Resource leak(?) when closing CloudSolrServer

[Phanindra R <ph...@gmail.com>]

Hi,

Our indexing job and expiration job run every ~60 minutes (for about 10
minutes) in the test environment, within same JVM. Every job creates a new
CloudSolrServer (decision was taken keeping other parts of system design in
mind) and invokes shutdown() after it's complete.

We have been seeing a strange issue which is the timeout exception from
Zookeeper which happens roughly *once in a week* and consistently remains
until an indexing system restart. Program is set to log if
cloudSolrServer.shutdown() fails but there's none in the logs.

I'm wondering whether it's related to our code or proper resource releasing
with CloudSolrServer?

*Details*
*SolrJ version:* 4.9.0

*Log statement from indexing*:
org.apache.solr.common.SolrException:
java.util.concurrent.TimeoutException: Could not connect to ZooKeeper <z1,
z2, z3> within 10000 ms

*Zookeeper log*
Too many connections from <indexing system ip> - max is 60

Please note that I'm re-posting this question with edited subject as it
didn't have any responses for a while.

Thanks.