Observability and Metrics

[C-Loftus]

For those who are using qlever in production or under load, I was curious what others are using for observability and performance metrics.

For instance, I was curious if anyone has set up a prometheus exporter or has found an open telemetry configuration which works well for tracing client queries.

From searches in discussions (#1744) it appears there is:

Not much available yet. We're considering adding tracing and heartbeat monitoring in the future.

so mainly asking out of curiosity to see if anyone in the community is working on anything individually

[plvsjason]

Not as far as getting it in to production yet as still not quite sure how to index my own data. But this is going to be a key aspect of any deployment so will add to the discussion as soon as I have more on this.

I am curious @C-Loftus if you have this running on its own server or in Kubernetes or docker? Plan here is to run multiple fail over instances etc so not sure if we can connect to discuss?

[plvsjason]

I will reach out via links as suggested, many thanks.

[ktk]

We do run it in HA, it's very straight forward. You can use the same index to spin up more instances in Kubernetes under load.

[C-Loftus]

We do run it in HA, it's very straight forward. You can use the same index to spin up more instances in Kubernetes under load.

What do you use for observability?

[ktk]

add /ping to "Backend-URL" in endpoint, for example: https://qlever.cs.uni-freiburg.de/api/osm-planet/ping

[ktk]

That won't help for tracing but we will add OpenTelemetry eventually.