[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Nexmark Phrase Triggering

(sorry, I missed this somehow)

The documentation for Nexmark was updated on Confluence: https://cwiki.apache.org/confluence/x/sZCzBQ 

For Performance tests to behave the same way, we need to add separate jobs and simply use a different dataset for storing the results (same as in Nexmark). Besides a JIRA ticket I created a while ago (https://issues.apache.org/jira/browse/BEAM-6012) I don't think more documentation is needed for this now.

pon., 26 lis 2018 o 21:05 Chamikara Jayalath <chamikara@xxxxxxxxxx> napisał(a):
Thanks Łukasz.

Should the solution be documented (in Beam testing guide ?) so that other performance tests can support manual triggering without affecting benchmark results in a similar manner ?

- Cham

On Thu, Nov 22, 2018 at 4:03 AM Łukasz Gajowy <lukasz.gajowy@xxxxxxxxx> wrote:
Hi all, 

BEAM-6011 is now resolved. If any of you think your changes require such additional build/performance checks, now you can run them by posting a comment on Github (eg. "Run Direct Runner Nexmark Tests") to see if everything is fine before merging a PR.


śr., 7 lis 2018 o 17:38 Andrew Pilloud <apilloud@xxxxxxxxxx> napisał(a):
My concern is that PR triggered runs should not publish results to the same table as runs on master. Looks like you already covered that in the bug.

There is also the issue that there is only one jenkins executor that can run the local runner jobs. This will only be a problem if the manual runs become frequent.


On Wed, Nov 7, 2018, 8:32 AM Łukasz Gajowy <lgajowy@xxxxxxxxxx wrote:

recent experience with Nexmark crashes made enabling Phrase Triggering in Nexmark suites even more urgent. If you have any opinions in this area feel free to share them.

Here's the link to a corresponding JIRA issue: https://issues.apache.org/jira/browse/BEAM-6011