mirror of
https://github.com/ClickHouse/ClickHouse.git
synced 2024-12-18 12:22:12 +00:00
38 lines
2.0 KiB
Markdown
38 lines
2.0 KiB
Markdown
# Monitoring
|
|
|
|
You can monitor:
|
|
|
|
- Utilization of hardware resources.
|
|
- ClickHouse server metrics.
|
|
|
|
## Resource Utilization
|
|
|
|
ClickHouse does not monitor the state of hardware resources by itself.
|
|
|
|
It is highly recommended to set up monitoring for:
|
|
|
|
- Load and temperature on processors.
|
|
|
|
You can use [dmesg](https://en.wikipedia.org/wiki/Dmesg), [turbostat](https://www.linux.org/docs/man8/turbostat.html) or other instruments.
|
|
|
|
- Utilization of storage system, RAM and network.
|
|
|
|
## ClickHouse Server Metrics
|
|
|
|
ClickHouse server has embedded instruments for self-state monitoring.
|
|
|
|
To track server events use server logs. See the [logger](server_settings/settings.md#server_settings-logger) section of the configuration file.
|
|
|
|
ClickHouse collects:
|
|
|
|
- Different metrics of how the server uses computational resources.
|
|
- Common statistics on query processing.
|
|
|
|
You can find metrics in the [system.metrics](system_tables.md#system_tables-metrics), [system.events](system_tables.md#system_tables-events), and [system.asynchronous_metrics](system_tables.md#system_tables-asynchronous_metrics) tables.
|
|
|
|
You can configure ClickHouse to export metrics to [Graphite](https://github.com/graphite-project). See the [Graphite section](server_settings/settings.md#server_settings-graphite) in the ClickHouse server configuration file. Before configuring export of metrics, you should set up Graphite by following their official [guide](https://graphite.readthedocs.io/en/latest/install.html).
|
|
|
|
Additionally, you can monitor server availability through the HTTP API. Send the `HTTP GET` request to `/`. If the server is available, it responds with `200 OK`.
|
|
|
|
To monitor servers in a cluster configuration, you should set the [max_replica_delay_for_distributed_queries](settings/settings.md#settings-max_replica_delay_for_distributed_queries) parameter and use the HTTP resource `/replicas_status`. A request to `/replicas_status` returns `200 OK` if the replica is available and is not delayed behind the other replicas. If a replica is delayed, it returns information about the gap.
|