Java client for "Too many open files"

Elastic Stack Elasticsearch
3

Oh, sorry, maybe I wasn't that clear on which side the "Too many open
files" appears; this is happening on the Elasticsearch Engine, not the
client, the client is hanging, not throwing any kind of exception ...

Thanks,
Alin

On Oct 24, 3:27 pm, Gautam Shyamantak gau...@datarpm.com wrote:

Check if you are closing the Client after using it or not.

On Mon, Oct 24, 2011 at 5:39 PM, Alin Popa alin.p...@gmail.com wrote:

Hi guys,

While working with Elasticsearch, we've noticed an issue regarding the
Java client when Elasticsearch Engine is crashing with "Too many open
files":

  • I know, installation instructions suggests that we need to increase
    the default number of opened files descriptors (http://
    Elasticsearch Platform — Find real-time answers at scale | Elastic), but
    let's say that I forgot to do that;
  • after a while, in the console of elasticsearch engine, I can see the
    follow stack trace which is pretty obvious:

[2011-10-24 14:53:46,582][WARN ]
[netty.channel.socket.nio.NioServerSocketPipelineSink] Failed to
accept a connection.
java.io.IOException: Too many open files
at sun.nio.ch.ServerSocketChannelImpl.accept0(Native Method)
at
sun.nio.ch.ServerSocketChannelImpl.accept(ServerSocketChannelImpl.java:
152)
at

org.elasticsearch.common.netty.channel.socket.nio.NioServerSocketPipelineSi nk
$Boss.run(NioServerSocketPipelineSink.java:244)
at

org.elasticsearch.common.netty.util.ThreadRenamingRunnable.run(ThreadRenami ngRunnable.java:
108)
at org.elasticsearch.common.netty.util.internal.DeadLockProofWorker
$1.run(DeadLockProofWorker.java:44)
at java.util.concurrent.ThreadPoolExecutor
$Worker.runTask(ThreadPoolExecutor.java:886)
at java.util.concurrent.ThreadPoolExecutor
$Worker.run(ThreadPoolExecutor.java:908)
at java.lang.Thread.run(Thread.java:662)

  • from this moment on, the elasticsearch server is down and everytime
    that I try to connect using the TransportClient (connecting to port
    9300), it hangs for ever;
  • I've tried setting the following options to the netty transport:
    network.tcp.connect_timeout, connect_timeout and
    transport.tcp.connect_timeout, each to value "3s" (as I've seen here:

https://github.com/elasticsearch/elasticsearch/blob/master/modules/el...
)
but no success.

So, my questions are:

  1. Am I setting the right timeout flag to the client ? If so, doesn't
    work as expected ?
  2. Am I looking into the wrong place ? Maybe I'm missing some other
    setting that is used for exactly this purpose ?

Thanks,
Alin