When updating on Mac OS X and elevation is required to update the client will not be able to update. The elevation UI is displayed but clicking the restart button only dismisses the UI instead of restarting and restarting just displays the elevation UI again.

This affects Nightly but it doesn't affect Firefox 66.x and I haven't checked Beta yet.

I'll look into this more tomorrow.

[Tracking Requested - why for this release]: macOS clients can end up in an update loop.

Last working build 20190416095014
https://hg.mozilla.org/mozilla-central/rev/a17f163841e956c10eb74cdbe27d7073facf6b7f
https://archive.mozilla.org/pub/firefox/nightly/2019/04/2019-04-16-09-50-14-mozilla-central/

First affected build 20190416220148
https://hg.mozilla.org/mozilla-central/rev/12a60898fdc12e4ab8735f228a77d6c1e6a600a9
http://archive.mozilla.org/pub/firefox/nightly/2019/04/2019-04-16-22-01-48-mozilla-central/

Relevant pushlog
https://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=a17f163841e956c10eb74cdbe27d7073facf6b7f&tochange=12a60898fdc12e4ab8735f228a77d6c1e6a600a9

This is quite possibly due to bug 1541062

I backed out bug 1541062 locally and confirmed that this bug was regressed by bug 1541062. One of the difficulties with this code is that it uses the OS provided ability to elevate and it does work during startup so it is to say the least difficult to test properly which is one of the reasons a test didn't catch this.

I think the best path forward for this is to back out bug 1541062 on both beta and nightly. I found this bug while trying to finish up bug 1515484 which also fixes bug 1541062 since it removes the code bug 1541062 modifies.

Jonas and Gijs, any objection to backing out bug 1541062 from beta and nightly and letting bug 1515484 take care of bug 1541062?

Just in case, I asked in #release-drivers whether 68 which has the patch for bug 1541062 will somehow make it to 67.0.5 and I've been told that it won't so marking 67.0.5 as unaffected

(In reply to Robert Strong (Robert they/them) [:rstrong] (use needinfo to contact me) from comment #4)

I backed out bug 1541062 locally and confirmed that this bug was regressed by bug 1541062. One of the difficulties with this code is that it uses the OS provided ability to elevate and it does work during startup so it is to say the least difficult to test properly which is one of the reasons a test didn't catch this.

I think the best path forward for this is to back out bug 1541062 on both beta and nightly. I found this bug while trying to finish up bug 1515484 which also fixes bug 1541062 since it removes the code bug 1541062 modifies.

Jonas and Gijs, any objection to backing out bug 1541062 from beta and nightly and letting bug 1515484 take care of bug 1541062?

I don't understand why we need to back out from Nightly if bug 1515484 removes the code anyway. Can we actually hit this in practice still, if it's being removed?

I think backing out from beta is fine, but I expect it'll need a rebased patch because of the pref with a list of exceptions.

(In reply to :Gijs (he/him) from comment #6)

(In reply to Robert Strong (Robert they/them) [:rstrong] (use needinfo to contact me) from comment #4)

I backed out bug 1541062 locally and confirmed that this bug was regressed by bug 1541062. One of the difficulties with this code is that it uses the OS provided ability to elevate and it does work during startup so it is to say the least difficult to test properly which is one of the reasons a test didn't catch this.

I think the best path forward for this is to back out bug 1541062 on both beta and nightly. I found this bug while trying to finish up bug 1515484 which also fixes bug 1541062 since it removes the code bug 1541062 modifies.

Jonas and Gijs, any objection to backing out bug 1541062 from beta and nightly and letting bug 1515484 take care of bug 1541062?

I don't understand why we need to back out from Nightly if bug 1515484 removes the code anyway. Can we actually hit this in practice still, if it's being removed?

It has more to do with how badly users break with this bug along with the time it will take to finish removing the UI but I'm ok with not backing it out on Nightly.

Gijs, will you handle the backout of bug 1541062 on Beta?

Beta/Release Uplift Approval Request

• User impact if declined: See comment #0 - broken updates!
• Is this code covered by automated tests?: No
• Has the fix been verified in Nightly?: No
• Needs manual test from QE?: No
• If yes, steps to reproduce:
• List of other uplifts needed: n/a
• Risk to taking this patch: Low
• Why is the change risky/not risky? (and alternatives if risky): It's a backout
• String changes made/needed: n/a

Attachment #9067422 [details] [diff] looks like the wrong file?

https://hg.mozilla.org/releases/mozilla-beta/rev/8a9ead04803465749d2cb41a9ca00cc72a050815

Gah, sorry, and thanks for sorting this out!

https://hg.mozilla.org/releases/mozilla-beta/rev/6ec3cb391f03

This is fixed via bug 1515484. There are other pre-existing issues with elevating on Mac that will hopefully be taken care of in bug 1521632

This bug has been identified as part of a pilot on determining root causes of blocking and dot release drivers.

It needs a root-cause set for it. Please see the list at https://docs.google.com/document/d/1FFEGsmoU8T0N8R9kk-MXWptOPtXXXRRIe4vQo3_HgMw/.

Add the root cause as a whiteboard tag in the form [rca - <cause> ] and remove the rca-needed keyword.

If you have questions, please contact :tmaity.

Hey Emma, can we kick this bug out of the RCA pilot? Potentially we could come back and do this in six months or so. We're really busy on other important work so we really don't have bandwidth for this.

Tanya, okay to remove this from the pilot?

(In reply to Emma Humphries, Bugmaster ☕️🎸🧞‍♀️✨ (she/her) [:emceeaich] (UTC-8) needinfo? me from comment #18)

Tanya, okay to remove this from the pilot?

Yes Emma, let's remove this one from the pilot for now.

Please specify a root cause for this bug. See :tmaity for more information.

As I'm on triage rotation I've been asked to specify a root cause here. I don't know why bug 1541062 broke elevated updates on macOS, and because the code was removed rather than fixed I don't think that analysis was ever done.

But it's pretty clear that the reason that this regression made it to beta was because of insufficient testing. The tests for this are perhaps impossible to automate, and we haven't established a schedule of manual tests to be run regularly, as far as I know.