1632424 - 1.83 - 5.96% perf_reftest_singletons external-string-pass.html (macosx1014-64-shippable) regression on push ce0291e3f74342d7543ad191b52f84f611012787 (Tue April 21 2020)

Status: RESOLVED WORKSFORME

(Core :: XPCOM, defect)

Description

[Alexandru Ionescu (needinfo me) [:alexandrui]]

Perfherder has detected a Talos performance regression from push ce0291e3f74342d7543ad191b52f84f611012787. As author of one of the patches included in that push, we need your help to address this regression.

Regressions:

6% perf_reftest_singletons external-string-pass.html macosx1014-64-shippable opt e10s stylo 703.25 -> 745.15
2% perf_reftest_singletons external-string-pass.html macosx1014-64-shippable opt e10s stylo 714.16 -> 727.23

Details of the alert can be found in the alert summary, including links to graphs and comparisons for each of the affected tests. Please follow our guide to handling regression bugs and let us know your plans within 3 business days, or the offending patch(es) will be backed out in accordance with our regression policy.

For more information on performance sheriffing please see our FAQ.

Comment 1

[Jean-Yves Avenard [:jya]]

must be a coincidence. There's nothing in this patch that could explain the measured behaviour.
Keeping ni? to look into it shortly.

Comment 2

[Alexandru Ionescu (needinfo me) [:alexandrui]]

I did a backfill on revision 06d3318904d2e to check for infra changes. Indeed this is a weird behavior.

Comment 3

[Rachel Tublitz [:rachel]]

Hi Jean-Yves, are you still looking into this one?

Comment 4

[Julien Cristau [:jcristau] (back April 22)]

From the graphs it's not clear to me there's actually a regression here, I'll call this worksforme.

Comment 5

[Alexandru Ionescu (needinfo me) [:alexandrui]]

The graph is pretty clear to me. I pushed to try a backout of this patch (not actually backed out). I will let you know the results. You will be able to see them here once the jobs are finished.

Comment 6

[Julien Cristau [:jcristau] (back April 22)]

That seems to show no difference?

Comment 7

[Alexandru Ionescu (needinfo me) [:alexandrui]]

That's right.

Comment 8

[Jean-Yves Avenard [:jya]]

Thanks for digging into this, it made no sense to me.

Comment 9

[Alexandru Ionescu (needinfo me) [:alexandrui]]

It's an infra change, most likely. I will push the last good job to try to make sure we have the correct baseline for try and autoland.