Failed during dump collection and unavailable_shards_exception

[mumlax]

As already mentioned in #552 (comment) I'm experiencing an "Exception during collection dump" because of a "ConnectionTimeout caused by ReadTimeoutError" (see logs and other details in linked comment).
I'm opening a new issue now, because a new circumstance appreared.
In the meantime, on some new tries to do a resync, I get a new error. Again an "Exception during collection dump" with the same Traceback, but at the end another reason.

OperationFailed: (u'611 document(s) failed to index.', [])

The array contains (I think 611) elements of the following form:

{u'index': {u'status': 503, u'_type': u'dbXY', u'_id': u'xyz', 
  u'error':  {u'reason': u'[indexAB][3] primary shard is not active Timeout: [1m], request: 
    [BulkShardRequest to [indexAB] containing [314] requests]', 
    u'type': u'unavailable_shards_exception'},
  u'_index': u'indexAB'}}

I double checked it, ElasticSearch is running.
Some time before I had played around with the BulkSize and set it to 3000. Because of the 'unavailable_shards_exception" I hoped to have found the reason and reset it to 1000. No improvement.
Sometimes the connector is stopped by the ReadTimeoutError, sometimes by the OperationFailed error, I can't detect a pattern when which error appears. I don't even know if these errors have the same source or if there are multiple problems...everything came with update to v2.5.0.

Any ideas on this problem(s)?

[ShaneHarvey]

The problem is that the Elasticsearch doc managers do not retry on connection/operation failure. I'm closing this in favor of the elastic2-doc-manager issue that tracks retrying operations on failure: yougov/elastic2-doc-manager#18

[mumlax]

Alright. I'll follow this issue. Thanks!
The reason for the long responsetime of ES was simple: diskspace was full, ooops ;) won't happen again.