Heap dump is really the gold standard for analysis, but if you don't want
to take a heap dump for some reason:
1. Decrease the cache sizes
2. Increase the index interval size
These in combination may reduce pressure on the heap enough so you do not
see these warnings in the log.
On Mon, Feb 27,
As a configuration issue, I haven't enable the heap dump directory.
Is there another way to find the cause to this and identify possible
configuration changes?
Thanks.
--
View this message in context:
http://cassandra-user-incubator-apache-org.3065146.n2.nabble.com/Cassndra-1-0-6-GC-query-tp73
Take a heap dump (there should be one from when you OOMed) and see
what is consuming your memory.
On Mon, Feb 27, 2012 at 3:45 PM, Roshan wrote:
> Hi Experts
>
> After getting an OOM error in production, I reduce the
> -XX:CMSInitiatingOccupancyFraction to .45 (from .75) and
> flush_largest_memta
Hi Experts
After getting an OOM error in production, I reduce the
-XX:CMSInitiatingOccupancyFraction to .45 (from .75) and
flush_largest_memtables_at to .45 (from .75). But still I am get an warning
message in production for the same Cassandra node regarding OOM. Also reduce
the concurrent compact
