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

namespace: postgres
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'
          labels:
            severity: warning
          annotations:
            description: |-
                Table {{ $labels.relname }} has not been auto vacuumed for 10 days
                  VALUE = {{ $value }}
                  LABELS = {{ $labels }}
            summary: Postgresql table not auto vacuumed (instance {{ $labels.instance }})
        - alert: PostgresqlTableNotAutoAnalyzed
          expr: '(pg_stat_user_tables_last_autoanalyze > 0) and (time() - pg_stat_user_tables_last_autoanalyze) > 24 * 60 * 60 * 10'
          labels:
            severity: warning
          annotations:
            description: |-
                Table {{ $labels.relname }} has not been auto analyzed for 10 days
                  VALUE = {{ $value }}
                  LABELS = {{ $labels }}
            summary: Postgresql table not auto analyzed (instance {{ $labels.instance }})
        - alert: PostgresqlDeadLocks
          expr: 'increase(pg_stat_database_deadlocks{datname!~"template.*|postgres"}[1m]) > 5'
          labels:
            severity: warning
          annotations:
            description: |-
                PostgreSQL has dead-locks
                  VALUE = {{ $value }}
                  LABELS = {{ $labels }}
            summary: Postgresql dead locks (instance {{ $labels.instance }})
        - alert: PostgresqlHighRollbackRate
          expr: 'sum by (namespace, datname) ((rate(pg_stat_database_xact_rollback{datid!="0",datname!~"template.*|postgres"}[3m])) / ((rate(pg_stat_database_xact_rollback{datid!="0",datname!~"template.*|postgres"}[3m])) + (rate(pg_stat_database_xact_commit{datid!="0",datname!~"template.*|postgres"}[3m])))) > 0.02'
          labels:
            severity: warning
          annotations:
            description: |-
                Ratio of transactions being aborted compared to committed is > 2 %
                  VALUE = {{ $value }}
                  LABELS = {{ $labels }}
            summary: Postgresql high rollback rate (instance {{ $labels.instance }})
        - alert: PostgresqlHighRateStatementTimeout
          expr: 'rate(postgresql_errors_total{type="statement_timeout"}[1m]) > 3'
          labels:
            severity: critical
          annotations:
            description: |-
                Postgres transactions showing high rate of statement timeouts
                  VALUE = {{ $value }}
                  LABELS = {{ $labels }}
            summary: Postgresql high rate statement timeout (instance {{ $labels.instance }})
        - alert: PostgresqlHighRateDeadlock
          expr: 'increase(postgresql_errors_total{type="deadlock_detected"}[1m]) > 1'
          labels:
            severity: critical
          annotations:
            description: |-
                Postgres detected deadlocks
                  VALUE = {{ $value }}
                  LABELS = {{ $labels }}
            summary: Postgresql high rate deadlock (instance {{ $labels.instance }})
        - 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:
            description: |-
                PostgreSQL dead tuples is too large
                  VALUE = {{ $value }}
                  LABELS = {{ $labels }}
            summary: Postgresql too many dead tuples (instance {{ $labels.instance }})
        - alert: PostgresqlTooManyLocksAcquired
          expr: '((sum(pg_locks_count)) / (pg_settings_max_locks_per_transaction * pg_settings_max_connections)) > 0.2'
          for: 2m
          labels:
            severity: critical
          annotations:
            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.
                  VALUE = {{ $value }}
                  LABELS = {{ $labels }}
            summary: Postgresql too many locks acquired (instance {{ $labels.instance }})
        - alert: PostgresqlBloatIndexHigh(>80%)
          expr: 'pg_bloat_btree_bloat_pct > 80 and on (idxname) (pg_bloat_btree_real_size > 1e+08)'
          for: 1h
          labels:
            severity: warning
          annotations:
            description: |-
                The index {{ $labels.idxname }} is bloated. You should execute `REINDEX INDEX CONCURRENTLY {{ $labels.idxname }};`
                  VALUE = {{ $value }}
                  LABELS = {{ $labels }}
            summary: Postgresql bloat index high (> 80%) (instance {{ $labels.instance }})
        - alert: PostgresqlBloatTableHigh(>80%)
          expr: 'pg_bloat_table_bloat_pct > 80 and on (relname) (pg_bloat_table_real_size > 2e+08)'
          for: 1h
          labels:
            severity: warning
          annotations:
            description: |-
                The table {{ $labels.relname }} is bloated. You should execute `VACUUM {{ $labels.relname }};`
                  VALUE = {{ $value }}
                  LABELS = {{ $labels }}
            summary: Postgresql bloat table high (> 80%) (instance {{ $labels.instance }})
        - alert: PostgresqlInvalidIndex
          expr: 'pg_genaral_index_info_pg_relation_size{indexrelname=~".*ccnew.*"}'
          for: 6h
          labels:
            severity: warning
          annotations:
            description: |-
                The table {{ $labels.relname }} has an invalid index: {{ $labels.indexrelname }}. You should execute `DROP INDEX {{ $labels.indexrelname }};`
                  VALUE = {{ $value }}
                  LABELS = {{ $labels }}
            summary: Postgresql invalid index (instance {{ $labels.instance }})