Details
-
Bug
-
Resolution: Fixed
-
Blocker
-
5.1.1, 5.5.0
-
Untriaged
-
Centos 64-bit
-
-
Yes
Description
I believe we have seen this problem before when Sitaram Vemulapalli made changes to go-couchbase (see MB-29034). Those changes have been reverted and the regressions have been recovered.
Sitaram's patch was re-introduced again in build 5.5.0-2546 and performance numbers got worse again.
See:
http://showfast.sc.couchbase.com/#/timeline/Linux/n1ql/Q1_Q3/MOI
http://showfast.sc.couchbase.com/#/timeline/Linux/n1ql/Q1_Q3/Plasma
http://showfast.sc.couchbase.com/#/timeline/Linux/n1ql/Q5_Q7/all
http://showfast.sc.couchbase.com/#/timeline/Linux/n1ql/array/all
http://showfast.sc.couchbase.com/#/timeline/Linux/n1ql/pagination/MOI
http://showfast.sc.couchbase.com/#/timeline/Linux/n1ql/soe/all
These test results are very consistent and I am very positive that we have a product issue.
Please let me know if you disagree with my assumptions.
Attachments
Issue Links
- is caused by
-
MB-29034 Query nodes activity impacted after failover of a data node due to network outage (net_tick_timeout)
- Closed