1353773 - Firefox does not do update when "relaunch to update" clicked

Status: RESOLVED WORKSFORME

(Toolkit :: Application Update, defect)

Description

[Dan Pernokis]

User Agent: Mozilla/5.0 (Windows NT 6.1; rv:52.0) Gecko/20100101 Firefox/52.0
Build ID: 20170323105023

Steps to reproduce:

Update mode set for manual.  Check HELP/ABOUT and run update if available.  Click RELAUNCH to restart FF and apply the update.


Actual results:

Update is downloaded OK.  RELAUNCH button is clicked.  FF window closes and re-opens.  No evidence of update running (no popups, scans, notices, whatever usually happens on screen *does not* happen).  Check HELP/ABOUT and notice version has not changed.

Have to manually close FF and manually restart.  Then update runs normally (popups, scans, notices, whatever usually happens happens), FF window opens, and HELP/ABOUT shows new version.  

First noticed several updates ago (at least back to FF 51), but always thought it was a procedural problem (like having two or more instances running (FF windows open).  At least a few times on laptop, FF did close but did not relaunch -- I had to do that manually.  When it happened on laptop this time, I documented what happened, then immediately repeated on tower -- expecting and getting exactly the same issue.



Expected results:

When RELAUNCH button is clicked, FF should properly close, apply pending updates, and relaunch automatically.

Comment 1

[Dan Pernokis]

Clarifications:

>> Have to manually close FF and manually restart.
I mean "relaunch FF" and not reboot/restart Windows.

>> When it happened on laptop this time...
I was upgrading to Version 52.0.2 from 52.0.1

Comment 2

[Robert Strong (they/them - no direct email)]

Hi Dan, this sounds like there is something preventing Firefox from exiting during restart. Please try updating using safe mode.
https://support.mozilla.org/t5/Procedures-to-diagnose-and-fix/Troubleshoot-Firefox-issues-using-Safe-Mode/ta-p/1687

Comment 3

[Dan Pernokis]

Hi Robert...

I was planning to regress FF to 52.0.1 (version immediately previous to the update) with the hopes of repeating the failure (or not).  I was going to regress, update normally (to ensure positively it would fail, so I'm not chasing a negative), regress again, then update in safe mode per your request.

I DL'd FF 52.0.1 (full setup) and had trouble running it (nothing happened) -- turns out I had a weird issue with Norton Security which reboot resolved.  So the regress worked OK after that and launched FF for me (not in safe mode).  And as soon as I clicked HELP/ABOUT, checked for update (52.0.2 pending), and clicked UPDATE -- it ran OK, relaunched OK, displayed the usual things, and updated OK, all without further intervention.

So the relaunch happened OK in regular (not safe) mode.  The problem might have been due to the Norton issue, but not on the laptop -- Norton was running OK at the time (I had just done a routine manual update).  I suspect more likely there is/was some driver/helper thing in memory at the time (Flash, for example) that wasn't there right now for the test but is present often enough to cause the failure to relaunch on a regular basis.



BTW, I checked some previous bug notes.  You & I had this failure happen on FF 49.0.2 as well (Bug 1312738), where you had me regress to 49.0.1 a few times -- but I can't really say it has happened often since, until the last few updates.  There was at that time also my bug 1303187, where a small FF stub would remain in memory after closing FF, then sometimes causing manual relaunches to fail (close FF and relaunch from icon).  Later there could be multiple stubs (3-4k or so typical size).  This issue seemed to resolve itself through a combo of FF and Norton updates -- I only rarely see it happen now.

I don't know whether there was a stub remainder that caused the new failure, but I had been in and out of FF many times, long time (2 weeks?) since last documented reboot, visiting one or two sites known to have "caused" the multiple stubs in the past.  Yet I have not had any FF popups to say it cannot start because it's already running, which I likely would have had.

>> sounds like there is something preventing Firefox from exiting during restart
Let me just say that, while troubleshooting bug 1303187, FF would regularly take many seconds to clear itself out of memory when closed, and it would then often refuse to start if it detected any stubs or artifacts remaining.  It might be that the timing is too tight in the update relaunch attempt, such that the old FF doesn't vacate quite fast enough and the updater doesn't get triggered.

Comment 4

[Robert Strong (they/them - no direct email)]

If it happens again before updating please either comment in this bug or file a new bug so we can gather some logs to check if it is timing out as you suspect. Thanks

Comment 5

[Dan Pernokis]

Will do.  Thanks.

Comment 6

[Dan Pernokis]

Robert, today I ran the newest update without any problems.

I successfully updated (upgraded) to the new FF 53.0 (32-bit) on two machines (tower and laptop) running Windows 7/32.
In both cases, when the time came to relaunch FF there was a momentary Windows wait circle for about 5 seconds.  I waited for the circle to disappear, the clicked to relaunch.  FF quit and relaunched OK, the update process appeared to be normal, and HELP/ABOUT indeed showed that the new version 53.0 was in and running.

So... It may just be as I suspected originally -- that I had too quickly clicked relaunch on previous occasions (ie- during the wait circle) and perhaps FF hadn't been fully ready to go down at that point.  See Comment 3 - Last half, "BTW..." (especially last paragraph).

Thanks again...