Uploading long stacks crashes OMERO

[Lipplab]

Hi everybody it seems we have a problem,
since we moved our server to OS X 10.9 Mavericks two months ago we experience some problems. Last week we udated the system to 5.0.2 and still the bugs persist. I allready increased java heap space to double the standard value.

When uploading larger tiffstacks of about 5000 images the client crashes and the upload fails. With small stacks of large size there is no problem. In the earlier versions of omero 5 the server was stuck too (no login possible any longer). Last changes I made were upgrade to jdk 7.40 and the omero upgrade 5.0.1->5.0.2.
Any suggestions?

Best regards

Benjamin

System: Mac Mini late late 2012, I7 2.6GHz, 16 GB RAM, MacOS X Mavericks 10.9.2

Attached is the client log. Most server logs (e.g. Blitz) could only be supplied externaly due to size restrictions. I could serve it via Dropbox?!

[jmoore]

Hi Benjamin,

thanks for the log. Your server is clearly not happy with the import. What are the current JVM settings for your servers? You can usually see these by using:

Code: Select all

jps -v


as the user running the server. One issue especially on upgrade is not upgrading the configuration values post-upgrade as described at https://www.openmicroscopy.org/site/support/omero5/sysadmins/windows/server-installation.html#jvm-memory-settings.

If that's not the issue, finding out a bit more about this fileset would likely be useful. How large is each tiff? How was it acquired? Etc.

Cheers,
~Josh

[Lipplab]

Hi Josh,
true I forgot to reinstate the right values. Nevertheless the problem was there in 5.0.1 even with adapted values. We'll test it with 5.0.2 now. The return for jps -v was
61755 Jps -Dapplication.home=/Library/Java/JavaVirtualMachine/jdk1.7.0_55.jdk/Contents/Home -Xms8m.
The files are Leica lif files by confocal imaging with a stack number of about 5000 images. I'll ask my boss for the total size. But as I mentioned the overall size does not matter. There is no issue with larger lif files which have lower stacknumber.

Thanks in advance.

Benjamin

[jmoore]

true I forgot to reinstate the right values. Nevertheless the problem was there in 5.0.1 even with adapted values. We'll test it with 5.0.2 now.

Great. Keep us posted.

The return for jps -v was
61755 Jps -Dapplication.home=/Library/Java/JavaVirtualMachine/jdk1.7.0_55.jdk/Contents/Home -Xms8m.

Hmmm....this doesn't include the servers, but if they were running with the defaults, that will definitely be too low.

The files are Leica lif files by confocal imaging with a stack number of about 5000 images. I'll ask my boss for the total size. But as I mentioned the overall size does not matter. There is no issue with larger lif files which have lower stacknumber.

Understood. Thanks!
~Josh.

[Lipplab]

Hi Josh,
it seems that did it. After the update and the renewal of the heap size settings we expereinced no new crash! Thanks a lot.

Best regards

Benjamin