZE']
if 'LSHOSTNAME' in DATA:
argv.append('-L=' + _check('LSHOSTNAME', DATA['LSHOSTNAME'], _HOSTPART_RE))
env = os.environ.copy()
env['NODETYPE'] = nodetype
log.info(
'sominion_setup_reactor: argv: %s (NODETYPE=%s)',
' '.join(shlex.quote(a) for a in argv),
shlex.quote(nodetype),
)
rc = subprocess.call(argv, shell=False, env=env)
cmd += " -L=" + DATA['LSHOSTNAME']
log.info('sominion_setup_reactor: Command: %s' % cmd)
rc = call(cmd, shell=True)
log.info('sominion_setup_reactor: rc: %s' % rc)
@@ -27,7 +27,6 @@ sool9_{{host}}:
log_file: /opt/so/log/salt/minion
grains:
hypervisor_host: {{host ~ "_" ~ role}}
sosmodel: HVGUEST
preflight_cmds:
- |
{%- set hostnames = [MANAGERHOSTNAME] %}
@@ -1,200 +0,0 @@
# Copyright Security Onion Solutions LLC and/or licensed to Security Onion Solutions LLC under one
# or more contributor license agreements. Licensed under the Elastic License 2.0 as shown at
# https://securityonion.net/license; you may not use this file except in compliance with the
# Elastic License 2.0.
# -*- coding: utf-8 -*-
"""
pg_notify_pillar Salt master engine that bridges so_pillar.change_queue
into the Salt event bus.
Architecture (see 008_change_notify.sql):
pillar_entry -- AFTER trigger --> change_queue (durable)
+ pg_notify('so_pillar_change') (wakeup)
|
LISTEN <-- this engine <-+
SELECT/UPDATE change_queue
|
fire_event('so/pillar/changed', ...)
|
reactor matches tag --> orch
Why a queue + notify rather than just notify: pg_notify is fire-and-forget
within a session. If the engine is down or the LISTEN connection is broken
when a write happens, the notification is lost forever. The change_queue
lets us recover on (re)connect, we drain everything still flagged
processed_at IS NULL.
Debounce: bulk operations (so-pillar-import, fresh installs) can fire
hundreds of notifications per second. The engine collects whatever lands in
a short window and emits one event per (scope, role, minion) tuple so the
reactor isn't stampeded.
"""
import json
import logging
import os
import select
import time
import salt.utils.event
log = logging.getLogger(__name__)
__virtualname__ = 'pg_notify_pillar'
DEFAULT_CHANNEL = 'so_pillar_change'
DEFAULT_DEBOUNCE_MS = 500
DEFAULT_RECONNECT_BACKOFF = 5
DEFAULT_BACKLOG_INTERVAL = 30
DEFAULT_BATCH_LIMIT = 500
EVENT_TAG = 'so/pillar/changed'
def __virtual__():
try:
import psycopg2 # noqa: F401
return __virtualname__
except ImportError:
return False, 'pg_notify_pillar engine requires psycopg2'
def start(dsn=None,
host='127.0.0.1',
port=5432,
dbname='securityonion',
user='so_pillar_master',
password=None,
channel=DEFAULT_CHANNEL,
debounce_ms=DEFAULT_DEBOUNCE_MS,
reconnect_backoff=DEFAULT_RECONNECT_BACKOFF,
backlog_interval=DEFAULT_BACKLOG_INTERVAL,
batch_limit=DEFAULT_BATCH_LIMIT,
password_file=None):
"""
Run the change-queue bridge until the master shuts the engine down.
Either pass a full ``dsn`` string, or supply discrete kwargs. The
password may also be read from ``password_file`` (mode 0400) so the
engine config in ``/etc/salt/master.d/`` doesn't have to embed it
inline only the file path.
"""
import psycopg2
import psycopg2.extensions
if dsn is None:
if password is None and password_file:
try:
with open(password_file, 'r') as fh:
password = fh.read().strip()
except (IOError, OSError) as exc:
log.error('pg_notify_pillar: cannot read password_file %s: %s',
password_file, exc)
return
dsn = _build_dsn(host=host, port=port, dbname=dbname,
user=user, password=password)
bus = salt.utils.event.get_master_event(
__opts__, __opts__['sock_dir'], listen=False)
log.info('pg_notify_pillar: starting (channel=%s debounce=%dms)',
channel, debounce_ms)
while True:
conn = None
try:
conn = psycopg2.connect(dsn)
conn.set_isolation_level(
psycopg2.extensions.ISOLATION_LEVEL_AUTOCOMMIT)
cur = conn.cursor()
cur.execute('LISTEN {0}'.format(channel))
log.info('pg_notify_pillar: connected; LISTEN %s', channel)
_drain(cur, bus, batch_limit)
while True:
ready, _, _ = select.select([conn], [], [], backlog_interval)
if not ready:
_drain(cur, bus, batch_limit)
continue
conn.poll()
_consume_notifies(conn)
if debounce_ms > 0:
time.sleep(debounce_ms / 1000.0)
conn.poll()
_consume_notifies(conn)
_drain(cur, bus, batch_limit)
except Exception as exc: # psycopg2.Error subclasses + OS errors
log.error('pg_notify_pillar: %s; reconnecting in %ds',
exc, reconnect_backoff)
finally:
if conn is not None:
try:
conn.close()
except Exception:
pass
time.sleep(reconnect_backoff)
def _build_dsn(host, port, dbname, user, password):
parts = ['host={0}'.format(host),
'port={0}'.format(port),
'dbname={0}'.format(dbname),
'user={0}'.format(user)]
if password:
parts.append('password={0}'.format(password))
return ' '.join(parts)
def _consume_notifies(conn):
# We don't use the payload directly — the queue table is the source of
# truth, and draining it covers any notifications we missed. So just
# discard them; their presence already proved there's something to drain.
while conn.notifies:
conn.notifies.pop(0)
def _drain(cur, bus, batch_limit):
"""Mark unprocessed change_queue rows processed and emit one event per
(scope, role_name, minion_id) group. SKIP LOCKED so multiple masters
sharing a Postgres don't double-process."""
cur.execute("""
UPDATE so_pillar.change_queue
SET processed_at = now()
WHERE id IN (
SELECT id FROM so_pillar.change_queue
WHERE processed_at IS NULL
ORDER BY id
FOR UPDATE SKIP LOCKED
LIMIT %s)
RETURNING id, scope, role_name, minion_id, pillar_path, op
""", (batch_limit,))
rows = cur.fetchall()
if not rows:
return
groups = {}
for row_id, scope, role_name, minion_id, pillar_path, op in rows:
key = (scope, role_name, minion_id)
groups.setdefault(key, []).append({
'queue_id': row_id,
'pillar_path': pillar_path,
'op': op,
})
for (scope, role_name, minion_id), changes in groups.items():
payload = {
'scope': scope,
'role_name': role_name,
'minion_id': minion_id,
'changes': changes,
}
log.debug('pg_notify_pillar: firing %s for %s',
EVENT_TAG, payload)
bus.fire_event(payload, EVENT_TAG)
-20
View File
@@ -1,27 +1,7 @@
engines_dirs:
- /etc/salt/engines
# All salt-master engines must be declared in this single file.
# Salt's master.d/*.conf merge replaces top-level lists rather than
# concatenating them, so a sibling .conf with its own `engines:` list
# would silently overwrite this one (only the last loaded file's list
# would survive). Anything new — including postsalt's pg_notify_pillar
# engine, gated on postgres:so_pillar:enabled below — gets appended
# here under the same `engines:` key.
engines:
{% if salt['pillar.get']('postgres:so_pillar:enabled', False) %}
- pg_notify_pillar:
host: {{ pillar.get('postgres', {}).get('host', '127.0.0.1') }}
port: {{ pillar.get('postgres', {}).get('port', 5432) }}
dbname: securityonion
user: so_pillar_master
password: {{ pillar['secrets']['pillar_master_pass'] }}
channel: so_pillar_change
debounce_ms: {{ pillar.get('postgres', {}).get('so_pillar', {}).get('engine_debounce_ms', 500) }}
reconnect_backoff: {{ pillar.get('postgres', {}).get('so_pillar', {}).get('engine_reconnect_backoff', 5) }}
backlog_interval: {{ pillar.get('postgres', {}).get('so_pillar', {}).get('engine_backlog_interval', 30) }}
batch_limit: {{ pillar.get('postgres', {}).get('so_pillar', {}).get('engine_batch_limit', 500) }}
{% endif %}
- checkmine:
interval: 60
- pillarWatch:
-3
View File
@@ -14,8 +14,6 @@
include:
- salt.minion
- salt.master.ext_pillar_postgres
- salt.master.pg_notify_pillar_engine
{% if 'vrt' in salt['pillar.get']('features', []) %}
- salt.cloud
- salt.cloud.reactor_config_hypervisor
@@ -63,7 +61,6 @@ engines_config:
file.managed:
- name: /etc/salt/master.d/engines.conf
- source: salt://salt/files/engines.conf
- template: jinja
# update the bootstrap script when used for salt-cloud
salt_bootstrap_cloud:
-46
View File
@@ -1,46 +0,0 @@
# Copyright Security Onion Solutions LLC and/or licensed to Security Onion Solutions LLC under one
# or more contributor license agreements. Licensed under the Elastic License 2.0 as shown at
# https://securityonion.net/license; you may not use this file except in compliance with the
# Elastic License 2.0.
# Drops /etc/salt/master.d/ext_pillar_postgres.conf so the salt-master loads
# pillar overlays from the so_pillar.* schema in so-postgres alongside the
# on-disk SLS pillar tree. Gated on the postgres:so_pillar:enabled feature
# flag (default false) so the file only appears once the schema is deployed
# and the importer has run at least once.
{% from 'allowed_states.map.jinja' import allowed_states %}
{% if 'salt.master' in allowed_states %}
{% if salt['pillar.get']('postgres:so_pillar:enabled', False) %}
ext_pillar_postgres_config:
file.managed:
- name: /etc/salt/master.d/ext_pillar_postgres.conf
- source: salt://salt/master/files/ext_pillar_postgres.conf.jinja
- template: jinja
- mode: '0640'
- user: root
- group: salt
- watch_in:
- service: salt_master_service
{% else %}
# When the flag is off make sure any previously-deployed config is removed
# so a rollback flips behavior cleanly.
ext_pillar_postgres_config_absent:
file.absent:
- name: /etc/salt/master.d/ext_pillar_postgres.conf
- watch_in:
- service: salt_master_service
{% endif %}
{% else %}
{{sls}}_state_not_allowed:
test.fail_without_changes:
- name: {{sls}}_state_not_allowed
{% endif %}
@@ -1,38 +0,0 @@
# /etc/salt/master.d/ext_pillar_postgres.conf
# Rendered by salt/salt/master/ext_pillar_postgres.sls.
# Reads the so_pillar.* schema in so-postgres and overlays it onto SLS pillar.
# SLS still renders first (ext_pillar_first: False) so bootstrap and mine-driven
# pillars work before Postgres is reachable; PG values overlay/override on top.
postgres:
host: {{ pillar.get('postgres', {}).get('host', '127.0.0.1') }}
port: {{ pillar.get('postgres', {}).get('port', 5432) }}
db: securityonion
user: so_pillar_master
pass: {{ pillar['secrets']['pillar_master_pass'] }}
ext_pillar_first: False
pillar_source_merging_strategy: smart
pillar_merge_lists: False
pillar_cache: True
pillar_cache_backend: disk
pillar_cache_ttl: {{ pillar.get('postgres', {}).get('so_pillar', {}).get('pillar_cache_ttl', 60) }}
# List form (not mapping form) so result rows merge into the pillar root rather
# than under a named subtree. Verified against salt/pillar/sql_base.py: list
# entries pass root=None to enter_root() which sets self.focus = self.result.
ext_pillar:
- postgres:
- query: "SELECT data FROM so_pillar.v_pillar_global WHERE %s IS NOT NULL ORDER BY sort_key, pillar_path"
as_json: True
ignore_null: True
- query: "SELECT data FROM so_pillar.v_pillar_role WHERE minion_id = %s ORDER BY sort_key, pillar_path"
as_json: True
ignore_null: True
- query: "SELECT data FROM so_pillar.v_pillar_minion WHERE minion_id = %s ORDER BY sort_key, pillar_path"
as_json: True
ignore_null: True
- query: "SELECT data FROM so_pillar.fn_pillar_secrets(%s)"
as_json: True
ignore_null: True
@@ -1,12 +0,0 @@
# /etc/salt/master.d/so_pillar_reactor.conf
# Wires the so/pillar/changed event tag — emitted by the pg_notify_pillar
# engine — to the so_pillar_changed reactor, which dispatches to
# orch.so_pillar_reload.
#
# Lives in its own file (rather than appended to reactor_hypervisor.conf)
# so the postgres:so_pillar:enabled flag can flip it on/off independently
# of hypervisor reactor wiring.
reactor:
- 'so/pillar/changed':
- /opt/so/saltstack/default/salt/reactor/so_pillar_changed.sls
@@ -1,80 +0,0 @@
# Copyright Security Onion Solutions LLC and/or licensed to Security Onion Solutions LLC under one
# or more contributor license agreements. Licensed under the Elastic License 2.0 as shown at
# https://securityonion.net/license; you may not use this file except in compliance with the
# Elastic License 2.0.
# Deploys the pg_notify_pillar engine module + its reactor config so the
# salt-master subscribes to so_pillar.change_queue and republishes changes
# on the salt event bus as so/pillar/changed. Reactor (so_pillar_changed.sls)
# matches that tag and dispatches the appropriate orch.
#
# The actual `engines:` declaration lives in salt/salt/files/engines.conf
# (jinja-rendered, also gated on postgres:so_pillar:enabled). It has to live
# in a single file because salt's master.d/*.conf merge replaces top-level
# lists rather than concatenating them — splitting `engines:` across multiple
# .conf files leaves only one loaded.
#
# Gated on the same postgres:so_pillar:enabled flag as the schema and
# ext_pillar config so the three components flip together.
{% from 'allowed_states.map.jinja' import allowed_states %}
{% if 'salt.master' in allowed_states %}
{% if salt['pillar.get']('postgres:so_pillar:enabled', False) %}
pg_notify_pillar_engine_module:
file.managed:
- name: /etc/salt/engines/pg_notify_pillar.py
- source: salt://salt/engines/master/pg_notify_pillar.py
- mode: '0644'
- user: root
- group: root
- makedirs: True
- watch_in:
- service: salt_master_service
pg_notify_pillar_reactor_config:
file.managed:
- name: /etc/salt/master.d/so_pillar_reactor.conf
- source: salt://salt/master/files/so_pillar_reactor.conf
- mode: '0644'
- user: root
- group: root
- watch_in:
- service: salt_master_service
{% else %}
# When the flag flips off, peel everything back so a rollback returns to
# pure-disk pillar with no orphan engine churning on a dead listen socket.
pg_notify_pillar_engine_module_absent:
file.absent:
- name: /etc/salt/engines/pg_notify_pillar.py
- watch_in:
- service: salt_master_service
pg_notify_pillar_engine_config_absent:
# No-op now: the engine config used to live in master.d/pg_notify_pillar_engine.conf
# but was folded into engines.conf to work around salt's master.d list-replace
# merge. Keep this state alive (no-op test.nop) so any old installs that
# still have the file get it cleaned up.
file.absent:
- name: /etc/salt/master.d/pg_notify_pillar_engine.conf
- watch_in:
- service: salt_master_service
pg_notify_pillar_reactor_config_absent:
file.absent:
- name: /etc/salt/master.d/so_pillar_reactor.conf
- watch_in:
- service: salt_master_service
{% endif %}
{% else %}
{{sls}}_state_not_allowed:
test.fail_without_changes:
- name: {{sls}}_state_not_allowed
{% endif %}
-5
View File
@@ -3,7 +3,6 @@ soc:
description: Enables or disables SOC. WARNING - Disabling this setting is unsupported and will cause the grid to malfunction. Re-enabling this setting is a manual effort via SSH.
forcedType: bool
advanced: True
readonly: True
telemetryEnabled:
title: SOC Telemetry
description: When this setting is enabled and the grid is not in airgap mode, SOC will provide feature usage data to the Security Onion development team via Google Analytics. This data helps Security Onion developers determine which product features are being used and can also provide insight into improving the user interface. When changing this setting, wait for the grid to fully synchronize and then perform a hard browser refresh on SOC, to force the browser cache to update and reflect the new setting.
@@ -891,16 +890,12 @@ soc:
suricata:
description: The template used when creating a new Suricata detection. [publicId] will be replaced with an unused Public Id.
multiline: True
forcedType: string
strelka:
description: The template used when creating a new Strelka detection.
multiline: True
forcedType: string
elastalert:
description: The template used when creating a new ElastAlert detection. [publicId] will be replaced with an unused Public Id.
multiline: True
forcedType: string
grid:
maxUploadSize:
description: The maximum number of bytes for an uploaded PCAP import file.
+1 -1
View File
@@ -261,7 +261,7 @@ strelka:
priority: 5
options:
limit: 1000
'ScanLNK':
'ScanLnk':
- positive:
flavors:
- 'lnk_file'
+1 -1
View File
@@ -15,7 +15,7 @@ from watchdog.observers import Observer
from watchdog.events import FileSystemEventHandler
with open("/opt/so/conf/strelka/filecheck.yaml", "r") as ymlfile:
cfg = yaml.safe_load(ymlfile)
cfg = yaml.load(ymlfile, Loader=yaml.Loader)
extract_path = cfg["filecheck"]["extract_path"]
historypath = cfg["filecheck"]["historypath"]
+1 -1
View File
@@ -99,7 +99,7 @@ strelka:
'ScanJpeg': *scannerOptions
'ScanJson': *scannerOptions
'ScanLibarchive': *scannerOptions
'ScanLNK': *scannerOptions
'ScanLnk': *scannerOptions
'ScanLsb': *scannerOptions
'ScanLzma': *scannerOptions
'ScanMacho': *scannerOptions
-1
View File
@@ -1,6 +1,5 @@
telegraf:
enabled: False
output: BOTH
config:
interval: '30s'
metric_batch_size: 1000
-44
View File
@@ -8,14 +8,6 @@
{%- set ZEEK_ENABLED = salt['pillar.get']('zeek:enabled', True) %}
{%- set MDENGINE = GLOBALS.md_engine %}
{%- set LOGSTASH_ENABLED = LOGSTASH_MERGED.enabled %}
{%- set TG_OUT = TELEGRAFMERGED.output | upper %}
{%- set PG_HOST = GLOBALS.manager_ip %}
{#- Per-minion telegraf creds live in the grid-wide telegraf/creds.sls pillar,
written by /usr/sbin/so-telegraf-cred on the manager. Each minion looks up
its own entry by grains.id. #}
{%- set PG_ENTRY = salt['pillar.get']('telegraf:postgres_creds:' ~ grains.id, {}) %}
{%- set PG_USER = PG_ENTRY.get('user', '') %}
{%- set PG_PASS = PG_ENTRY.get('pass', '') %}
# Global tags can be specified here in key="value" format.
[global_tags]
role = "{{ GLOBALS.role.split('-') | last }}"
@@ -80,7 +72,6 @@
# OUTPUT PLUGINS #
###############################################################################
{%- if TG_OUT in ['INFLUXDB', 'BOTH'] %}
# Configuration for sending metrics to InfluxDB
[[outputs.influxdb_v2]]
urls = ["https://{{ INFLUXDBHOST }}:8086"]
@@ -94,41 +85,6 @@
tls_key = "/etc/telegraf/telegraf.key"
## Use TLS but skip chain & host verification
# insecure_skip_verify = false
{%- endif %}
{%- if TG_OUT in ['POSTGRES', 'BOTH'] and PG_USER and PG_PASS %}
# Configuration for sending metrics to PostgreSQL.
# options='-c role=so_telegraf' makes every connection SET ROLE to the shared
# group role so tables created on first write are owned by so_telegraf, and
# all per-minion members can INSERT/SELECT them via role inheritance.
# fields_as_jsonb/tags_as_jsonb keep metric tables at a fixed column count so
# high-cardinality inputs (docker, procstat, kafka) don't blow past the
# Postgres 1600-column-per-table limit.
[[outputs.postgresql]]
connection = "host={{ PG_HOST }} port=5432 user={{ PG_USER }} password={{ PG_PASS }} dbname=so_telegraf sslmode=verify-full sslrootcert=/etc/telegraf/ca.crt options='-c role=so_telegraf'"
schema = "telegraf"
tags_as_foreign_keys = true
tags_as_jsonb = true
fields_as_jsonb = true
# Every metric table is a daily time-range partitioned parent managed by
# pg_partman. Retention drops old partitions instead of row-by-row DELETEs.
{% raw %}
# pg_partman 5.x requires the control column (time) to be NOT NULL, so
# ALTER it before create_parent(). And create_parent() splits
# p_parent_table on '.' to look up raw identifiers, so the literal must
# be 'schema.name' (not '"schema"."name"' as .table|quoteLiteral emits).
# IF NOT EXISTS keeps the three templates idempotent so a Telegraf
# restart after any DB-side surgery re-runs them safely.
create_templates = [
'''CREATE TABLE IF NOT EXISTS {{ .table }} ({{ .columns }}) PARTITION BY RANGE ("time")''',
'''ALTER TABLE {{ .table }} ALTER COLUMN "time" SET NOT NULL''',
'''SELECT partman.create_parent(p_parent_table := {{ printf "%s.%s" .table.Schema .table.Name | quoteLiteral }}, p_control := 'time', p_type := 'range', p_interval := '1 day', p_premake := 3) WHERE NOT EXISTS (SELECT 1 FROM partman.part_config WHERE parent_table = {{ printf "%s.%s" .table.Schema .table.Name | quoteLiteral }})'''
]
tag_table_create_templates = [
'''CREATE TABLE IF NOT EXISTS {{ .table }} ({{ .columns }}, PRIMARY KEY (tag_id))'''
]
{% endraw %}
{%- endif %}
###############################################################################
# PROCESSOR PLUGINS #
-9
View File
@@ -4,15 +4,6 @@ telegraf:
forcedType: bool
advanced: True
helpLink: influxdb
output:
description: Selects the backend(s) Telegraf writes metrics to. INFLUXDB keeps the current behavior; POSTGRES writes to the grid's Postgres instance; BOTH dual-writes for migration validation.
options:
- INFLUXDB
- POSTGRES
- BOTH
global: True
advanced: True
helpLink: influxdb
config:
interval:
description: Data collection interval.
-5
View File
@@ -68,7 +68,6 @@ base:
- backup.config_backup
- nginx
- influxdb
- postgres
- soc
- kratos
- hydra
@@ -96,7 +95,6 @@ base:
- backup.config_backup
- nginx
- influxdb
- postgres
- soc
- kratos
- hydra
@@ -125,7 +123,6 @@ base:
- registry
- nginx
- influxdb
- postgres
- strelka.manager
- soc
- kratos
@@ -156,7 +153,6 @@ base:
- registry
- nginx
- influxdb
- postgres
- strelka.manager
- soc
- kratos
@@ -185,7 +181,6 @@ base:
- manager
- nginx
- influxdb
- postgres
- strelka.manager
- soc
- kratos
-2
View File
@@ -1,5 +1,4 @@
{% from 'vars/elasticsearch.map.jinja' import ELASTICSEARCH_GLOBALS %}
{% from 'vars/postgres.map.jinja' import POSTGRES_GLOBALS %}
{% from 'vars/sensor.map.jinja' import SENSOR_GLOBALS %}
{% set ROLE_GLOBALS = {} %}
@@ -7,7 +6,6 @@
{% set EVAL_GLOBALS =
[
ELASTICSEARCH_GLOBALS,
POSTGRES_GLOBALS,
SENSOR_GLOBALS
]
%}
-2
View File
@@ -1,5 +1,4 @@
{% from 'vars/elasticsearch.map.jinja' import ELASTICSEARCH_GLOBALS %}
{% from 'vars/postgres.map.jinja' import POSTGRES_GLOBALS %}
{% from 'vars/sensor.map.jinja' import SENSOR_GLOBALS %}
{% set ROLE_GLOBALS = {} %}
@@ -7,7 +6,6 @@
{% set IMPORT_GLOBALS =
[
ELASTICSEARCH_GLOBALS,
POSTGRES_GLOBALS,
SENSOR_GLOBALS
]
%}
+1 -3
View File
@@ -1,14 +1,12 @@
{% from 'vars/elasticsearch.map.jinja' import ELASTICSEARCH_GLOBALS %}
{% from 'vars/logstash.map.jinja' import LOGSTASH_GLOBALS %}
{% from 'vars/postgres.map.jinja' import POSTGRES_GLOBALS %}
{% set ROLE_GLOBALS = {} %}
{% set MANAGER_GLOBALS =
[
ELASTICSEARCH_GLOBALS,
LOGSTASH_GLOBALS,
POSTGRES_GLOBALS
LOGSTASH_GLOBALS
]
%}
+1 -3
View File
@@ -1,14 +1,12 @@
{% from 'vars/elasticsearch.map.jinja' import ELASTICSEARCH_GLOBALS %}
{% from 'vars/logstash.map.jinja' import LOGSTASH_GLOBALS %}
{% from 'vars/postgres.map.jinja' import POSTGRES_GLOBALS %}
{% set ROLE_GLOBALS = {} %}
{% set MANAGERSEARCH_GLOBALS =
[
ELASTICSEARCH_GLOBALS,
LOGSTASH_GLOBALS,
POSTGRES_GLOBALS
LOGSTASH_GLOBALS
]
%}
-16
View File
@@ -1,16 +0,0 @@
{# Copyright Security Onion Solutions LLC and/or licensed to Security Onion Solutions LLC under one
or more contributor license agreements. Licensed under the Elastic License 2.0 as shown at
https://securityonion.net/license; you may not use this file except in compliance with the
Elastic License 2.0. #}
{% import 'vars/init.map.jinja' as INIT %}
{%
set POSTGRES_GLOBALS = {
'postgres': {}
}
%}
{% if salt['file.file_exists']('/opt/so/saltstack/local/pillar/postgres/auth.sls') %}
{% do POSTGRES_GLOBALS.postgres.update({'auth': INIT.PILLAR.postgres.auth}) %}
{% endif %}
-2
View File
@@ -1,6 +1,5 @@
{% from 'vars/elasticsearch.map.jinja' import ELASTICSEARCH_GLOBALS %}
{% from 'vars/logstash.map.jinja' import LOGSTASH_GLOBALS %}
{% from 'vars/postgres.map.jinja' import POSTGRES_GLOBALS %}
{% from 'vars/sensor.map.jinja' import SENSOR_GLOBALS %}
{% set ROLE_GLOBALS = {} %}
@@ -9,7 +8,6 @@
[
ELASTICSEARCH_GLOBALS,
LOGSTASH_GLOBALS,
POSTGRES_GLOBALS,
SENSOR_GLOBALS
]
%}
+32 -134
View File
@@ -202,10 +202,10 @@ check_service_status() {
systemctl status $service_name > /dev/null 2>&1
local status=$?
if [ $status -gt 0 ]; then
info "$service_name is not running"
info " $service_name is not running"
return 1;
else
info "$service_name is running"
info " $service_name is running"
return 0;
fi
@@ -821,7 +821,6 @@ create_manager_pillars() {
soc_pillar
idh_pillar
influxdb_pillar
postgres_pillar
logrotate_pillar
patch_pillar
nginx_pillar
@@ -1054,14 +1053,8 @@ generate_passwords(){
HYDRAKEY=$(get_random_value)
HYDRASALT=$(get_random_value)
REDISPASS=$(get_random_value)
POSTGRESPASS=$(get_random_value)
SOCSRVKEY=$(get_random_value 64)
IMPORTPASS=$(get_random_value)
# postsalt: salt-master connects to so_pillar.* as so_pillar_master, and the
# so-postgres container needs a symmetric key for pgcrypto-encrypted secrets.
# Both are generated here so they survive reinstall like the other secrets.
PILLARMASTERPASS=$(get_random_value)
SO_PILLAR_KEY=$(get_random_value 64)
}
generate_interface_vars() {
@@ -1362,12 +1355,6 @@ influxdb_pillar() {
" token: $INFLUXTOKEN" > $local_salt_dir/pillar/influxdb/token.sls
}
postgres_pillar() {
title "Create the postgres pillar file"
touch $adv_postgres_pillar_file
touch $postgres_pillar_file
}
make_some_dirs() {
mkdir -p /nsm
mkdir -p "$default_salt_dir"
@@ -1377,7 +1364,7 @@ make_some_dirs() {
mkdir -p $local_salt_dir/salt/firewall/portgroups
mkdir -p $local_salt_dir/salt/firewall/ports
for THEDIR in bpf elasticsearch ntp firewall redis backup influxdb postgres strelka sensoroni soc docker zeek suricata nginx telegraf logstash soc manager kratos hydra idh elastalert stig global kafka versionlock hypervisor vm; do
for THEDIR in bpf elasticsearch ntp firewall redis backup influxdb strelka sensoroni soc docker zeek suricata nginx telegraf logstash soc manager kratos hydra idh elastalert stig global kafka versionlock hypervisor vm; do
mkdir -p $local_salt_dir/pillar/$THEDIR
touch $local_salt_dir/pillar/$THEDIR/adv_$THEDIR.sls
touch $local_salt_dir/pillar/$THEDIR/soc_$THEDIR.sls
@@ -1554,8 +1541,13 @@ clear_previous_setup_results() {
reinstall_init() {
info "Putting system in state to run setup again"
# Always include both services. check_service_status skips units that aren't present.
local salt_services=( "salt-master" "salt-minion" )
if [[ $install_type =~ ^(MANAGER|EVAL|MANAGERSEARCH|MANAGERHYPE|STANDALONE|FLEET|IMPORT)$ ]]; then
local salt_services=( "salt-master" "salt-minion" )
else
local salt_services=( "salt-minion" )
fi
local service_retry_count=20
{
# remove all of root's cronjobs
@@ -1571,51 +1563,31 @@ reinstall_init() {
salt-call state.apply ca.remove -linfo --local --file-root=../salt
# Stop salt services and force-kill any lingering salt processes (including orphans
# from an earlier reinstall attempt where the unit file is gone but processes survive)
# so dnf remove salt can run cleanly
# Kill any salt processes (safely)
for service in "${salt_services[@]}"; do
# Stop the service in the background so we can exit after a certain amount of time
if check_service_status "$service"; then
info "Stopping $service via systemctl"
systemctl stop "$service"
systemctl stop "$service" &
fi
done
local pid=$!
# Unconditionally force-kill any remaining salt binaries — these may be orphaned
# from a prior aborted reinstall (no unit file, so systemctl can't see them).
for salt_bin in salt-master salt-minion salt-call salt-cloud; do
if pgrep -f "/usr/bin/${salt_bin}" > /dev/null 2>&1; then
info "Force-killing lingering $salt_bin processes"
pkill -9 -ef "/usr/bin/${salt_bin}" 2>/dev/null
fi
done
# Catch stray `salt` CLI children from saltutil.kill_all_jobs / state.apply invocations
pkill -9 -ef "/usr/bin/python3 /bin/salt" 2>/dev/null
local count=0
while check_service_status "$service"; do
if [[ $count -gt $service_retry_count ]]; then
echo "Could not stop $service after 1 minute, exiting setup."
# Give the kernel a moment to reap the killed processes before dnf removes the binaries
local kill_wait=0
while pgrep -f "/usr/bin/salt-" > /dev/null 2>&1; do
if [[ $kill_wait -gt 10 ]]; then
info "Salt processes still present after SIGKILL + 10s wait; proceeding anyway"
pgrep -af "/usr/bin/salt-" | while read -r line; do info " lingering: $line"; done
break
fi
sleep 1
((kill_wait++))
# Stop the systemctl process trying to kill the service, show user a message, then exit setup
kill -9 $pid
fail_setup
fi
sleep 5
((count++))
done
done
# Clear the 'failed' state SIGKILL left on the units before removing the package
systemctl reset-failed salt-master.service salt-minion.service 2>/dev/null || true
# Remove all salt configs
dnf -y remove salt
rm -rf /etc/salt/ /var/cache/salt/
# Drop systemd's in-memory references to the now-removed units
systemctl daemon-reload
# Uninstall local Elastic Agent, if installed
elastic-agent uninstall -f
rm -rf /etc/salt/engines/* /etc/salt/grains /etc/salt/master /etc/salt/master.d/* /etc/salt/minion /etc/salt/minion.d/* /etc/salt/pki/* /etc/salt/proxy /etc/salt/proxy.d/* /var/cache/salt/
if command -v docker &> /dev/null; then
# Stop and remove all so-* containers so files can be changed with more safety
@@ -1639,7 +1611,10 @@ reinstall_init() {
backup_dir /nsm/hydra "$date_string"
backup_dir /nsm/influxdb "$date_string"
} 2>&1 | tee -a "$setup_log"
# Uninstall local Elastic Agent, if installed
elastic-agent uninstall -f
} >> "$setup_log" 2>&1
info "System reinstall init has been completed."
}
@@ -1706,24 +1681,6 @@ remove_package() {
fi
}
ensure_pyyaml() {
title "Ensuring python3-pyyaml is installed"
if rpm -q python3-pyyaml >/dev/null 2>&1; then
info "python3-pyyaml already installed"
return 0
fi
info "python3-pyyaml not found, attempting to install"
set -o pipefail
dnf -y install python3-pyyaml 2>&1 | tee -a "$setup_log"
local result=$?
set +o pipefail
if [[ $result -ne 0 ]] || ! rpm -q python3-pyyaml >/dev/null 2>&1; then
error "Failed to install python3-pyyaml (exit=$result)"
fail_setup
fi
info "python3-pyyaml installed successfully"
}
# When updating the salt version, also update the version in securityonion-builds/images/iso-task/Dockerfile and salt/salt/master.defaults.yaml and salt/salt/minion.defaults.yaml
# CAUTION! SALT VERSION UDDATES - READ BELOW
# When updating the salt version, also update the version in:
@@ -1875,67 +1832,8 @@ secrets_pillar(){
printf '%s\n'\
"secrets:"\
" import_pass: $IMPORTPASS"\
" influx_pass: $INFLUXPASS"\
" pillar_master_pass: $PILLARMASTERPASS"\
" postgres_pass: $POSTGRESPASS" > $local_salt_dir/pillar/secrets.sls
elif ! grep -q '^[[:space:]]*pillar_master_pass:' $local_salt_dir/pillar/secrets.sls; then
# Existing install pre-postsalt — append the new key without disturbing
# the values already on disk. Keys we already wrote stay; only the new
# pillar_master_pass is added.
info "Appending pillar_master_pass to existing Secrets Pillar"
if [ -z "$PILLARMASTERPASS" ]; then
PILLARMASTERPASS=$(get_random_value)
fi
printf ' pillar_master_pass: %s\n' "$PILLARMASTERPASS" >> $local_salt_dir/pillar/secrets.sls
" influx_pass: $INFLUXPASS" > $local_salt_dir/pillar/secrets.sls
fi
# postsalt: write the so_pillar pgcrypto master key to a 0400 file owned by
# root. The key itself is never read by Salt — schema_pillar.sls loads it
# into the so-postgres container via ALTER ROLE so_pillar_secret_owner SET
# so_pillar.master_key = '<key>'; the file just lets the value survive
# container restarts.
if [ ! -f /opt/so/conf/postgres/so_pillar.key ]; then
info "Generating so_pillar pgcrypto master key"
mkdir -p /opt/so/conf/postgres
if [ -z "$SO_PILLAR_KEY" ]; then
SO_PILLAR_KEY=$(get_random_value 64)
fi
# Subshell-scope the umask so it doesn't leak into subsequent so-setup
# (and salt-call) file writes. Without the (...) wrapper the umask 077
# persists for the rest of the install and every state-rendered config
# file under /opt/so/conf lands at 0600 — which breaks containers that
# bind-mount their config and run as a non-root user (the influxdb
# container, in particular, exits with "permission denied" on
# /conf/config.yaml after the gosu drop).
(
umask 077
printf '%s' "$SO_PILLAR_KEY" > /opt/so/conf/postgres/so_pillar.key
)
chmod 0400 /opt/so/conf/postgres/so_pillar.key
chown root:root /opt/so/conf/postgres/so_pillar.key
fi
}
# postsalt: flip postgres:so_pillar:enabled to True in the local pillar so
# the schema_pillar / ext_pillar_postgres / pg_notify_pillar engine states
# all activate during the install highstate. Without this the entire
# PG-canonical pillar stack short-circuits on its default-False gate and
# the install ends in legacy disk-pillar mode — defeating the point of
# being on postsalt at all. The companion enabled=False rollback just
# rewrites this file or removes the flag.
enable_so_pillar_postgres() {
local pillar_dir=/opt/so/saltstack/local/pillar/postgres
mkdir -p "$pillar_dir"
cat > "$pillar_dir/adv_postgres.sls" <<'EOPILLAR'
# postsalt: enable PG-canonical pillar mode. Generated by setup/so-functions
# during install. Flip to False here (or delete this file) to roll back to
# disk-pillar mode without wiping the so_pillar.* schema in so-postgres.
postgres:
so_pillar:
enabled: True
EOPILLAR
chown -R socore:socore "$pillar_dir"
chmod 0644 "$pillar_dir/adv_postgres.sls"
}
set_network_dev_status_list() {
+2 -26
View File
@@ -66,9 +66,6 @@ set_timezone
# Let's see what OS we are dealing with here
detect_os
# Ensure python3-pyyaml is available before any code that may need so-yaml/PyYAML
ensure_pyyaml
# Check to see if this is the setup type of "desktop".
is_desktop=
@@ -222,7 +219,7 @@ if [ -n "$test_profile" ]; then
WEBUSER=onionuser@somewhere.invalid
WEBPASSWD1=0n10nus3r
WEBPASSWD2=0n10nus3r
NODE_DESCRIPTION="${HOSTNAME} - ${install_type} - ${MSRVIP_OFFSET}"
NODE_DESCRIPTION="${HOSTNAME} - ${install_type} - ${MAINIP}"
update_sudoers_for_testing
fi
@@ -795,31 +792,10 @@ if ! [[ -f $install_opt_file ]]; then
checkin_at_boot
set_initial_firewall_access
initialize_elasticsearch_indices "so-case so-casehistory so-assistant-session so-assistant-chat"
# run a final highstate before enabling scheduled highstates.
# run a final highstate before enabling scheduled highstates.
# this will ensure so-elasticsearch-ilm-policy-load and so-elasticsearch-templates-load have a chance to run after elasticfleet is setup
info "Running final highstate for setup"
logCmd "salt-call state.highstate -l info"
# postsalt: enable PG-canonical pillar mode now that the install is
# fully on disk. We can't flip the flag earlier — ext_pillar overlay
# would replace the elasticsearch subtree (and others) with what's
# in PG before the importer has run, dropping secrets-allowlisted
# subkeys like elasticsearch.auth.users.so_elastic_user.pass that
# elasticsearch.enabled.sls needs to render. Order:
# 1. drop adv_postgres.sls flipping the flag
# 2. refresh_pillar so the next state sees enabled=True
# 3. apply postgres.schema_pillar — deploys schema, ALTERs role
# passwords, installs psycopg2 into salt's bundled python,
# runs so-pillar-import, writes /opt/so/conf/so-yaml/mode=postgres
# 4. apply salt.master — re-renders engines.conf with the
# pg_notify_pillar engine block, drops master.d ext_pillar
# config, watch_in restarts salt-master, ext_pillar takes over
info "Enabling postsalt PG-canonical pillar mode"
enable_so_pillar_postgres
logCmd "salt-call saltutil.refresh_pillar"
logCmd "salt-call state.apply postgres.schema_pillar -l info"
logCmd "salt-call state.apply salt.master -l info"
logCmd "salt-call schedule.enable -linfo --local"
verify_setup
else
-6
View File
@@ -202,12 +202,6 @@ export influxdb_pillar_file
adv_influxdb_pillar_file="$local_salt_dir/pillar/influxdb/adv_influxdb.sls"
export adv_influxdb_pillar_file
postgres_pillar_file="$local_salt_dir/pillar/postgres/soc_postgres.sls"
export postgres_pillar_file
adv_postgres_pillar_file="$local_salt_dir/pillar/postgres/adv_postgres.sls"
export adv_postgres_pillar_file
logrotate_pillar_file="$local_salt_dir/pillar/logrotate/soc_logrotate.sls"
export logrotate_pillar_file
+1 -2
View File
@@ -71,8 +71,7 @@ log_has_errors() {
grep -vE "remove_failed_vm.sls" | \
grep -vE "failed to copy: httpReadSeeker" | \
grep -vE "Error response from daemon: failed to resolve reference" | \
grep -vE "log-.*-pipeline_failed_attempts" | \
grep -vE " -v ON_ERROR_STOP=1" &> "$error_log"
grep -vE "log-.*-pipeline_failed_attempts" &> "$error_log"
if [[ $? -eq 0 ]]; then
# This function succeeds (returns 0) if errors are detected