1729298 - 5.32 - 4% tart / tart + 2 more (OSX) regression on Wed September 1 2021

Status: RESOLVED WORKSFORME

(Core :: Networking, defect)

Description

[Andra Esanu (needinfo me)]

Perfherder has detected a talos performance regression from push c33b8afdbd6f6bf3e152016d6bf3c5fd18625a80. As author of one of the patches included in that push, we need your help to address this regression.

Regressions:

Ratio	Test	Platform	Options	Absolute values (old vs new)
5%	tart	macosx1015-64-shippable-qr	e10s stylo webrender-sw	2.10 -> 2.21
5%	tart	macosx1015-64-shippable-qr	e10s stylo webrender	2.12 -> 2.22
5%	tart	macosx1015-64-shippable-qr	e10s fission stylo webrender	2.12 -> 2.21
4%	tart	macosx1015-64-shippable-qr	e10s stylo webrender-sw	2.10 -> 2.19

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

[Dragana Damjanovic [:dragana]]

I just go back from PTO. I will take a look.

Comment 2

[Dragana Damjanovic [:dragana]]

This is caused by our es setup that uses a proxy. The most majority of users does not use a proxy and will not experience the regression.

I will take a look what is really causing it.
The patch only access nss to disable the TLS early-data that will not be used anyway. I am wondering why it is causing 10ms delay at all. Maybe it is nss locks that are causing this.

Comment 3

[BugBot [:suhaib / :marco/ :calixte]]

Set release status flags based on info from the regressing bug 1718719

Comment 4

[Marco Castelluccio [:marco]]

(In reply to Dragana Damjanovic [:dragana] from comment #2)

This is caused by our es setup that uses a proxy. The most majority of users does not use a proxy and will not experience the regression.

I will take a look what is really causing it.
The patch only access nss to disable the TLS early-data that will not be used anyway. I am wondering why it is causing 10ms delay at all. Maybe it is nss locks that are causing this.

Dragana, did you have time to look into this?

Comment 5

[Ryan VanderMeulen [:RyanVM]]

This went back to pre-regression values when bug 1713350 landed and has been stable ever since. I think that's probably the best we can ask for at this point.