Hello,
Here's my threadpool configuration (test conf):
https://gist.github.com/3762162
And here's the stats elasticsearch:
https://gist.github.com/3762104
We can see that "threads:" is 37, but in the "thread_pool" part we see that
only 7 are supposed to be used.
My question is, from where comes such a big difference? Is that all the
Network/Filesystem IO background threads from the JVM?
I have to run elasticsearch on a machine that only uses 512Mb of RAM (no
clusterization, only one server for the app). I reduced the number of
shards to 1 and set replica to 0 to save what I can in terms of memory, but
it seems that limiting the number of threads in each threadpool is a dead
end reguarding saving memory.
Philippe
--