1144970 - (emu-x86-kk-tests) [meta] Enable emulator-x86-kk and tests on treeherder/TaskCluster

Status: RESOLVED WONTFIX

(Taskcluster :: General, defect)

Comment 1

[Thomas Zimmermann [:tzimmermann] [:tdz]]

What's the purpose of this bug report wrt bug 1107678?

Comment 2

[Jed Davis [:jld] ⟨⏰|UTC-7⟩ ⟦he/him⟧]

(In reply to Thomas Zimmermann [:tzimmermann] [:tdz] from comment #1)
> What's the purpose of this bug report wrt bug 1107678?

It looks like bug 1107678 is for the ARM KK emulator and this one is for the x86 KK emulator.  If so, it would probably make sense for this bug to depend on bug 1107678, because I'd expect most things that block emulator-kk would also block emulator-x86-kk.

Comment 3

[Thomas Zimmermann [:tzimmermann] [:tdz]]

Oh, right. m( I didn't read the summary attentively. Sorry.

Comment 4

[Hsin-Yi Tsai (she/her) [:hsinyi]]

Bug 1144944 enabled emulator-x86-kk on taskcluster, but try were all orange [1].

According to the error log, runtestsb2g.py was run with option '--emulator=arm' but looks that should be '--emulator=x86' My trial on [2] seemed support my guess. Let's investigate more.

[1] https://hg.mozilla.org/try/diff/0ef4b01700d0/testing/config/mozharness/b2g_emulator_config.py
[2] https://treeherder.allizom.org/#/jobs?repo=try&revision=0ef4b01700d0

Comment 5

[Hsin-Yi Tsai (she/her) [:hsinyi]]

(In reply to Hsin-Yi Tsai [:hsinyi] from comment #4)
> Bug 1144944 enabled emulator-x86-kk on taskcluster, but try were all orange
> [1].
> 
> According to the error log, runtestsb2g.py was run with option
> '--emulator=arm' but looks that should be '--emulator=x86' My trial on [2]
> seemed support my guess. Let's investigate more.
> 
> [1]
> https://hg.mozilla.org/try/diff/0ef4b01700d0/testing/config/mozharness/
> b2g_emulator_config.py

Sorry, the link should be [1] https://treeherder.allizom.org/#/jobs?repo=try&revision=18452606021a

> [2] https://treeherder.allizom.org/#/jobs?repo=try&revision=0ef4b01700d0

Comment 6

[James Lal [:lightsofapollo]]

@kli- Do you have time to take a look ? 

Ideally we could teach mozharness to detect the emulator type implicitly rather then requiring manually passing these types (doing this manually is not terrible but requires duplicating a bunch of configs ;/)

Note that @hsinyi's hack did not also hardcode reftests so we don't know how green they are... The early results here look really good though!

After looking through some logs it appears that the xpcshell tests in particular fail frequently due to emulator startup errors (or file pushing errors) which is probably an easy fix.

Comment 7

[Hsin-Yi Tsai (she/her) [:hsinyi]]

(In reply to James Lal [:lightsofapollo] from comment #6)
> @kli- Do you have time to take a look ? 
> 
> Ideally we could teach mozharness to detect the emulator type implicitly
> rather then requiring manually passing these types (doing this manually is
> not terrible but requires duplicating a bunch of configs ;/)
> 

Thanks for the suggestion. John from my team will help take a look and follow the above comment first. He is still working on another bug but could be switching to this soon.

> Note that @hsinyi's hack did not also hardcode reftests so we don't know how
> green they are... The early results here look really good though!
> 
> After looking through some logs it appears that the xpcshell tests in
> particular fail frequently due to emulator startup errors (or file pushing
> errors) which is probably an easy fix.

Comment 8

[Wesley Huang [:wesley_huang] (EPM) (NI me)]

[Tracking Requested - why for this release]:

Comment 9

[Dustin J. Mitchell [:dustin] (he/him)]

Edgar, I've assigned this to you as part of cleaning up this "General" component.  If this doesn't still need to be tracked, can we close?  Or if there's a better assignee, please reassign.

Comment 10

[Hsin-Yi Tsai (she/her) [:hsinyi]]

I am inferring we aren't going to fix this. 
Julie/Fabrice, please feel free to reopen if there's still need for CD.