Monitoring with Datadog

[freshfly63]

We’d like to monitor our Metabase instance. Ideal metrics could include:

• SQL query duration
• SQL result row count
• HTTP response code count
• HTTP response size in bytes
• etc.

Can we get these metrics from Metabase?

Ideally we’ll send these to Datadog (where we monitor everything else). So if they are available through a Prometheus endpoint, or a statsd client, or exported via JMX, then that would work.

Thank you!

[freshfly63]

After reading core.clj and middleware.clj, looks like our best bet might be parsing the Metabase logs:

• https://github.com/metabase/metabase/blob/master/src/metabase/middleware.clj#L339
• https://github.com/metabase/metabase/blob/master/src/metabase/core.clj#L87
• https://github.com/metabase/metabase/blob/master/src/metabase/core.clj#L202

I don’t see any existing metrics code, or any way to enable metabase or embedded jetty metrics with an env var.

We’d prefer to not parse the logs because the format may change in future versions and it wasn’t intended for machine consumption.

Please let us know if there’s another way!

[freshfly63]

Since Metabase doesn’t export metrics, our plan now is…

• Use haproxy or traefik in front of metabase, then use datadog’s haproxy integration to get stats on web requests
• Use datadog’s postgresql integration to get query stats directly from the database
• Not to parse logs