Today, we are happy to announce the bugfix release of Elasticsearch 1.3.4, based on Lucene 4.9.1. You can download it and read the full changes list here: Elasticsearch 1.3.4.
Elasticsearch 1.3.4 is the latest stable release. Users of Elasticsearch 1.3.3 who have 100 or more shards per node should upgrade.
Elasticsearch 1.3.3 included a change to the management threadpool (used, amongst other things, to collect node and index statistics) to make it a fixed pool with a queue size limited to 100. The change was added to prevent a queue explosion in the case of slow statistics collection, which was experienced by some of our bigger users. Usually, collecting statistics is very fast, but a bug introduced in a previous release slowed down the file system check enough to cause a notable delay on nodes with many shards.
Unfortunately, limiting the size of the management thread pool means that stats calls can fail on nodes with more than 100 shards, because the queue fills up. This release reverts the change to the management thread pool, and in a future release we will look at other ways of handling this scenario.