health probe

[cpanato]

Hello
this is a nice product, i like it thanks for that

One question does the api have any health endpoint to check? I would like to setup k8s Liveness and Readiness Probes but did not found in the documentation

thanks!

[FZambia]

Hello, thanks.

At moment Centrifugo has no health probe endpoint, there is also similar request in Centrifuge library which is the core of Centrifugo - centrifugal/centrifuge#27 - with a bit more details.

Could you tell me which things you expect to be checked when calling this endpoint?

[sj26]

Imho, a fixed "200 OK" response would be sufficient. It should exercise that the application is available to service requests. Generally health checks should not check upstream services (i.e. redis).

[FZambia]

ok, will include health endpoint into next release then if it already has benefits in this state. At work we have services that check external storages in liveness probe. I suppose checking external services has some reasoning in certain scenarios - broken reconnect, possible deadlock in drivers, connectivity lost forever for the whole underlying machine. But this can be an extra functionality added later.

[cpanato]

thanks, sounds great. please let me know if need help or where can i help :)

[sj26]

That's true, Kubernetes has two different probes which differentiate "this service is alive" versus "this service is healthy and all upstreams are okay"

https://kubernetes.io/docs/tasks/configure-pod-container/configure-liveness-readiness-probes/

[FZambia]

Released with v2.1.0, docs here

[sj26]

Thanks!

[cpanato]

thanks, I've updated the helm charts to use that