Every 50 minutes, get_cmd time has increased very significantly

I have announced this issue for a long time. But It is difficult to figure out the cause lol …
Is there anyone has experienced like this issue?

I observed this issue using cbstats.
Steps.

  • /opt/couchbase/bin/cbstats localhost:11210 -b bucketName reset
  • watch -d -n .1 “/opt/couchabse/bin/cbstats localhost:11210 -b bucketName timings”

Every 50 minute (1:50, 2:50,3:50…) , “get_cmd” time is increased significantly.
When there is a normal 1ms under, but there is over 100ms at 50min.

I have this issue on jira below. Please help me… My service is sensitive to this problem because it needs to get results within 30ms .

http://issues.couchbase.com/browse/MB-13032?focusedCommentId=111387&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-111387