[SOLR-5495] Recovery strategy for leader partitioned from replica case.

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 4.9
• Assignee: Timothy Potter
• Reporter: Mark Miller
• Created: 23/Nov/13 17:27
• Updated: 11/Jul/14 15:18
• Resolved: 24/Jun/14 16:58

Description

We need to work out a strategy for the case of:

Leader and replicas can still talk to ZooKeeper, Leader cannot talk to replica.

We punted on this in initial design, but I'd like to get something in.

Issue Links

• is related to: SOLR-6189 If the node hosting a replica is not "live", then leader-initiated recovery may publish the "down" state more than once.
• relates to: SOLR-5482 We should add an option to the ChaosMonkey's to do more complicated partition failures on Linux, as well as simulate hard fails of Jetty.