diff options
Diffstat (limited to 'doc/administration/monitoring/prometheus')
-rw-r--r-- | doc/administration/monitoring/prometheus/gitlab_exporter.md | 24 | ||||
-rw-r--r-- | doc/administration/monitoring/prometheus/gitlab_metrics.md | 1 |
2 files changed, 25 insertions, 0 deletions
diff --git a/doc/administration/monitoring/prometheus/gitlab_exporter.md b/doc/administration/monitoring/prometheus/gitlab_exporter.md index 8e4d87cfa78..5add842bccc 100644 --- a/doc/administration/monitoring/prometheus/gitlab_exporter.md +++ b/doc/administration/monitoring/prometheus/gitlab_exporter.md @@ -30,3 +30,27 @@ To enable the GitLab exporter in an Omnibus GitLab instance: Prometheus automatically begins collecting performance data from the GitLab exporter exposed at `localhost:9168`. + +## Use a different Rack server + +>- Introduced in [Omnibus GitLab 13.8](https://gitlab.com/gitlab-org/omnibus-gitlab/-/merge_requests/4896). +>- WEBrick is now the default Rack server instead of Puma. + +By default, the GitLab exporter runs on [WEBrick](https://github.com/ruby/webrick), a single-threaded Ruby web server. +You can choose a different Rack server that better matches your performance needs. +For instance, in multi-node setups that contain a large number of Prometheus scrapers +but only a few monitoring nodes, you may decide to run a multi-threaded server such as Puma instead. + +To change the Rack server to Puma: + +1. Edit `/etc/gitlab/gitlab.rb`. +1. Add, or find and uncomment, the following line, and set it to `puma`: + + ```ruby + gitlab_exporter['server_name'] = 'puma' + ``` + +1. Save the file and [reconfigure GitLab](../../restart_gitlab.md#omnibus-gitlab-reconfigure) + for the changes to take effect. + +The supported Rack servers are `webrick` and `puma`. diff --git a/doc/administration/monitoring/prometheus/gitlab_metrics.md b/doc/administration/monitoring/prometheus/gitlab_metrics.md index c5e0570515d..4493c1677bc 100644 --- a/doc/administration/monitoring/prometheus/gitlab_metrics.md +++ b/doc/administration/monitoring/prometheus/gitlab_metrics.md @@ -210,6 +210,7 @@ configuration option in `gitlab.yml`. These metrics are served from the | `limited_capacity_worker_max_running_jobs` | Gauge | 13.5 | Maximum number of running jobs | `worker` | | `limited_capacity_worker_remaining_work_count` | Gauge | 13.5 | Number of jobs waiting to be enqueued | `worker` | | `destroyed_job_artifacts_count_total` | Counter | 13.6 | Number of destroyed expired job artifacts | | +| `destroyed_pipeline_artifacts_count_total` | Counter | 13.8 | Number of destroyed expired pipeline artifacts | | ## Database load balancing metrics **(PREMIUM ONLY)** |