cube icon indicating copy to clipboard operation
cube copied to clipboard

Support K8s-style health check endpoints for Cube Store nodes

Open igorlukanin opened this issue 4 years ago • 3 comments

Is your feature request related to a problem? Please describe. Cube supports K8s-style health check endpoints while Cube Store nodes don't. Hence, it's not easy to understand if the Cube Store cluster and individual nodes have initialized and started properly.

Describe the solution you'd like Provide and document the same set of endpoints for Cube Store.

Describe alternatives you've considered —

Additional context Relevant discussion in Slack: https://cube-js.slack.com/archives/C0206EXEVAS/p1621326447001400?thread_ts=1621322103.000300&cid=C0206EXEVAS

igorlukanin avatar May 19 '21 09:05 igorlukanin

My workplace is considering using CubeJS as the API layer over our Redshift Data warehouse, and I am currently figuring out how to deploy CubeJS in our Kubernetes cluster and get it to a production ready state. The examples included in this repo make my task pretty straight forward.

However, one area that is lacking in information is how to setup proper liveness and readiness probes for each pod. It seems the /readyz and /livez endpoints are only exposed by the main API pod and refresh worker pods. Any guidance on what to do for the cubestore related pods?

saharshsingh avatar Mar 08 '23 01:03 saharshsingh

Hey, guys, is it there something like a /livez endpoint for the cube_store image ? I'm running it in production and need to check that the process is up and running. How can I do it ?

klezovics avatar Nov 28 '23 14:11 klezovics

@klezovics there is livenessProbe defined for cubestore router. However I don't see there any liveness probe for cubestore worker. The probe for worker should be clarified I guess

svdimchenko avatar Jan 22 '24 16:01 svdimchenko