Details
-
Bug
-
Resolution: Not a Bug
-
Major
-
6.5.0
-
Triton cluster
-
Untriaged
-
Centos 64-bit
-
No
Description
Despite the statement in MB-31848 the throughput is actually degraded after that change.
I'm getting 80k q/sec versus 100k in Daily test for Q1
The latency profile is interesting. There are no more slow requests in higher percentiles but overall latency got way higher. Which also caused throughput degradation:
Full comparison:
Logs from 1719:
services:
triton-srv-01-ip6.perf.couchbase.com:kv
triton-srv-02-ip6.perf.couchbase.com:kv
triton-srv-03-ip6.perf.couchbase.com:kv
triton-srv-04-ip6.perf.couchbase.com:kv
triton-srv-05-ip6.perf.couchbase.com:index
triton-srv-06-ip6.perf.couchbase.com:n1ql
Q1 performance in Daily:
Attachments
Issue Links
- is caused by
-
MB-31848 Improve servicers throughput
- Closed