Error number 0 in 11: Uncaught exception when resizing images + many other problems

Help / Discuss
#1

Hi all

10.11.2 user here, at the end of my wits. I have to mention that our setup here is not really simple: We’ve got three app servers, two oracle db servers, two instances of XWiki per app server, each pointing to one specific database instance. Since I didn’t really figure out the clustering problems I faced today, I wanted to have at least one XWiki instance up and running, on one server. No fancy clustering. Sidenote: Offline setup.

I picked one server and noticed that some things were not working due to NFS issues (but everything else did work), so I switched to another server. There, NFS runs smoothly, but I get this when displaying an image with ?width=400 appended:

Error number 0 in 11: Uncaught exception
...
Caused by: java.lang.NoClassDefFoundError: 
     com/github/jaiimageio/stream/RawImageInputStream

While preparing this question here, some more exceptions showed up:

[DefaultQuartzScheduler_Worker-8] ERROR o.q.c.JobRunShell              - Job 
   DEFAULT.xwiki:Scheduler.NotificationEmailHourlySender_0 threw an unhandled Exception:
   java.lang.RuntimeException: Failed to load component for type [interface 
   org.xwiki.context.ExecutionContextManager] for hint [default]

They might be related, I can’t tell. As well, the editor doesn’t seem to work - spinning wheel.

The way I set it up was according to the offline installation manual, so the offline repository is there. The installation went through fine. I then sent over our look and feel and some starting pages via REST, which also worked perfectly fine.

What really puzzles me is that the SAME config on the SAME database on a DIFFERENT server works.

Any hint highly appreciated!

Cheers

André

#2

This usually means that some JARs in WEB-INF/lib are not correct. Could it be that you upgraded XWiki by unzipping on top of an existing instance?

You need to be sure to not do that since otherwise the old JARs would still be there.

Can you confirm if that’s your problem?

In any case it means the xwiki WAR is not valid and you should redo it.

#3

Hi Vincent

Actually no. I dropped all tables in the DB, recreated the permanent directory structure and threw in a fresh WAR.

I’ll double check, but this will take a few minutes because my teammates need to work on the database.

I’ll keep you posted.

Cheers

André

Edit: PS: This was also my first guess, but somehow it doesn’t make sense. I took the same WAR as on the other system.

#4

Hi Vincent

Update - I bounced the tomcat (which I shouldn’t do on a production system) and it’s working now.

Is it possible that tomcat keeps some things cached? I also noticed a few other issues that could be related:

  • Even if the wiki is down (no webapp present in the directory), hibernate keeps the database session pool up (which lead to “too many concurrent sessions”)
  • Some files inside the solr runtime directory seem to be locked after removing the webapp. I noticed because I wasn’t able to rm them.

What is the cleanest and best way to bring the wiki down?

Cheers

André

.oO( maybe I should open another thread …)

#5

Stop the servlet container.