atbore-phx / pgpool2_exporter

Prometheus exporter for Pgpool-II.

This is a fork of pgpool/pgpool2_exporter, which utilizes gcloud SQL and the pg_stat_replication table to retrieve replication metrics.

When pgpool is deployed in a cloud environment like gcloud (as implemented in this fork), replicas are often scaled up or down. If a node remains unused or goes down, the metric pool_nodes will hang, potentially causing the Prometheus scraping timeout to be reached.

The original version defines pool_nodes as follows:

		"pool_nodes": {
			"hostname":          {LABEL, "Backend hostname"},
			"port":              {LABEL, "Backend port"},
			"role":              {LABEL, "Role (primary or standby)"},
			"status":            {GAUGE, "Backend node Status (1 for up or waiting, 0 for down or unused)"},
			"select_cnt":        {COUNTER, "SELECT statement counts issued to each backend"},
			"replication_delay": {GAUGE, "Replication delay"},
    }

pool_nodes is replaced as follows:

• hostname, port, role, status, select_cnt are considered redundant since they are already present in another metric called pool_backend_stats.

• replication_delay is substituted with three new metrics, namely: flush_lag, write_lag, replay_lag, which are obtained by inspecting the pg_stat_replication table on the primary PostgreSQL instance.

  the new pool_backend_stats:

"pool_backend_stats": {
			"hostname":   {LABEL, "Backend hostname"},
			"port":       {LABEL, "Backend port"},
			"role":       {LABEL, "Role (primary or standby)"},
			"status":     {GAUGE, "Backend node Status (1 for up or waiting, 0 for down or unused)"},
			"select_cnt": {COUNTER, "SELECT statement counts issued to each backend"},
			"insert_cnt": {COUNTER, "INSERT statement counts issued to each backend"},
			"update_cnt": {COUNTER, "UPDATE statement counts issued to each backend"},
			"delete_cnt": {COUNTER, "DELETE statement counts issued to each backend"},
			"ddl_cnt":    {COUNTER, "DDL statement counts issued to each backend"},
			"other_cnt":  {COUNTER, "other statement counts issued to each backend"},
			"panic_cnt":  {COUNTER, "Panic message counts returned from backend"},
			"fatal_cnt":  {COUNTER, "Fatal message counts returned from backend)"},
			"error_cnt":  {COUNTER, "Error message counts returned from backend"},
			"flush_lag":  {GAUGE, "flush delay"},
			"write_lag":  {GAUGE, "write delay"},
			"replay_lag": {GAUGE, "reply delay"},
		},

Pgpool-II Exporter

Prometheus exporter for Pgpool-II metrics.

Supported Pgpool-II 3.6 and later.

Building and running

Build

$ make

Running

Running using environment variables:

$ export DATA_SOURCE_NAME="postgresql://<user>:<password>@<hostname>:<port>/<dbname>?sslmode=<sslmode>"
OR
# If your password contains special characters, use the environment variables below:
export DATA_SOURCE_USER="<user>"
export DATA_SOURCE_PASS="<password>"
export DATA_SOURCE_URI="<hostname>:<port>/<dbname>?sslmode=<sslmode>"

$ ./pgpool2_exporter <flags>

To see all available configuration flags:

$ ./pgpool2_exporter --help

Flags

• help Show context-sensitive help (also try --help-long and --help-man).

• version Print version information.

• web.listen-address Address on which to expose metrics and web interface. (default ":9719").

• web.telemetry-path Path under which to expose metrics. (default "/metrics")

• log.level Set logging level: one of debug, info, warn, error.

• log.format Set the log format: one of logfmt, json.

Docker

This package is available for Docker. The following environment variables configure the docker container:

• POSTGRES_USERNAME PostgreSQL user name. Default is postgres.

• POSTGRES_PASSWORD PostgreSQL user password. Default is postgres.

• POSTGRES_DATABASE Database name. Default is postgres.

• PGPOOL_SERVICE Pgpool-II hostname. Default is localhost.

• PGPOOL_SERVICE_PORT Pgpool-II port number. Default is 9999.

• SSLMODE Whether or not to use SSL. Default is disable. Valid values: disable, require, verify-ca, verify-full.

docker run --name pgpool2_exporter \
  --net=host --rm \
  -e POSTGRES_USERNAME=<username> \
  -e POSTGRES_PASSWORD=<password> \
  -e POSTGRES_DATABASE=<database> \
  -e PGPOOL_SERVICE=<hostname> \
  -e PGPOOL_SERVICE_PORT=<port> \
  -e SSLMODE=<sslmode> \
  pgpool/pgpool2_exporter:latest

Metrics

name	Pgpool-II Version	Description
pgpool2_frontend_total	3.6+	Number of total child processes
pgpool2_frontend_used	3.6+	Number of used child processes
pgpool2_frontend_used_ratio	3.6+	Ratio of used child processes to total child processes (0.0 to 1.0)
pgpool2_pool_nodes_status	3.6+	Backend node Status (1 for up or waiting, 0 for down or unused)
pgpool2_pool_nodes_replication_delay	3.6+	Replication delay
pgpool2_pool_nodes_select_cnt	3.6+	SELECT query counts issued to each backend
pgpool2_pool_cache_cache_hit_ratio	3.6+	Query cache hit ratio
pgpool2_pool_cache_num_cache_entries	3.6+	Number of used cache entries
pgpool2_pool_cache_num_hash_entries	3.6+	Number of total hash entries
pgpool2_pool_cache_used_hash_entries	3.6+	Number of used hash entries
pgpool2_pool_backend_stats_select_cnt	4.2+	SELECT statement counts issued to each backend
pgpool2_pool_backend_stats_insert_cnt	4.2+	INSERT statement counts issued to each backend
pgpool2_pool_backend_stats_update_cnt	4.2+	UPDATE statement counts issued to each backend
pgpool2_pool_backend_stats_delete_cnt	4.2+	DELETE statement counts issued to each backend
pgpool2_pool_backend_stats_ddl_cnt	4.2+	DDL statement counts issued to each backend
pgpool2_pool_backend_stats_other_cnt	4.2+	other statement counts issued to each backend
pgpool2_pool_backend_stats_panic_cnt	4.2+	Panic message counts returned from backend
pgpool2_pool_backend_stats_fatal_cnt	4.2+	Fatal message counts returned from backend
pgpool2_pool_backend_stats_error_cnt	4.2+	Error message counts returned from backend
pgpool2_pool_health_check_stats_total_count	4.2+	Number of health check count in total
pgpool2_pool_health_check_stats_success_count	4.2+	Number of successful health check count in total
pgpool2_pool_health_check_stats_fail_count	4.2+	Number of failed health check count in total
pgpool2_pool_health_check_stats_skip_count	4.2+	Number of skipped health check count in total
pgpool2_pool_health_check_stats_retry_count	4.2+	Number of retried health check count in total
pgpool2_pool_health_check_stats_average_retry_count	4.2+	Number of average retried health check count in a health check session
pgpool2_pool_health_check_stats_max_retry_count	4.2+	Number of maximum retried health check count in a health check session
pgpool2_pool_health_check_stats_max_duration	4.2+	Maximum health check duration in Millie seconds
pgpool2_pool_health_check_stats_min_duration	4.2+	Minimum health check duration in Millie seconds
pgpool2_pool_health_check_stats_average_duration	4.2+	Average health check duration in Millie seconds