View Revisions: Issue #3010

Summary 0003010: Batches get stranded in ready state in staging on a cluster because the ack that moves them to done occurs on a different server
Revision 2017-03-10 15:24 by chenson
Description Batches get stranded in ready state in staging on a cluster because the ack that moves them to done occurs on a different server.

Getting rid of the READY state in favor of just CREATE and DONE. When a batch is finished extracting it will go to directly to DONE. This should be fine with the new staging purging strategy of waiting until the outgoing or incoming batch has been purged. If the old purging strategy in employed, then a batch will be purged after the ttl hits. This means that the ttl needs to be longer than the longest amount of time it takes to extract. This is ok because it is advanced behavior (you have to explicitly turn on the old purge strategy).
Revision 2017-03-10 15:24 by chenson
Description Batches get stranded in ready state in staging on a cluster because the ack that moves them to done occurs on a different server.

Getting rid of the READY state in favor of just CREATE and DONE. When a batch is finished extracting it will go to directly to DONE. This should be fine with the new staging purging strategy of waiting until the outgoing or incoming batch has been purged. If the old purging strategy in employed, then a batch will be purged after the ttl hits. This means that the ttl needs to be longer than the longest amount of time it takes to extract. This is ok because it is advanced behavior.