Details
-
Improvement
-
Status: Open
-
Major
-
Resolution: Unresolved
-
None
-
None
-
None
-
None
Description
TolerantUpdateProcessor is being added in SOLR-445 but it's not entirely obvious what the behavior should be when using something like this is used in conjunction with Shard Splitting.
In particular what should / shouldn't happen if an update error occurs on a subShardLeader (while the shard is actively being split) after the update already succeded on the original shard leader. when TUP is not used, this error is propogated back to the client – but if TUP is being used, then should the subShardLeader' error be propogated back as a tolerated error, or a hard failure?
Attachments
Issue Links
- relates to
-
SOLR-445 Update Handlers abort with bad documents
- Closed