Hello,
We are trying to start up a tribe node that would connect to ~60 clusters
all consisting of several nodes.
We use unicast from the tribe server to discover all of the cluster master
nodes. Master nodes are fixed (if a master fails, then other machines in
the cluster will be out of function anyway).
The connection step takes a while, where servers appears to open a large
number of threads (~1000 AFAICT, by looking at /proc info) before failing
with this:
[2014-06-06 18:27:51,122][ERROR][bootstrap ] [Ghaur]
{1.2.0}: Startup Failed ...
- ElasticsearchException[unable to create new native thread]
OutOfMemoryError[unable to create new native thread]
We attempted to change the java -Xss parameter to adjust the stack size,
however it didn't have any effect.
We understood from the code that it creates a thread to wait on a
connection to each node in each cluster. In our case this can go up to
several hundred machines (although they aren't yet all installed).
If the tribe works by querying each individual node, then it might not
scale for our system. Instead we were hoping that it can work by proxying
queries through the master node. Is something alike possible?
Any help or advice is appreciated.
Thanks,
Srecko
--
You received this message because you are subscribed to the Google Groups
"elasticsearch" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To view this discussion on the web visit
https://groups.google.com/d/msgid/elasticsearch/0f14795f-f13d-4101-9726-0d573e3cba1c%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
