[CONTROLLER-1777] Possible delete performance degradation Created: 03/Oct/17  Updated: 07/May/19  Resolved: 07/May/19

Status: Resolved
Project: controller
Component/s: clustering
Affects Version/s: Carbon
Fix Version/s: None

Type: Bug
Reporter: Vratko Polak Assignee: Vratko Polak
Resolution: Cannot Reproduce Votes: 0
Labels: None
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified
Environment:

Operating System: All
Platform: All


External issue ID: 9236

 Description   

After merging CONTROLLER-1746 fix [0] into Nitrogen, CSIT dsbenchmark test started showing that performance of delete operation is now both worse and more volatile. See test duration graph [1].

It is possible that the new behavior is a logical consequence of the bug fix and there is nothing to improve, but the volatility increase suggests there might be a performance bug somewhere.

[0] https://git.opendaylight.org/gerrit/61284
[1] https://jenkins.opendaylight.org/releng/view/controller/job/controller-csit-3node-periodic-benchmark-all-nitrogen/plot/getPlot?index=6&width=900&height=900



 Comments   
Comment by Robert Varga [ 14/Nov/18 ]

vrpolak is this still present?

Comment by Vratko Polak [ 14/Nov/18 ]

Not sure whether the average performance is still worse than in early Nitrogen, but the results are still volatile. Current graph of delete times is here: [2].

[2] https://jenkins.opendaylight.org/releng/view/controller/job/controller-csit-3node-benchmark-all-neon/plot/getPlot?index=6&width=900&height=900

Comment by Tom Pantelis [ 14/Nov/18 ]

I think the fluctuations are probably due to the amount of load in the infra/env at that time - a volatile env is not ideal for benchmarking/performance testing.

Comment by Vratko Polak [ 14/Nov/18 ]

I remember that in early Nitrogen the volatility was low and it jumped sharply. I do not recall any infra change at that time.

We should be able to eliminate the environment effect by re-testing Carbon, if the current CSIT code can run that.

Comment by Robert Varga [ 07/May/19 ]

Cannot reproduce now, let's reopen if occurs again.

Generated at Wed Feb 07 19:56:26 UTC 2024 using Jira 8.20.10#820010-sha1:ace47f9899e9ee25d7157d59aa17ab06aee30d3d.