Files
securityonion/salt/reactor/push_pillar.sls
Josh Patterson 53f9ebcd46 FIX: queue auto-applied state runs instead of failing on conflict
orch.push_batch passed `kwarg: {queue: 2}` to salt.state, but in Salt
3006 queue is a top-level salt.state argument and salt.state always sets
the minion's queue kwarg from it (default False), so the kwarg block was
silently dropped and every pushed state ran with queue=False. The drainer
dispatches a separate async orchestration each 15s pass, so settings saved
more than ~15s apart overlap on the same minion and every run after the
first fails immediately with 'The function "state.sls" is running as PID
...'. The change then waits for the next scheduled highstate.

Seen on a 3.4.0 standalone: hydra.enabled, telegraf.output, and two soc
settings (including soc.config.licenseKey) were saved within 30s. The soc
state was dispatched while the telegraf state was still running and was
rejected, so the license key was not applied.

Use `queue: True`, as orch.deploy_newnode already does. An int is treated
as max_queue and still falls through to the conflict error once that many
state runs are active.

The failure was only visible in the master log, since the drainer
dispatches with --async and logged only "dispatch accepted". The drainer
now:
  - logs each dispatched action
  - parses the orchestration jid from salt-run's stderr (the only place
    --async reports it) and records it under /opt/so/state/push_dispatched
  - on later passes looks each jid up with jobs.lookup_jid and logs either
    "push succeeded" or an ERROR with the failed step, the per-minion
    failed states or rejection text, and the triggering paths
Lookups run outside the pending-intent lock since the reactors share it.

The beacon now logs each audit_settings row it emits and the reactor logs
the audit row id, so a single change can be traced from audit_settings to
its push result.

Adds so-push-drainer_test.py; the drainer is now held to the 100% coverage
requirement in python-test.

Verified on the standalone: a soc push dispatched while a 90s state run
was in progress queued behind it (queue=True in the job args), completed,
and the drainer logged "push succeeded" for its jid. The new result
parsing reports the original soc conflict and the hydra license failure
from the job cache.
2026-09-30 16:18:37 -04:00

178 lines
6.1 KiB
YAML+Jinja

#!py
# Reactor invoked by the postgres_pillar_beacon when SOC records settings changes in
# the securityonion.audit_settings table (see salt/_beacons/postgres_pillar_beacon.py). The beacon
# emits one event per new row carrying setting_id and node_id.
#
# Two branches, keyed on node_id:
# A) node_id populated -> the change is scoped to that one minion. Look up the
# app in pillar_push_map.yaml and write an intent that runs the app's mapped
# state(s) targeted to just that node.
# B) node_id empty -> grid-wide app change. Look up the app in
# pillar_push_map.yaml and write an intent with the entry's actions as-is.
#
# The app name is the first dotted segment of setting_id (e.g. "telegraf.output"
# -> "telegraf"), which matches the pillar_push_map.yaml keys 1:1.
#
# Reactors never dispatch directly. The so-push-drainer schedule picks up
# ready intents, dedupes across pending files, and dispatches orch.push_batch.
import fcntl
import json
import logging
import os
import time
from salt.client import Caller
import yaml
LOG = logging.getLogger(__name__)
PENDING_DIR = '/opt/so/state/push_pending'
LOCK_FILE = os.path.join(PENDING_DIR, '.lock')
MAX_PATHS = 20
# The pillar_push_map.yaml is shipped via salt:// but the reactor runs on the
# master, which mounts the default saltstack tree at this path.
PUSH_MAP_PATH = '/opt/so/saltstack/default/salt/reactor/pillar_push_map.yaml'
_PUSH_MAP_CACHE = {'mtime': 0, 'data': None}
def _load_push_map():
try:
st = os.stat(PUSH_MAP_PATH)
except OSError:
LOG.warning('push_pillar: %s not found', PUSH_MAP_PATH)
return {}
if _PUSH_MAP_CACHE['mtime'] != st.st_mtime:
try:
with open(PUSH_MAP_PATH, 'r') as f:
_PUSH_MAP_CACHE['data'] = yaml.safe_load(f) or {}
except Exception:
LOG.exception('push_pillar: failed to load %s', PUSH_MAP_PATH)
_PUSH_MAP_CACHE['data'] = {}
_PUSH_MAP_CACHE['mtime'] = st.st_mtime
return _PUSH_MAP_CACHE['data'] or {}
def _push_enabled():
try:
caller = Caller()
return bool(caller.cmd('pillar.get', 'salt:auto_apply:enabled', True))
except Exception:
LOG.exception('push_pillar: pillar.get salt:auto_apply:enabled failed, assuming enabled')
return True
def _write_intent(key, actions, path):
now = time.time()
try:
os.makedirs(PENDING_DIR, exist_ok=True)
except OSError:
LOG.exception('push_pillar: cannot create %s', PENDING_DIR)
return
intent_path = os.path.join(PENDING_DIR, '{}.json'.format(key))
lock_fd = os.open(LOCK_FILE, os.O_CREAT | os.O_RDWR, 0o644)
try:
fcntl.flock(lock_fd, fcntl.LOCK_EX)
intent = {}
if os.path.exists(intent_path):
try:
with open(intent_path, 'r') as f:
intent = json.load(f)
except (IOError, ValueError):
intent = {}
intent.setdefault('first_touch', now)
intent['last_touch'] = now
intent['actions'] = actions
paths = intent.get('paths', [])
if path and path not in paths:
paths.append(path)
paths = paths[-MAX_PATHS:]
intent['paths'] = paths
tmp_path = intent_path + '.tmp'
with open(tmp_path, 'w') as f:
json.dump(intent, f)
os.rename(tmp_path, intent_path)
except Exception:
LOG.exception('push_pillar: failed to write intent %s', intent_path)
finally:
try:
fcntl.flock(lock_fd, fcntl.LOCK_UN)
finally:
os.close(lock_fd)
def _app_from_setting(setting_id):
# setting_id is e.g. 'telegraf.output' -> 'telegraf', 'ntp.config.servers' -> 'ntp'
if not setting_id:
return None
return setting_id.split('.', 1)[0] or None
def _node_actions(entry, node_id):
# Copy the app's mapped actions but retarget each one to the single node.
# Preserves the state/highstate selection and any batch/batch_wait overrides.
actions = []
for action in entry:
if not isinstance(action, dict):
continue
node_action = dict(action)
node_action['tgt'] = node_id
node_action['tgt_type'] = 'glob'
actions.append(node_action)
return actions
def run():
if not _push_enabled():
LOG.info('push_pillar: push disabled, skipping')
return {}
# The postgres_pillar_beacon nests its payload under data['data']; fall back to the
# top level so the reactor is robust to either shape.
event = data.get('data', data) # noqa: F821 -- data provided by reactor
setting_id = event.get('setting_id', '')
audit_id = event.get('id')
node_id = (event.get('node_id') or '').strip()
app = _app_from_setting(setting_id)
if not app:
LOG.debug('push_pillar: ignoring event with no app segment: setting_id=%s', setting_id)
return {}
push_map = _load_push_map()
entry = push_map.get(app)
if not entry:
LOG.warning(
'push_pillar: app "%s" is not in pillar_push_map.yaml; change will be '
'picked up at the next scheduled highstate (setting_id=%s audit_id=%s)',
app, setting_id, audit_id,
)
return {}
# Branch A: per-node change -> retarget the app's states to just that node.
if node_id:
actions = _node_actions(entry, node_id)
if not actions:
LOG.warning('push_pillar: no usable actions for app "%s" (setting_id=%s)', app, setting_id)
return {}
_write_intent(
'node_{}_{}'.format(node_id, app), actions,
'audit:{}@{}'.format(setting_id, node_id),
)
LOG.info('push_pillar: per-node intent updated for %s on %s (setting_id=%s audit_id=%s)',
app, node_id, setting_id, audit_id)
return {}
# Branch B: grid-wide app change -> use the map entry's actions as-is.
actions = list(entry) # copy to avoid mutating the cache
_write_intent('pillar_{}'.format(app), actions, 'audit:{}'.format(setting_id))
LOG.info('push_pillar: app intent updated for %s (setting_id=%s audit_id=%s)', app, setting_id, audit_id)
return {}