[SOLR-9036] Solr slave is doing full replication (entire index) of index after master restart

• Type: Bug
• Status: Resolved
• Priority: Critical
• Resolution: Fixed
• Affects Version/s: 5.3.1, 6.0
• Fix Version/s: 5.5.2, 5.6, 6.0.1, 6.1, 7.0
• Component/s: replication (java)
• Labels: impact-high
• Assignee: Shalin Shekhar Mangar
• Reporter: Shalin Shekhar Mangar
• Created: 25/Apr/16 15:19
• Updated: 02/Oct/19 17:24
• Resolved: 04/Oct/16 02:27

Description

This was first described in the following email:
https://mail-archives.apache.org/mod_mbox/lucene-solr-user/201604.mbox/[e-mail]%3E

I tried Solr 5.3.1 and Solr 6 and I can reproduce the problem. If the master comes back online before the next polling interval then the slave finds itself in sync with the master but if the master is down for at least one polling interval then the slave pulls the entire full index from the master even if the index has not changed on the master.

Issue Links

• is broken by: SOLR-7134 Replication can still cause index corruption.
• is related to: SOLR-12201 TestReplicationHandler.doTestIndexFetchOnMasterRestart(): unexpected replication failures