Options to repair elasticsearch when BulkResponse hangs


#1

I have a customer instance of elasticsearch 1.5.2 where a Java client process (using the Java API) is consistently experiencing a thread hang. The client process is using an ES transport client, and issuing a BulkRequest. The thread appears to be blocked in the following call:

BulkResponse bulkResponse = (BulkResponse) response.actionGet();

This may be distantly related to the following issue (which should've been fixed in 1.5.2):

I have a single-node elasticsearch server instance running on Windows Server 2008, so relocating of a shard is probably not a culprit.

If indeed there is some issue with the ES file system, however, what are my choices to repair the elasticsearch instance, and what can I look for to extract more evidence? The elasticsearch logs (both server and client) are in debug mode, and there's nothing immediate standing out.


(system) #2