Constanly crashing

Hi all, my XWiki installation is constantly crashing. Several times each hour. For two users which don’t even really use the Wiki this is absolutely inacceptable.

Problem is - I don’t know why. The only thing I did was LDAP authentication, after trying (and failing) with OpenID connect authentication (I disabled the latter plugin again after LDAP worked).

I host XWiki in Kubernetes using my XWiki helm chart, and my configuration can be found here.

I would appreciate any help! This is intended to be our department’s wiki - under those circumstances it’s just not possible :frowning:

Thanks for any help!

I kept the log running now and the only exceptions I saw so far were the ones I appended to my gist. To me, they don’t look “crash-worthy”.

I had the thought about resources - what is the resource usage of a single, unused XWiki instance?

Anyone? :slight_smile:

What do you call a crash?

I’ve checked the logs and there’s a broken pipe between Tomcat and browser or any server in between your browser and Tomcat. So you should check/fix this. XWiki is trying to return data to the client side and Tomcat is saying that the connection to the client side is dead (in general this means the client side disconnected).

See https://www.xwiki.org/xwiki/bin/view/Documentation/AdminGuide/Installation/#HHardwareandSoftwarerequirements but I really don’t think that’s your issue seen the broken pipe.

What do you call a crash?

Well. The wiki becomes unresponsive, and after a couple of seconds I get the “Wiki initializing (%…)” screen. To me that’s a crash, but you’re right that was not fully precise.

I’ve checked the logs and there’s a broken pipe between Tomcat and browser or any server in between your browser and Tomcat. So you should check/fix this. XWiki is trying to return data to the client side and Tomcat is saying that the connection to the client side is dead (in general this means the client side disconnected).

Very weird. Nothing fancy going on here. Any hints on how that can happen?

UPDATE: maybe I should be more preceise again :slight_smile: . Some context: I run XWiki in Kubernetes, using the “standard” nginx-ingress-controller from Helm, and ACME encryption using cert-manager. All of that runs in Azure, with an initial load balancer in front of everything.

Nothing here is unusually configured, there are no long running requests, and the cluster is basically empty execpt for XWiki (as of now :wink: ).