Improve monitoring for prod
This commit is contained in:
Urtzi Alfaro
@@ -0,0 +1,306 @@
|
||||
---
|
||||
apiVersion: apps/v1
|
||||
kind: Deployment
|
||||
metadata:
|
||||
name: postgres-exporter
|
||||
namespace: monitoring
|
||||
labels:
|
||||
app: postgres-exporter
|
||||
spec:
|
||||
replicas: 1
|
||||
selector:
|
||||
matchLabels:
|
||||
app: postgres-exporter
|
||||
template:
|
||||
metadata:
|
||||
labels:
|
||||
app: postgres-exporter
|
||||
spec:
|
||||
containers:
|
||||
- name: postgres-exporter
|
||||
image: prometheuscommunity/postgres-exporter:v0.15.0
|
||||
ports:
|
||||
- containerPort: 9187
|
||||
name: metrics
|
||||
env:
|
||||
- name: DATA_SOURCE_NAME
|
||||
valueFrom:
|
||||
secretKeyRef:
|
||||
name: postgres-exporter
|
||||
key: data-source-name
|
||||
# Enable extended metrics
|
||||
- name: PG_EXPORTER_EXTEND_QUERY_PATH
|
||||
value: "/etc/postgres-exporter/queries.yaml"
|
||||
# Disable default metrics (we'll use custom ones)
|
||||
- name: PG_EXPORTER_DISABLE_DEFAULT_METRICS
|
||||
value: "false"
|
||||
# Disable settings metrics (can be noisy)
|
||||
- name: PG_EXPORTER_DISABLE_SETTINGS_METRICS
|
||||
value: "false"
|
||||
volumeMounts:
|
||||
- name: queries
|
||||
mountPath: /etc/postgres-exporter
|
||||
resources:
|
||||
requests:
|
||||
memory: "64Mi"
|
||||
cpu: "50m"
|
||||
limits:
|
||||
memory: "128Mi"
|
||||
cpu: "200m"
|
||||
livenessProbe:
|
||||
httpGet:
|
||||
path: /
|
||||
port: 9187
|
||||
initialDelaySeconds: 30
|
||||
periodSeconds: 10
|
||||
readinessProbe:
|
||||
httpGet:
|
||||
path: /
|
||||
port: 9187
|
||||
initialDelaySeconds: 5
|
||||
periodSeconds: 5
|
||||
volumes:
|
||||
- name: queries
|
||||
configMap:
|
||||
name: postgres-exporter-queries
|
||||
|
||||
---
|
||||
apiVersion: v1
|
||||
kind: ConfigMap
|
||||
metadata:
|
||||
name: postgres-exporter-queries
|
||||
namespace: monitoring
|
||||
data:
|
||||
queries.yaml: |
|
||||
# Custom PostgreSQL queries for bakery-ia metrics
|
||||
|
||||
pg_database:
|
||||
query: |
|
||||
SELECT
|
||||
datname,
|
||||
numbackends as connections,
|
||||
xact_commit as transactions_committed,
|
||||
xact_rollback as transactions_rolled_back,
|
||||
blks_read as blocks_read,
|
||||
blks_hit as blocks_hit,
|
||||
tup_returned as tuples_returned,
|
||||
tup_fetched as tuples_fetched,
|
||||
tup_inserted as tuples_inserted,
|
||||
tup_updated as tuples_updated,
|
||||
tup_deleted as tuples_deleted,
|
||||
conflicts as conflicts,
|
||||
temp_files as temp_files,
|
||||
temp_bytes as temp_bytes,
|
||||
deadlocks as deadlocks
|
||||
FROM pg_stat_database
|
||||
WHERE datname NOT IN ('template0', 'template1', 'postgres')
|
||||
metrics:
|
||||
- datname:
|
||||
usage: "LABEL"
|
||||
description: "Name of the database"
|
||||
- connections:
|
||||
usage: "GAUGE"
|
||||
description: "Number of backends currently connected to this database"
|
||||
- transactions_committed:
|
||||
usage: "COUNTER"
|
||||
description: "Number of transactions in this database that have been committed"
|
||||
- transactions_rolled_back:
|
||||
usage: "COUNTER"
|
||||
description: "Number of transactions in this database that have been rolled back"
|
||||
- blocks_read:
|
||||
usage: "COUNTER"
|
||||
description: "Number of disk blocks read in this database"
|
||||
- blocks_hit:
|
||||
usage: "COUNTER"
|
||||
description: "Number of times disk blocks were found in the buffer cache"
|
||||
- tuples_returned:
|
||||
usage: "COUNTER"
|
||||
description: "Number of rows returned by queries in this database"
|
||||
- tuples_fetched:
|
||||
usage: "COUNTER"
|
||||
description: "Number of rows fetched by queries in this database"
|
||||
- tuples_inserted:
|
||||
usage: "COUNTER"
|
||||
description: "Number of rows inserted by queries in this database"
|
||||
- tuples_updated:
|
||||
usage: "COUNTER"
|
||||
description: "Number of rows updated by queries in this database"
|
||||
- tuples_deleted:
|
||||
usage: "COUNTER"
|
||||
description: "Number of rows deleted by queries in this database"
|
||||
- conflicts:
|
||||
usage: "COUNTER"
|
||||
description: "Number of queries canceled due to conflicts with recovery"
|
||||
- temp_files:
|
||||
usage: "COUNTER"
|
||||
description: "Number of temporary files created by queries"
|
||||
- temp_bytes:
|
||||
usage: "COUNTER"
|
||||
description: "Total amount of data written to temporary files by queries"
|
||||
- deadlocks:
|
||||
usage: "COUNTER"
|
||||
description: "Number of deadlocks detected in this database"
|
||||
|
||||
pg_replication:
|
||||
query: |
|
||||
SELECT
|
||||
CASE WHEN pg_is_in_recovery() THEN 1 ELSE 0 END as is_replica,
|
||||
EXTRACT(EPOCH FROM (now() - pg_last_xact_replay_timestamp()))::INT as lag_seconds
|
||||
metrics:
|
||||
- is_replica:
|
||||
usage: "GAUGE"
|
||||
description: "1 if this is a replica, 0 if primary"
|
||||
- lag_seconds:
|
||||
usage: "GAUGE"
|
||||
description: "Replication lag in seconds (only on replicas)"
|
||||
|
||||
pg_slow_queries:
|
||||
query: |
|
||||
SELECT
|
||||
datname,
|
||||
usename,
|
||||
state,
|
||||
COUNT(*) as count,
|
||||
MAX(EXTRACT(EPOCH FROM (now() - query_start))) as max_duration_seconds
|
||||
FROM pg_stat_activity
|
||||
WHERE state != 'idle'
|
||||
AND query NOT LIKE '%pg_stat_activity%'
|
||||
AND query_start < now() - interval '30 seconds'
|
||||
GROUP BY datname, usename, state
|
||||
metrics:
|
||||
- datname:
|
||||
usage: "LABEL"
|
||||
description: "Database name"
|
||||
- usename:
|
||||
usage: "LABEL"
|
||||
description: "User name"
|
||||
- state:
|
||||
usage: "LABEL"
|
||||
description: "Query state"
|
||||
- count:
|
||||
usage: "GAUGE"
|
||||
description: "Number of slow queries"
|
||||
- max_duration_seconds:
|
||||
usage: "GAUGE"
|
||||
description: "Maximum query duration in seconds"
|
||||
|
||||
pg_table_stats:
|
||||
query: |
|
||||
SELECT
|
||||
schemaname,
|
||||
relname,
|
||||
seq_scan,
|
||||
seq_tup_read,
|
||||
idx_scan,
|
||||
idx_tup_fetch,
|
||||
n_tup_ins,
|
||||
n_tup_upd,
|
||||
n_tup_del,
|
||||
n_tup_hot_upd,
|
||||
n_live_tup,
|
||||
n_dead_tup,
|
||||
n_mod_since_analyze,
|
||||
last_vacuum,
|
||||
last_autovacuum,
|
||||
last_analyze,
|
||||
last_autoanalyze
|
||||
FROM pg_stat_user_tables
|
||||
WHERE schemaname = 'public'
|
||||
ORDER BY n_live_tup DESC
|
||||
LIMIT 20
|
||||
metrics:
|
||||
- schemaname:
|
||||
usage: "LABEL"
|
||||
description: "Schema name"
|
||||
- relname:
|
||||
usage: "LABEL"
|
||||
description: "Table name"
|
||||
- seq_scan:
|
||||
usage: "COUNTER"
|
||||
description: "Number of sequential scans"
|
||||
- seq_tup_read:
|
||||
usage: "COUNTER"
|
||||
description: "Number of tuples read by sequential scans"
|
||||
- idx_scan:
|
||||
usage: "COUNTER"
|
||||
description: "Number of index scans"
|
||||
- idx_tup_fetch:
|
||||
usage: "COUNTER"
|
||||
description: "Number of tuples fetched by index scans"
|
||||
- n_tup_ins:
|
||||
usage: "COUNTER"
|
||||
description: "Number of tuples inserted"
|
||||
- n_tup_upd:
|
||||
usage: "COUNTER"
|
||||
description: "Number of tuples updated"
|
||||
- n_tup_del:
|
||||
usage: "COUNTER"
|
||||
description: "Number of tuples deleted"
|
||||
- n_tup_hot_upd:
|
||||
usage: "COUNTER"
|
||||
description: "Number of tuples HOT updated"
|
||||
- n_live_tup:
|
||||
usage: "GAUGE"
|
||||
description: "Estimated number of live rows"
|
||||
- n_dead_tup:
|
||||
usage: "GAUGE"
|
||||
description: "Estimated number of dead rows"
|
||||
- n_mod_since_analyze:
|
||||
usage: "GAUGE"
|
||||
description: "Number of rows modified since last analyze"
|
||||
|
||||
pg_locks:
|
||||
query: |
|
||||
SELECT
|
||||
mode,
|
||||
locktype,
|
||||
COUNT(*) as count
|
||||
FROM pg_locks
|
||||
GROUP BY mode, locktype
|
||||
metrics:
|
||||
- mode:
|
||||
usage: "LABEL"
|
||||
description: "Lock mode"
|
||||
- locktype:
|
||||
usage: "LABEL"
|
||||
description: "Lock type"
|
||||
- count:
|
||||
usage: "GAUGE"
|
||||
description: "Number of locks"
|
||||
|
||||
pg_connection_pool:
|
||||
query: |
|
||||
SELECT
|
||||
state,
|
||||
COUNT(*) as count,
|
||||
MAX(EXTRACT(EPOCH FROM (now() - state_change))) as max_state_duration_seconds
|
||||
FROM pg_stat_activity
|
||||
GROUP BY state
|
||||
metrics:
|
||||
- state:
|
||||
usage: "LABEL"
|
||||
description: "Connection state"
|
||||
- count:
|
||||
usage: "GAUGE"
|
||||
description: "Number of connections in this state"
|
||||
- max_state_duration_seconds:
|
||||
usage: "GAUGE"
|
||||
description: "Maximum time a connection has been in this state"
|
||||
|
||||
---
|
||||
apiVersion: v1
|
||||
kind: Service
|
||||
metadata:
|
||||
name: postgres-exporter
|
||||
namespace: monitoring
|
||||
labels:
|
||||
app: postgres-exporter
|
||||
spec:
|
||||
type: ClusterIP
|
||||
ports:
|
||||
- port: 9187
|
||||
targetPort: 9187
|
||||
protocol: TCP
|
||||
name: metrics
|
||||
selector:
|
||||
app: postgres-exporter
|
||||
Reference in New Issue
Block a user