ClickHouse Managed Postgres metrics reference

This page lists every metric exposed by the ClickHouse Managed Postgres Prometheus endpoint. For setup and authentication, see the Prometheus endpoint page.

Common labels

Every metric carries the following labels:

Label	Description
`clickhouse_org`	Organization ID
`postgres_service`	Postgres service ID
`postgres_service_name`	Postgres service name

Some metrics add a label for the dimension they break out (e.g. mode on CPU metrics, state on connections, database on database size, wait on wait events). Those are listed alongside each metric.

Information metric

PostgresServiceInfo is a gauge that always equals 1 and carries the service’s current status and version on its labels. Use it to join status onto other metrics or to alert on a service leaving the running state.

Metric	Type	Extra labels	Description
`PostgresServiceInfo`	gauge	`postgres_status`, `postgres_version`	One series per service; value is always `1`.

postgres_status reports the current lifecycle state of the service (for example running, creating, stopped). postgres_version reports the major Postgres version (e.g. 17, 18).

Capacity

Static limits provisioned for the service. These change only when the service is resized.

Metric	Type	Unit	Description
`PostgresServer_CPUCores`	gauge	cores	CPU cores allocated to the service.
`PostgresServer_MemoryLimitBytes`	gauge	bytes	Memory allocated to the service.
`PostgresServer_StorageLimitBytes`	gauge	bytes	Storage allocated to the service.

Resource utilization

Metric	Type	Extra labels	Description
`PostgresServer_CPUSeconds_Total`	counter	`mode`	CPU time consumed, broken out by mode: `user`, `system`, `iowait`, `softirq`, `steal`, `irq`, `nice`, `idle`.
`PostgresServer_MemoryUsedPercent`	gauge		Memory in use, as a percentage of `PostgresServer_MemoryLimitBytes`.
`PostgresServer_MemoryCachePercent`	gauge		Memory used by cache and buffers, as a percentage of total memory.
`PostgresServer_FilesystemUsedPercent`	gauge		Filesystem space used, as a percentage of total storage.

To compute CPU usage as a percentage, take the rate of PostgresServer_CPUSeconds_Total over the modes you care about and divide by PostgresServer_CPUCores.

Disk and network I/O

Metric	Type	Unit	Description
`PostgresServer_DiskReads_Total`	counter	ops	Disk read operations completed.
`PostgresServer_DiskWrites_Total`	counter	ops	Disk write operations completed.
`PostgresServer_NetworkReceiveBytes_Total`	counter	bytes	Bytes received over the network.
`PostgresServer_NetworkTransmitBytes_Total`	counter	bytes	Bytes transmitted over the network.

Database activity

Cumulative counters since service start. Use rate() or irate() to turn them into per-second values.

Metric	Type	Description
`PostgresServer_TuplesFetched_Total`	counter	Rows fetched by queries.
`PostgresServer_TuplesInserted_Total`	counter	Rows inserted.
`PostgresServer_TuplesUpdated_Total`	counter	Rows updated.
`PostgresServer_TuplesDeleted_Total`	counter	Rows deleted.
`PostgresServer_TransactionsCommitted_Total`	counter	Transactions committed.
`PostgresServer_TransactionsRolledBack_Total`	counter	Transactions rolled back.
`PostgresServer_Deadlocks_Total`	counter	Deadlocks detected.

Query performance

Counters from recorded Query Insights events, excluding UTILITY operations.

Metric	Type	Unit	Description
`PostgresQuery_Calls_Total`	counter	events	Recorded query events.
`PostgresQuery_DurationMicroseconds_Total`	counter	microseconds	Total recorded query duration.
`PostgresQuery_ErrorEvents_Total`	counter	events	Recorded events with nonzero error severity, including warnings.
`PostgresQuery_FailureEvents_Total`	counter	events	Recorded events with severity `ERROR` or higher.
`PostgresQuery_Rows_Total`	counter	rows	Rows returned or affected by recorded queries.
`PostgresQuery_SharedBlocksHit_Total`	counter	blocks	Shared blocks hit by recorded queries.
`PostgresQuery_SharedBlocksRead_Total`	counter	blocks	Shared blocks read by recorded queries.
`PostgresQuery_WALBytes_Total`	counter	bytes	WAL bytes generated by recorded queries.
`PostgresQuery_SelectCalls_Total`	counter	events	Recorded `SELECT` events.
`PostgresQuery_InsertCalls_Total`	counter	events	Recorded `INSERT` events.
`PostgresQuery_UpdateCalls_Total`	counter	events	Recorded `UPDATE` events.
`PostgresQuery_DeleteCalls_Total`	counter	events	Recorded `DELETE` events.
`PostgresQuery_MergeCalls_Total`	counter	events	Recorded `MERGE` events.
`PostgresQuery_OtherCalls_Total`	counter	events	Recorded events outside `SELECT`, `INSERT`, `UPDATE`, `DELETE`, and `MERGE`.

Connections, cache, and database size

Metric	Type	Extra labels	Description
`PostgresServer_ActiveConnections`	gauge	`state`	Connections by state (e.g. `active`, `idle`).
`PostgresServer_CacheHitRatio`	gauge		Buffer cache hit ratio: blocks served from cache vs. total blocks accessed, as a percentage.
`PostgresServer_DatabaseSizeBytes`	gauge	`database`	Disk size of each database, in bytes. Includes the `postgres` default and any user-created databases.

Wait events

Gauges from the wait samples that ClickHouse Managed Postgres takes on the primary every 30 seconds. These are the same samples behind the wait charts in Query insights. A series appears only while there’s something to report: an idle service has no wait series, and PostgresServer_LockWaitMaxSeconds appears only while a session is blocked on a lock.

Metric	Type	Extra labels	Description
`PostgresServer_ActiveSessionsByWait`	gauge	`wait`	Active sessions in each wait event, averaged over the recent samples that reported it. `wait` is the wait event as `type:event`, for example `Lock:relation` or `IO:WalSync`, or `None` for sessions running with no wait event.
`PostgresServer_LockWaitMaxSeconds`	gauge		Longest time a session has been blocked on a lock, in seconds.

To count the sessions blocked on locks, sum the Lock wait events:

sum by (postgres_service_name) (PostgresServer_ActiveSessionsByWait{wait=~"Lock:.*"})

Prometheus endpoint — setup, authentication, and scraping
Dashboard — built-in cloud console charts
Query insights — wait charts and the Waits table in the cloud console
OpenAPI guide — API key creation and lookup of organization and service IDs

Get started

Features

Guides

Pricing

ClickHouse Integration

Migrations

Monitoring

FAQ

ClickHouse Managed Postgres metrics reference

Common labels

Information metric

Capacity

Resource utilization

Disk and network I/O

Database activity

Query performance

Connections, cache, and database size

Wait events

​Common labels

​Information metric

​Capacity

​Resource utilization

​Disk and network I/O

​Database activity

​Query performance

​Connections, cache, and database size

​Wait events

​Related pages

Common labels

Information metric

Capacity

Resource utilization

Disk and network I/O

Database activity

Query performance

Connections, cache, and database size

Wait events

Related pages