seanderickson
commented
10 years ago Further work needs to be done to ensure that the object graph that this report is pulling into memory is optimized. Regarding the issues of the heap size:
in [HMS #356663] Memory settings for tomcat servers on orchestra on 2012-04-12, David Hummel writes:
On Wed, Apr 11, 2012 at 11:48:59PM -0400, Sean Erickson via RT wrote:
For our screensaver.med.harvard.edu instance, we are seeing intermittent Java heap errors when pulling datasets out of the database. We have done profiling for this problem in other cases in the past, and have found memory optimizations that have helped minimize issues. However, I'm wondering in this case if an increase (to say, 2 GB) would alleviate our issues, which have become more varied and random.
Is there RAM available on these machines that would allow us to increase the java heap? What would the availability be?
We don't have the capacity to allocate large amounts of additional memory to many Tomcat instances at this time, but we do have a bit of headroom on the Tomcat servers. I've increased the maximum heap size from 1.5GB to 2.0GB on instances:
dev.screensaver.med.harvard.edu stage.screensaver.med.harvard.edu screensaver.med.harvard.edu
This will be reflected the next time you restart these instances. Let me know if this helps to alleviate the issues you've been seeing.
combinations of the columns:
"date of first activity" "date of last activity" with "assay readout type"
when added to the report, cause an out of memory exception.
It is possible that the memory requirements of this report are just greater than the heap capacity.