News/2023 Cloud VPS metrics changes
Jump to navigation Jump to search
This page is currently a draft.
More information and discussion about changes to this draft on the talk page.
This page details information about the changes to Cloud VPS metrics systems in early 2023.
What is changing?
Relevant changes that happened previously
- Debian 11 (released mid 2021) stopped including Diamond, which produced machine-level stats for Graphite.
- From November 2022, a Prometheus instance (at prometheus.wmcloud.org) has been collecting machine-level statistics using prometheus-node-exporter on all VMs.
What is changing now?
- Date TBD: The remaining Cloud VPS instances will have Diamond removed.
- Date TBD: The Cloud VPS Graphite server will be shut down.
- Date TBD: The Grafana server at grafana-cloud.wikimedia.org and grafana-labs.wikimedia.org will be shut down and replaced by a new server running on grafana.wmcloud.org.
What should I do?
Migrate dashboards (and other tools) to use Prometheus data
You should migrate any affected Grafana dashboards and other tools you find useful to load data from Prometheus instead of loading it from Graphite.
If your project needs Graphite for a reason other than per-machine statistics and can't move to Prometheus, you need to run your own Graphite instance instead of relying on a WMCS-hosted instance.
Migrate dashboards from grafana-cloud.wikimedia.org to grafana.wmcloud.org
You should move any Grafana dashboards you find useful on grafana-cloud.wikimedia.org to grafana.wmcloud.org.
Why are we doing this?
- We are replacing Diamond metrics with Prometheus based ones because Diamond uses Python 2, which is no longer supported on latest versions of the Debian operating system.
- We are shutting down the Graphite server because we don't find it very useful after Diamond will be removed.
- We are replacing the old Grafana instance because new versions of Grafana no longer support the setup we currently have.