Hi,
I have a simple setup: a java web application communicating with a remote
elasticsearch server (via the java client); is there any way to configure
the thread pool of this client?
I am performing a stress testing session and I'm seeing a very strange
behavior with the way Tomcat allocates threads: on heavy load, it allocates
about 100 threads and doesn't go over that.
I have of course tuned Tomcat to be able to go well over 100 connections,
which it does - other stress testing sessions will indeed create over 300
threads and server as many connections.
I will mention that the operation I'm performing is a simple geospatial
search (in a polygon), on an index of 20000 location entities, with the
latest Elasticsearch (0.19.2).
Any help on this is appreciated.
Thanks.
Eugen.
Thread pools on the client side are basically meaningless, as they are
"work" thread pools (for search, index, and so on). I don't know why tomcat
won't create more than 100 threads in your case, but it does not seem like
it relates to elasticsearch, or at least I can't see how it does.
On Mon, Apr 9, 2012 at 1:19 AM, Eugen Paraschiv hanriseldon@gmail.comwrote:
Hi,
I have a simple setup: a java web application communicating with a remote
elasticsearch server (via the java client); is there any way to configure
the thread pool of this client?
I am performing a stress testing session and I'm seeing a very strange
behavior with the way Tomcat allocates threads: on heavy load, it allocates
about 100 threads and doesn't go over that.
I have of course tuned Tomcat to be able to go well over 100 connections,
which it does - other stress testing sessions will indeed create over 300
threads and server as many connections.
I will mention that the operation I'm performing is a simple geospatial
search (in a polygon), on an index of 20000 location entities, with the
latest Elasticsearch (0.19.2).
Any help on this is appreciated.
Thanks.
Eugen.