[ome-users] 100 % CPU Usage on 6 Cores for 3 Days

[Andreas Mueller]

  Hi Mark,

  thanks a lot for your work !!

  The surface01OG488_Ste13_01.vsi was imported - one and a half year ago
  - with OMERO 5.3.4. In the mean time we don't touched the image.

  On monday I was searching for a picture for a presentation and I found
  a lot of that old vsi-files. I clicked them all and was woundering
  why the system slows down and got now previews.

  In my htop I found 6 cores with 100%. I tryed to do:

    bin/omero sessions who

  but got no response. So I decided to kill -9 the java process (a
  normal kill didn't work)    # Monday 2pm

  After a

    bin/omero admin restart

  everything seemed to be fine. I guessed that the issue had something
  to do with the vsi-files and so I clicked them again.

  Again:  6 cores with 100%.
  I took my bag and went home; in the next morning: 6 cores 100%.
  (the whole night !!)

  But this time I could made a

    bin/omero admin restart

  After that I wrote the mails to you.


  Best Regards
  Andreas


On 16.05.19 09:50, Mark Carroll wrote:
> Dear Andreas,
> 
> I imported surface01OG488_Ste13_01.vsi into OMERO 5.4.10 and it took an
> hour on my laptop to build pyramids but in the end I can view the two
> images just fine and play the viewer through the timepoints. The
> pyramids wouldn't keep the Blitz process busy though, that'd be the
> PixelData thread instead.
> 
> I noticed that your Blitz-0.log includes a bunch of KryoException events
> about "Encountered unregistered class ID" which I don't see locally. I
> don't think those are related to your very busy server though.
> 
> I notice that from 2pm Monday the log also notes some warnings from
> SelfCorrectingDataSource, following that there is an ongoing series of
> BitronixRuntimeException: broadly, from that time onward something seems
> to be going badly wrong with your OMERO server's connection to the
> database. It may be that restarting either or both would fix the problem
> but I appreciate that impact upon your users.
> 
> I wish I could pinpoint something that might have caused the problem.
> Let's hope it never happens again but if it does then maybe the trigger
> will become clearer. I don't suppose anybody else in the community
> recognizes this pattern of errors?
> 
> -- Mark
> 
> The University of Dundee is a registered Scottish Charity, No: SC015096
> _______________________________________________
> ome-users mailing list
> ome-users at lists.openmicroscopy.org.uk
> https://lists.openmicroscopy.org.uk/mailman/listinfo/ome-users