Open Microscopy Environment

Greetings,

We successfully installed 4.1.1b and now 4.2.0b on a testing server (centos 5.3) and everything was fine.

I recently installed 4.2.0b on a new production server and running into a problem. The blitz error log has this error repeated:

010-08-13 13:06:35,318 ERROR [ ome.services.util.ServiceHandler] (r_Worker-6) Method interface ome.services.util.Executor$Work.doWork invocation took 30138
2010-08-13 13:06:35,318 ERROR [.services.blitz.repo.AbstractRepositoryI] (r_Worker-6) Unexpected error in called executor on takeover

with exception coming from org.quartz.simpl.SimpleThreadPool$WorkerThread.run(SimpleThreadPool.java:520)

One thing we've done differently on this server is make /OMERO an NFS mount, but otherwise it is almost identical. Another minor difference, while both servers have OpenJDK 1.6 the working test server also has gcj 1.4.2.

I can't find any help on this with search. Maybe someone here has an idea?

Thanks so much,
Van

As long as GCJ isn't the active JVM it should not present a problem. Can you attach your var/log/Blitz-0.log with the error messages so that we can get understand the context surrounding the errors?

Attached is Blitz-0.log, although now there are different errors including null pointer exception. I'm still verifiying all system components are the same because this just feels to me like a system config problem. Just haven't been able to find the diff yet

Here's another Blitz-0.log with the error I originally mentioned.

The client seems to be connecting to the server properly and we've uploaded images, etc. So I'm not even sure this is a problem.

Thanks,
Van

Ok, it looks like this is an NFS problem.

For some reason the /OMERO/.omero/repository/f3774895-8b33-481b-815c-157f07b2d4f7/repo_uuid is created, but it is zero length and blitz write to it.

If we make /OMERO a local dir everything starts up fine.

We tried changing NFS to cifs on the server side and that didn't help.

Anyone have clues on how to configure NFS for Blitz?

We made 2 changes and it's working much better now. Sysadmin made some change to the way locking over NFS was handled and we did the NFS mount as omero user instead of root.

We're still getting another error now, Blitz logfile attached.

Reply posted at http://www.openmicroscopy.org/community/viewtopic.php?f=5&t=457&p=1396#p1396

Cheers,
~Josh