On a 32GB RAM server, I often run out of system memory leading to the 
kernel killing the orientdb process (to clarify, this is not a Java heap 
outofmemory error). Seems to happen when
- large indexes are being rebuilt
- heavy writes (multiple scenarios)

Before the orientdb process is killed by the kernel, the VSZ RSS values 
explode. However, every value shown in jconsole is under the desired 
limits. Currently running heap value of -Xmx7072m . I've read many posts 
saying to keep this low, but seems to be necessary to avoid Java Heap 
memory exhaustion. Also gc pauses are not an issue for us on these write 
heavy servers.  Also running -Dstorage.diskCache.bufferSize=21000.  Also 
running -XX:+PerfDisableSharedMem. How would I go about getting more 
specific information about the memory usage? I've read some posts 
mentioning "memory dump", but how exactly would I generate that?

-- 

--- 
You received this message because you are subscribed to the Google Groups 
"OrientDB" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
For more options, visit https://groups.google.com/d/optout.

Reply via email to