Details
Description
go-memcached upr_feed.go UprGetFailoverLog is very slow - a few minutes to return the failover logs if the number of vbuckets passed in is big. This cause goxdcr start replication seems to hang. It affects pause-resume test in QE's suit.
The implementation of UprGetFailoverLog seems to sending\receiving failover log one by one for each vb in sequential. Can it parallel the processing to speed it up?
Attachments
Issue Links
- relates to
-
MB-13581 Avoid calling GetUprFailoverLog if checkpoint record contains failover uuid
- Closed