Merge pull request #15810 from Security-Onion-Solutions/feature/postgres

Split postgres auth pillar from per-minion telegraf creds

pillar/telegraf/creds.sls

@@ -0,0 +1,12 @@
+# 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: {}

pillar/top.sls

@@ -17,6 +17,7 @@ base:
     - sensoroni.adv_sensoroni
     - telegraf.soc_telegraf
     - telegraf.adv_telegraf
+    - telegraf.creds
     - versionlock.soc_versionlock
     - versionlock.adv_versionlock
     - soc.license

salt/manager/tools/sbin/so-minion

@@ -281,22 +281,18 @@ function deleteMinionFiles () {
 	fi
 }
 
-# Remove this minion's postgres Telegraf credential from both the aggregate
+# Remove this minion's postgres Telegraf credential from the shared creds
-# pillar and the postgres database. Paired with add_telegraf_to_minion:
+# pillar and drop the matching role in Postgres. Always returns 0 so a dead
-# add/delete cycle both here and in the DB. Always returns 0 so a dead or
+# or unreachable so-postgres doesn't block minion deletion — in that case we
-# 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}"
-	local AGGREGATE=/opt/so/saltstack/local/pillar/postgres/auth.sls
 
 	log "INFO" "Removing postgres telegraf cred for $MINION_ID"
 
-	if [[ -f "$AGGREGATE" ]]; then
+	so-telegraf-cred remove "$MINION_ID" >/dev/null 2>&1 || true
-		so-yaml.py remove "$AGGREGATE" "postgres.auth.users.telegraf_${MINION_SAFE}" >/dev/null 2>&1 || true
-	fi
 
 	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
@@ -311,7 +307,7 @@ END
 \$\$;
 EOSQL
 		then
-			log "WARN" "Failed to drop postgres role $PG_USER; aggregate pillar entry was removed — drop manually if the role persists"
+			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"
@@ -580,35 +576,15 @@ function add_telegraf_to_minion() {
         return 1
     fi
 
-    # Provision the per-minion postgres Telegraf credential so telegraf.conf
+    # Provision the per-minion postgres Telegraf credential in the shared
-    # renders correctly on the minion's first highstate and postgres.telegraf_users
+    # telegraf/creds.sls pillar. so-telegraf-cred is the only writer; it
-    # picks up the matching aggregate entry on the next manager apply.
+    # 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
-    # Writes:
+    # manager creates/updates the DB role from the same pillar.
-    #   - postgres.telegraf.{user,pass} into the minion's own pillar file
+    so-telegraf-cred add "$MINION_ID"
-    #     (distributed to only this minion via pillar/top.sls).
+    if [ $? -ne 0 ]; then
-    #   - postgres.auth.users.telegraf_<safe>.{user,pass} into the aggregate
+        log "ERROR" "Failed to provision postgres telegraf cred for $MINION_ID"
-    #     pillar so postgres.telegraf_users CREATE ROLE finds it.
+        return 1
-    #
-    # An existing password is reused if the aggregate already has one (re-add),
-    # so rerunning so-minion for the same minion keeps the cred stable.
-    local MINION_SAFE
-    MINION_SAFE=$(echo "$MINION_ID" | tr '.-' '__' | tr '[:upper:]' '[:lower:]')
-    local PG_USER="so_telegraf_${MINION_SAFE}"
-    local AGGREGATE=/opt/so/saltstack/local/pillar/postgres/auth.sls
-    local PG_PASS=""
-    if [[ -f "$AGGREGATE" ]]; then
-        PG_PASS=$(so-yaml.py get -r "$AGGREGATE" "postgres.auth.users.telegraf_${MINION_SAFE}.pass" 2>/dev/null || true)
-    fi
-    if [[ -z "$PG_PASS" ]]; then
-        PG_PASS=$(tr -dc 'A-Za-z0-9~!@#^&*()_=+[]|;:,.<>?-' < /dev/urandom | head -c 72)
-    fi
-
-    so-yaml.py replace "$PILLARFILE" postgres.telegraf.user "$PG_USER" >/dev/null
-    so-yaml.py replace "$PILLARFILE" postgres.telegraf.pass "$PG_PASS" >/dev/null
-    if [[ -f "$AGGREGATE" ]]; then
-        so-yaml.py replace "$AGGREGATE" "postgres.auth.users.telegraf_${MINION_SAFE}.user" "$PG_USER" >/dev/null
-        so-yaml.py replace "$AGGREGATE" "postgres.auth.users.telegraf_${MINION_SAFE}.pass" "$PG_PASS" >/dev/null
     fi
 }
 

salt/manager/tools/sbin/so-telegraf-cred

@@ -0,0 +1,54 @@
+#!/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 "$CREDS")"
+    if [[ ! -f "$CREDS" ]]; then
+        (umask 027 && printf 'telegraf:\n  postgres_creds: {}\n' > "$CREDS")
+        chown socore:socore "$CREDS" 2>/dev/null || true
+        chmod 640 "$CREDS"
+    fi
+}
+
+OP=$1
+MID=$2
+[[ -z "$OP" || -z "$MID" ]] && usage
+
+case "$OP" in
+    add)
+        SAFE=$(echo "$MID" | tr '.-' '__' | tr '[:upper:]' '[:lower:]')
+        seed_creds_file
+        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

salt/manager/tools/sbin/soup

@@ -490,32 +490,16 @@ up_to_3.1.0() {
 post_to_3.1.0() {
   /usr/sbin/so-kibana-space-defaults
 
-  # One-time backfill for minions that existed before the postgres Telegraf
+  # Backfill the Telegraf creds pillar for every accepted minion. so-telegraf-cred
-  # feature shipped. postgres.auth's up_minions fallback loop generates any
+  # add is idempotent — it no-ops when an entry already exists — so this is safe
-  # missing aggregate pillar entries; postgres.telegraf_users CREATEs the
+  # to run on every soup. The subsequent state.apply creates/updates the matching
-  # matching DB roles; then the bash loop below copies each minion's cred
+  # Postgres roles from the reconciled pillar.
-  # into its own pillar file. Going forward, so-minion owns add/delete for
+  echo "Reconciling Telegraf Postgres creds for accepted minions."
-  # every new minion, so this backfill is only needed on the upgrade boundary.
+  for mid in $(salt-key --out=json --list=accepted 2>/dev/null | jq -r '.minions[]?' 2>/dev/null); do
-  echo "Provisioning Telegraf Postgres users for existing minions."
+    [[ -n "$mid" ]] || continue
-  salt-call --local state.apply postgres.auth,postgres.telegraf_users queue=True || true
+    /usr/sbin/so-telegraf-cred add "$mid" || echo "  warning: so-telegraf-cred add $mid failed" >&2
-
+  done
-  AGGREGATE_PILLAR=/opt/so/saltstack/local/pillar/postgres/auth.sls
+  salt-call --local state.apply postgres.telegraf_users queue=True || true
-  MINIONS_DIR=/opt/so/saltstack/local/pillar/minions
-  if [[ -f "$AGGREGATE_PILLAR" && -d "$MINIONS_DIR" ]]; then
-    for pillar_file in "$MINIONS_DIR"/*.sls; do
-      [[ -f "$pillar_file" ]] || continue
-      mid=$(basename "$pillar_file" .sls)
-      [[ "$mid" == adv_* ]] && continue
-      safe=$(echo "$mid" | tr '.-' '__' | tr '[:upper:]' '[:lower:]')
-      existing_user=$(so-yaml.py get -r "$pillar_file" postgres.telegraf.user 2>/dev/null || true)
-      [[ "$existing_user" == "so_telegraf_${safe}" ]] && continue
-      user=$(so-yaml.py get -r "$AGGREGATE_PILLAR" "postgres.auth.users.telegraf_${safe}.user" 2>/dev/null || true)
-      pass=$(so-yaml.py get -r "$AGGREGATE_PILLAR" "postgres.auth.users.telegraf_${safe}.pass" 2>/dev/null || true)
-      [[ -z "$user" || -z "$pass" ]] && continue
-      so-yaml.py replace "$pillar_file" postgres.telegraf.user "$user" >/dev/null
-      so-yaml.py replace "$pillar_file" postgres.telegraf.pass "$pass" >/dev/null
-    done
-  fi
 
   POSTVERSION=3.1.0
 }

salt/orch/deploy_newnode.sls

@@ -25,8 +25,33 @@ 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

salt/postgres/auth.sls

@@ -13,24 +13,8 @@
   {% 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)) %}
 
-  {# Per-minion Telegraf Postgres credentials. Merge currently-up minions with any #}
+# Admin cred only. Per-minion Telegraf creds live in telegraf/creds.sls,
-  {# previously-known entries in pillar so existing passwords persist across runs. #}
+# managed by /usr/sbin/so-telegraf-cred (called from so-minion).
-  {% set existing = salt['pillar.get']('postgres:auth:users', {}) %}
-  {% set up_minions = salt['saltutil.runner']('manage.up') or [] %}
-  {% set telegraf_users = {} %}
-  {% for key, entry in existing.items() %}
-    {%- if key.startswith('telegraf_') and entry.get('user') and entry.get('pass') %}
-      {%- do telegraf_users.update({key: entry}) %}
-    {%- endif %}
-  {% endfor %}
-  {% for mid in up_minions %}
-    {%- set safe = mid | replace('.','_') | replace('-','_') | lower %}
-    {%- set key = 'telegraf_' ~ safe %}
-    {%- if key not in telegraf_users %}
-      {%- do telegraf_users.update({key: {'user': 'so_telegraf_' ~ safe, 'pass': salt['random.get_str'](72, chars=CHARS)}}) %}
-    {%- endif %}
-  {% endfor %}
-
 postgres_auth_pillar:
   file.managed:
     - name: /opt/so/saltstack/local/pillar/postgres/auth.sls
@@ -43,11 +27,6 @@ postgres_auth_pillar:
               so_postgres_user:
                 user: so_postgres
                 pass: "{{ so_postgres_user_pass }}"
-              {% for key, entry in telegraf_users.items() %}
-              {{ key }}:
-                user: {{ entry.user }}
-                pass: "{{ entry.pass }}"
-              {% endfor %}
     - show_changes: False
 {% else %}
 

salt/postgres/telegraf_users.sls

@@ -10,7 +10,7 @@
 
 {# 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 (from the reactor orch or from soup) still has that ID in scope. Salt
+   on its own (e.g. from orch.deploy_newnode) still has that ID in scope. Salt
    de-duplicates the circular include. #}
 include:
   - postgres.enabled
@@ -96,9 +96,9 @@ postgres_telegraf_group_role:
     - require:
       - cmd: postgres_create_telegraf_db
 
-{%   set users = salt['pillar.get']('postgres:auth:users', {}) %}
+{%   set creds = salt['pillar.get']('telegraf:postgres_creds', {}) %}
-{%   for key, entry in users.items() %}
+{%   for mid, entry in creds.items() %}
-{%     if key.startswith('telegraf_') and entry.get('user') and entry.get('pass') %}
+{%     if entry.get('user') and entry.get('pass') %}
 {%       set u = entry.user %}
 {%       set p = entry.pass | replace("'", "''") %}
 

salt/telegraf/etc/telegraf.conf

@@ -10,12 +10,12 @@
 {%- set LOGSTASH_ENABLED = LOGSTASH_MERGED.enabled %}
 {%- set TG_OUT = TELEGRAFMERGED.output | upper %}
 {%- set PG_HOST = GLOBALS.manager_ip %}
-{#- Per-minion telegraf creds are written into the minion's own pillar file
+{#- Per-minion telegraf creds live in the grid-wide telegraf/creds.sls pillar,
-    (/opt/so/saltstack/local/pillar/minions/<id>.sls) by postgres.auth on the
+    written by /usr/sbin/so-telegraf-cred on the manager. Each minion looks up
-    manager. Each minion only sees its own password — the aggregate map in
+    its own entry by grains.id. #}
-    postgres:auth:users is manager-scoped. #}
+{%- set PG_ENTRY = salt['pillar.get']('telegraf:postgres_creds:' ~ grains.id, {}) %}
-{%- set PG_USER = salt['pillar.get']('postgres:telegraf:user', '') %}
+{%- set PG_USER = PG_ENTRY.get('user', '') %}
-{%- set PG_PASS = salt['pillar.get']('postgres:telegraf:pass', '') %}
+{%- set PG_PASS = PG_ENTRY.get('pass', '') %}
 # Global tags can be specified here in key="value" format.
 [global_tags]
   role = "{{ GLOBALS.role.split('-') | last }}"