MessageSize and server OOM (possibly) related problems

[manics]

If you're switching JDKs it might be worth trying the latest version (Java 7, either Oracle or OpenJDK).

[dpwrussell]

If you're switching JDKs it might be worth trying the latest version (Java 7, either Oracle or OpenJDK).

Done. Now using openjdk-7.

Try again please Helio.

[hroque]

Well it failed again but this time with an "Failed" import message that then led to a crash when I refreshed.
Douglas this was done around 16h so you would have already restarted the server due to the LDAP no?

Here is the log file.

https://www.dropbox.com/s/5lpe76zc11u7p ... nsight.log

Cheers,
Helio

[manics]

Thanks Helio.
We'll need to see the server logs from Douglas too.

[dpwrussell]

Thanks Helio.
We'll need to see the server logs from Douglas too.

https://www.openmicroscopy.org/qa2/qa/feedback/7717/

[jmoore]

Thanks for the logs, Helio. This time there was a successful insertion of the metadata into the DB but it took some 230 minutes during which insight lost its connection.We'll look into a fix for that, but in the mean time if possible you might try kicking off a CLI import of the same file set as workaround.

Again, thanks for all your patience in helping us to track this down!
Cheers,
~Josh

[dpwrussell]

230 minutes!? Is there that much metadata being inserted into the database or is it some other aspect of the import which delays the completion past when Insight has timed out. If it's 230 minutes of inserts then either the inserts are crazy slow or we're going to be creating a pretty gigantic database!

[hroque]

but in the mean time if possible you might try kicking off a CLI import of the same file set as workaround.

Sorry but what is a CLI import.

Also about the metadata I don't think there is such amount of metadata but there is metadata in every single image file (~70000 files) so it takes forever to go trough all of them and read each single file. I think...

Cheers,
Helio

[jmoore]

Sorry but what is a CLI import.

By CLI import I meant using `$OMERO_DIST/bin/omero import` from the server, itself.

Also about the metadata I don't think there is such amount of metadata but there is metadata in every single image file (~70000 files) so it takes forever to go trough all of them and read each single file. I think...

Quite likely. I'm trying to produce a data set of a similar size here to see if I can reproduce.

Regards,
~Josh