Well, it would be awkard to get right. At what point should a commit be sent? When the queue was empty? That could be after every document when incoming docs were slow. And what about the other shards? CDCR doesn't really know when _all_ the docs in _all_ the shards have been sent. On the autocommit interval on the source? I guess this latter is possible, but why bother when you can do the same with a well tested mechanism? And if the source did send commits, how would those interact with autocommits on the target? We discourage clients sending explicit commits after all.
In short, it doesn't seem like it adds enough value to be worth it. Best, Erick On Tue, Nov 22, 2016 at 12:49 PM, gayatri.umesh <gayatri.um...@mathworks.com> wrote: > Thank you again Erick. Added autoSoftCommit settings in target solrconfig and > it works now. > > As CDCR does not auto-commit on the target upon replication, is there a > specific reason for this? > > Thanks, > Gayatri > > > > -- > View this message in context: > http://lucene.472066.n3.nabble.com/Solr6-CDCR-indexing-doc-to-source-replicates-to-target-doc-not-searchable-in-target-tp4306717p4306975.html > Sent from the Solr - User mailing list archive at Nabble.com.
