1411304 - Fix sccache hit rate perfherder metric to use lowerIsBetter

Status: RESOLVED FIXED

(Firefox Build System :: General, defect, P2)

Description

[(not currently active) Ted Mielczarek]

We submit sccache stats to perfherder:
https://dxr.mozilla.org/mozilla-central/rev/ce1a86d3b4db161c95d1147676bbed839d7a4732/testing/mozharness/mozharness/mozilla/building/buildbase.py#1938

In bug 1411212 we saw:
Improvements:

100%  sccache hit rate summary windows-mingw32-32 debug taskcluster-m4.4xlarge     1.00 -> 0.00

...when this is actually a regression. Apparently we need to specify `lowerIsBetter: true` in the perfherder suite to make this work properly.

Comment 1

[Mike Hommey [:glandium]]

(In reply to Ted Mielczarek [:ted.mielczarek] from comment #0)
> We submit sccache stats to perfherder:
> https://dxr.mozilla.org/mozilla-central/rev/
> ce1a86d3b4db161c95d1147676bbed839d7a4732/testing/mozharness/mozharness/
> mozilla/building/buildbase.py#1938
> 
> In bug 1411212 we saw:
> Improvements:
> 
> 100%  sccache hit rate summary windows-mingw32-32 debug
> taskcluster-m4.4xlarge     1.00 -> 0.00
> 
> ...when this is actually a regression. Apparently we need to specify
> `lowerIsBetter: true` in the perfherder suite to make this work properly.

O_o but lower is *not* better.

Comment 2

[Ionuț Goldan [:igoldan]]

Attachment: Bug 1411304 - lowerIsBetter not true for sccache hit rate

Comment 3

[Joel Maher ( :jmaher ) (UTC -8)]

Comment on attachment 9003046 [details]
Bug 1411304 - lowerIsBetter not true for sccache hit rate

Joel Maher ( :jmaher ) (UTC-4) has approved the revision.

Comment 4

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/ed26ed848a8e
lowerIsBetter not true for sccache hit rate r=jmaher

Comment 5

[Noemi Erli[:noemi_erli]]

https://hg.mozilla.org/mozilla-central/rev/ed26ed848a8e

Comment 6

[Tom Prince [:tomprince]]

https://hg.mozilla.org/releases/mozilla-esr60/rev/8a07d9ed34d3