Hello devs. The purpose of this email is to point you to the results we have and to problems the results show.
We have several jobs jobs running in releng which shows now more "readable" graphs from jobs benchmarking the datastore. The history is now more than 10 job runs and I'd like to point you to some results which are not expected. We have totally 12 jobs, 6 for 1node setup, 6 for 3nodes setup; for each stream Be,B,C and "only" and "all" feature install scope. 1node jobs are basically ok, graphs don't show any problems and lines keep the trend https://jenkins.opendaylight.org/releng/view/controller/job/controller-csit-1node-periodic-benchmark-only-boron/plot/ https://jenkins.opendaylight.org/releng/view/controller/job/controller-csit-1node-periodic-benchmark-only-carbon/plot/ 3node graphs are more "variable". PUT and MERGE operations look ok, they keep the trends, but READ and DELETE vary. https://jenkins.opendaylight.org/releng/view/controller/job/controller-csit-3node-periodic-benchmark-only-boron/plot/ https://jenkins.opendaylight.org/releng/view/controller/job/controller-csit-3node-periodic-benchmark-only-carbon/plot/ Lines described as FOL_CONFIG_ or FOL_OPERATIONAL are the results of the benchmark running on the default config or operational shard follower. Lines with CONFIG_ and OPERATIONAL_ are from the leader node. Given graphs don't look well, therefore you may want to open the graph in a new browser tab and play with width and height params. https://jenkins.opendaylight.org/releng/view/controller/job/controller-csit-3node-periodic-benchmark-only-carbon/plot/getPlot?index=4&width=750&height=450 vs. https://jenkins.opendaylight.org/releng/view/controller/job/controller-csit-3node-periodic-benchmark-only-carbon/plot/getPlot?index=4&width=1750&height=1450 The most visible problem is the READ operation. https://jenkins.opendaylight.org/releng/view/controller/job/controller-csit-3node-periodic-benchmark-only-boron/plot/getPlot?index=1&width=1750&height=1450 The time of the operation may vary from less than 1s till 160s and is related to operations done on both leader and follower nodes. All graphs and logs are available in jenkins for your further investigation. Peter Gubka
_______________________________________________ controller-dev mailing list controller-dev@lists.opendaylight.org https://lists.opendaylight.org/mailman/listinfo/controller-dev
