search

kubesphere / ks-jenkins

Jenkins distribution for Kubesphere
Apache License 2.0
25 stars 30 forks source link

liveness probe and readiness probe of ks-jenkins failed #70

Open showjason opened 2 years ago

showjason commented 2 years ago

OS info: CentOS Linux release 7.9.2009 (Core) , 4C/16G

Kubernetes version: image

Docker version: image

KubeSphere version: v3.1.1

Issue Description: After I finish KS installation offline, I enable DevOps by following the step of Enable DevOps after Installation

Here list the snapshots: clusterconfiguration image ks-jenkins pod status and events: image image log of ks-jenkins pod image Helm list image I find that jenkin uc is not deployed, is this the reason why ks-jenkins is not ready Please, experts come up with some solutions or ideas!

JohnNiang commented 2 years ago

/kind support Thanks for reaching out here. There are several step you can try:

LinuxSuRen commented 2 years ago

I find that jenkin uc is not deployed, is this the reason why ks-jenkins is not ready

All Jenkins plugins are installed in the ks-jenkins container image since v3.x. So, we don't need a private uc (Jenkins update center) anymore.

showjason commented 2 years ago

/kind support Thanks for reaching out here. There are several step you can try:

  • Describe the Pod ks-jenkins and check the exit code of Container ks-jenkins image

  • Check latest events image

  • Check full logs of previous Pod ks-jenkins image

@JohnNiang, I did some checks as you said, but still could not get obvious error info. Here screenshots listed above is what I do, please provide some in-deep analysis or directions

showjason commented 2 years ago

I find that jenkin uc is not deployed, is this the reason why ks-jenkins is not ready

All Jenkins plugins are installed in the ks-jenkins container image since v3.x. So, we don't need a private uc (Jenkins update center) anymore.

Ok, thanks for quick reply!

LinuxSuRen commented 2 years ago

Try remove liveness probe and readiness probe, then see if it can start successfully.

showjason commented 2 years ago

Try remove liveness probe and readiness probe, then see if it can start successfully.

It's working now after removing liveness probe and readiness probe. However, I am confused that http://<POD_IP>:8080/login is only used by probes to do health check or also plays other role in jenkins. If the latter, even though ks-jenkins is ready, http://<POD_IP>:8080/login still refuses connection, will there be potential risk?

LinuxSuRen commented 2 years ago

I guess your cluster might have a network issue.

showjason commented 2 years ago

I guess your cluster might have a network issue.

It's weird, in my cluster, every component is running well, I run curl -v http://10.244.179.7:8080/login in my cluster and got response 200. image

Anyhow, it is runing now, if any new issues I found, I will be back, and ask for support. @LinuxSuRen, thanks!

wyuxiao729 commented 2 years ago

The same problem was assumed to be caused by the NTP server, but I tried to set jenkinsJavaOpts_Xms and jenkinsJavaOpts_Xmx to less than 1000M, and the problem was resolved。