1219751 - Change the the depth limit of the thread hangs stack to use the 99th percentile

Status: RESOLVED FIXED

(Toolkit :: Telemetry, defect, P1)

Description

[Alessio Placitelli [:Dexter]]

We should change the maximum length of the thread hangs stack to the value of the 99th percentile rather than the 95th percentile as the most interesting stacks tend to be longer.

We need to run an updated spark job [0] to find the 99th percentile and then change the constant in [1].

[0] - https://gist.github.com/Dexterp37/fc4a043fb442d4e4be90
[1] - https://dxr.mozilla.org/mozilla-central/rev/1e700005a0ddf2b17803213e1f3f8d78a7a618b8/xpcom/threads/BackgroundHangMonitor.cpp#36

Comment 1

[Roberto Agostino Vitillo (:rvitillo)]

Let's set the limit to 30 (99.9%).

Comment 2

[Alessio Placitelli [:Dexter]]

Attachment: bug1219751.patch

Comment 3

[Alessio Placitelli [:Dexter]]

https://treeherder.mozilla.org/#/jobs?repo=try&revision=87cdd3ba280d

Comment 4

[Alessio Placitelli [:Dexter]]

https://hg.mozilla.org/integration/fx-team/rev/134bb8fdb24418730e1ed5940fafc42cdfc5e972
Bug 1219751 - Change the the depth limit of the thread hangs stack to use the 99th percentile. r=gfritzsche

Comment 5

[Wes Kocher (:KWierso) (Not reading bugmail; email directly if needed)]

https://hg.mozilla.org/mozilla-central/rev/134bb8fdb244

Comment 6

[Carsten Book [:Tomcat]]

https://hg.mozilla.org/releases/mozilla-b2g44_v2_5/rev/134bb8fdb244

Comment 7

[Carsten Book [:Tomcat]]

removing the b2g 2.5 flag since this commit has been reverted due to an incorrect merge, sorry for the confusion

Comment 8

[Vladan Djeric (:vladan)]

Can we uplift this to Aurora 44 & Beta 43?

See https://bugzilla.mozilla.org/show_bug.cgi?id=1211411#c11

Comment 9

[Alessio Placitelli [:Dexter]]

Comment on attachment 8680654 [details] [diff] [review]
bug1219751.patch

Approval Request Comment
[Feature/regressing bug #]: Tweaks the constant that controls the number of thread hang stacks sent.
[User impact if declined]: This isn't a user-facing feature. However, without this patches bug 1222894 could potentially cause sending a lot of oversized pings, wasting both storage and processing time server side, thus increasing costs.
[Describe test coverage new/current, TreeHerder]: This has been on m-c for 2 weeks without any reported issue.
[Risks and why]: Low risk, the changed constant controls how many stacks we are sending.
[String/UUID change made/needed]: None.

Please note that this is part of an uplift stack required for bug 1222894. This is the stack (the patches correctly apply in order to mozilla-beta): 1213780, 1211411, 1215540, 1219751.

Comment 10

[Alessio Placitelli [:Dexter]]

Comment on attachment 8680654 [details] [diff] [review]
bug1219751.patch

This also needs to be uplifted on Aurora: the rest of the stack is already there and it's just missing this one patch.

Comment 11

[Liz Henry (:lizzard) (relman/hg->git project)]

Marking this as affected for 43 and 44 since we intend to uplift.

Comment 12

[Liz Henry (:lizzard) (relman/hg->git project)]

Comment on attachment 8680654 [details] [diff] [review]
bug1219751.patch

Please uplift this patch to both aurora and beta.

Comment 13

[Vladan Djeric (:vladan)]

https://hg.mozilla.org/releases/mozilla-aurora/rev/8efe7a54bafe
https://hg.mozilla.org/releases/mozilla-beta/rev/6cabb1a43af6

Comment 14

[Carsten Book [:Tomcat]]

https://hg.mozilla.org/releases/mozilla-b2g44_v2_5/rev/8efe7a54bafe

Comment 15

[Carsten Book [:Tomcat]]

(In reply to Vladan Djeric (:vladan) -- please needinfo! PTO Nov 16/19/20 from comment #13)
> https://hg.mozilla.org/releases/mozilla-aurora/rev/8efe7a54bafe
> https://hg.mozilla.org/releases/mozilla-beta/rev/6cabb1a43af6

setting the flags for this checkin