infra/services/monitoring/lgtm/alerts/postgres.yml

groups:
- name: PostgreSQL
  rules:

    - alert: PostgresqlTableNotAutoVacuumed
      expr: '(pg_stat_user_tables_last_autovacuum > 0) and (time() - pg_stat_user_tables_last_autovacuum) > 60 * 60 * 24 * 10'
      for: 0m
      labels:
        severity: warning
      annotations:
        summary: Postgresql table not auto vacuumed (instance {{ $labels.instance }})
        description: "Table {{ $labels.relname }} has not been auto vacuumed for 10 days\n  VALUE = {{ $value }}\n  LABELS = {{ $labels }}"

    - alert: PostgresqlTableNotAutoAnalyzed
      expr: '(pg_stat_user_tables_last_autoanalyze > 0) and (time() - pg_stat_user_tables_last_autoanalyze) > 24 * 60 * 60 * 10'
      for: 0m
      labels:
        severity: warning
      annotations:
        summary: Postgresql table not auto analyzed (instance {{ $labels.instance }})
        description: "Table {{ $labels.relname }} has not been auto analyzed for 10 days\n  VALUE = {{ $value }}\n  LABELS = {{ $labels }}"

    - alert: PostgresqlDeadLocks
      expr: 'increase(pg_stat_database_deadlocks{datname!~"template.*|postgres"}[1m]) > 5'
      for: 0m
      labels:
        severity: warning
      annotations:
        summary: Postgresql dead locks (instance {{ $labels.instance }})
        description: "PostgreSQL has dead-locks\n  VALUE = {{ $value }}\n  LABELS = {{ $labels }}"

    - alert: PostgresqlHighRollbackRate
      expr: 'sum by (namespace,datname) ((rate(pg_stat_database_xact_rollback{datname!~"template.*|postgres",datid!="0"}[3m])) / ((rate(pg_stat_database_xact_rollback{datname!~"template.*|postgres",datid!="0"}[3m])) + (rate(pg_stat_database_xact_commit{datname!~"template.*|postgres",datid!="0"}[3m])))) > 0.02'
      for: 0m
      labels:
        severity: warning
      annotations:
        summary: Postgresql high rollback rate (instance {{ $labels.instance }})
        description: "Ratio of transactions being aborted compared to committed is > 2 %\n  VALUE = {{ $value }}\n  LABELS = {{ $labels }}"

    - alert: PostgresqlHighRateStatementTimeout
      expr: 'rate(postgresql_errors_total{type="statement_timeout"}[1m]) > 3'
      for: 0m
      labels:
        severity: critical
      annotations:
        summary: Postgresql high rate statement timeout (instance {{ $labels.instance }})
        description: "Postgres transactions showing high rate of statement timeouts\n  VALUE = {{ $value }}\n  LABELS = {{ $labels }}"

    - alert: PostgresqlHighRateDeadlock
      expr: 'increase(postgresql_errors_total{type="deadlock_detected"}[1m]) > 1'
      for: 0m
      labels:
        severity: critical
      annotations:
        summary: Postgresql high rate deadlock (instance {{ $labels.instance }})
        description: "Postgres detected deadlocks\n  VALUE = {{ $value }}\n  LABELS = {{ $labels }}"

    - alert: PostgresqlTooManyDeadTuples
      expr: '((pg_stat_user_tables_n_dead_tup > 10000) / (pg_stat_user_tables_n_live_tup + pg_stat_user_tables_n_dead_tup)) >= 0.1'
      for: 2m
      labels:
        severity: warning
      annotations:
        summary: Postgresql too many dead tuples (instance {{ $labels.instance }})
        description: "PostgreSQL dead tuples is too large\n  VALUE = {{ $value }}\n  LABELS = {{ $labels }}"

    - alert: PostgresqlTooManyLocksAcquired
      expr: '((sum (pg_locks_count)) / (pg_settings_max_locks_per_transaction * pg_settings_max_connections)) > 0.20'
      for: 2m
      labels:
        severity: critical
      annotations:
        summary: Postgresql too many locks acquired (instance {{ $labels.instance }})
        description: "Too many locks acquired on the database. If this alert happens frequently, we may need to increase the postgres setting max_locks_per_transaction.\n  VALUE = {{ $value }}\n  LABELS = {{ $labels }}"

    - alert: PostgresqlBloatIndexHigh(>80%)
      expr: 'pg_bloat_btree_bloat_pct > 80 and on (idxname) (pg_bloat_btree_real_size > 100000000)'
      for: 1h
      labels:
        severity: warning
      annotations:
        summary: Postgresql bloat index high (> 80%) (instance {{ $labels.instance }})
        description: "The index {{ $labels.idxname }} is bloated. You should execute `REINDEX INDEX CONCURRENTLY {{ $labels.idxname }};`\n  VALUE = {{ $value }}\n  LABELS = {{ $labels }}"

    - alert: PostgresqlBloatTableHigh(>80%)
      expr: 'pg_bloat_table_bloat_pct > 80 and on (relname) (pg_bloat_table_real_size > 200000000)'
      for: 1h
      labels:
        severity: warning
      annotations:
        summary: Postgresql bloat table high (> 80%) (instance {{ $labels.instance }})
        description: "The table {{ $labels.relname }} is bloated. You should execute `VACUUM {{ $labels.relname }};`\n  VALUE = {{ $value }}\n  LABELS = {{ $labels }}"

    - alert: PostgresqlInvalidIndex
      expr: 'pg_genaral_index_info_pg_relation_size{indexrelname=~".*ccnew.*"}'
      for: 6h
      labels:
        severity: warning
      annotations:
        summary: Postgresql invalid index (instance {{ $labels.instance }})
        description: "The table {{ $labels.relname }} has an invalid index: {{ $labels.indexrelname }}. You should execute `DROP INDEX {{ $labels.indexrelname }};`\n  VALUE = {{ $value }}\n  LABELS = {{ $labels }}"