Update was off, something turned it on. Then the computer crashed in sleep
Categories
(Toolkit :: Application Update, defect)
Tracking
()
People
(Reporter: steevo, Unassigned)
References
Details
User Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:77.0) Gecko/20100101 Firefox/77.0
Steps to reproduce:
I had auto updates off, set to notify only. I was receiving constant annoying update nags.
I turned auto updates off because when I got an auto update and firefox restarted I had a significant delay, "Firefox is updating and will start in a few moments."
This is because of very poor design of that update notification system I had to turn it off.
So yesterday, I got a failure to resume from sleep, and I had to crash the computer with the power button and restart. Moving the mouse or tapping on the keyboard would not resume it though it was clearly on but was sending no video.
On restart I got the "Firefox is updating" popup. I thought, that's odd, auto updates were off.
I looked in even viewer to see if there are any interesting events, but the only interesting events were these:
The system firmware has changed the processor's memory type range registers (MTRRs) across a sleep state transition (S4). This can result in reduced resume performance.
Actual results:
The computer would not resume from sleep, and when I restarted it I had the notice that Firefox is updating and will start in a few moments.
I suspect the failure to resume from sleep was related to the Firefox update, which I had turned off. But something turned it on.
Expected results:
Firefox should not update automatically when updates are off.
Firefox update should not prevent resume from sleep.
|
||
Comment 1•5 years ago
|
||
Bugbug thinks this bug should belong to this component, but please revert this change in case of error.
|
|
||
Comment 2•5 years ago
|
||
The severity field is not set for this bug.
:nalexander, could you have a look please?
For more information, please visit auto_nag documentation.
|
||
Comment 3•5 years ago
|
||
Steve, thanks for the report. (And sorry for the long delay getting back to you for more information.)
We need some more information to have a better idea of what might have happened. First, it's not clear to me that auto-updates are re-enabled. The way that Firefox updates is (usually) to download an update, "stage" it, and then on the next restart to move the "staged" update into the main location and us that. (That's what causes the delay, generally -- it's just how long it takes to move the files around on your device.) It rather sounds like you had an update downloaded (or downloading, which then finished and was downloaded) when you flipped the auto-enable switch. Then, potentially much later, the computer restarts and Firefox restarts, completing that pending update.
Could you verify whether the auto-updates toggle is on for me? There's also an "Update History" log in about:preferences. This was such a long time ago, but could you tell me whether you've seen lots of updates in the last 3 weeks, or just the one unexpected one?
Finally, the issues with resuming from sleep don't sound related to Firefox updates. They could be, of course -- but I have no plausible explanation for an interaction as yet.
Thanks!
- My about firefox says auto updates are disabled by your administrator (me) by the registry fix.
BTW, I do consider that an inappropriate method of stopping auto updates because it also prevents any manual updates. That is bad architecture. Disabling auto updates should not prevent manual updates. That was a bad idea. - I don't see the update history log. Can you be more clear, and I can post it.
- I have had instances where this Windows 10 Dell system fails to resume from sleep, I just noticed the commonality this one time when the firefox update. I have no proof at all that it is related, but admittedly the firefox update sometimes takes so long it must be consuming resources.
|
|
||
Comment 5•5 years ago
|
||
Sorry for the long delay in replying -- busy, busy!
(In reply to Steve from comment #4)
- My about firefox says auto updates are disabled by your administrator (me) by the registry fix.
OK.
BTW, I do consider that an inappropriate method of stopping auto updates because it also prevents any manual updates. That is bad architecture. Disabling auto updates should not prevent manual updates. That was a bad idea.
If you opt out of auto-updates, you're responsible for updating. The policy setting (registry value) is for administrators who are maintaining machines in some other way -- say, by re-imaging them on some cadence.
- I don't see the update history log. Can you be more clear, and I can post it.
If you navigate to about:preferences and then search for "updates", you should have a button that says "Show Update History...".
- I have had instances where this Windows 10 Dell system fails to resume from sleep, I just noticed the commonality this one time when the firefox update. I have no proof at all that it is related, but admittedly the firefox update sometimes takes so long it must be consuming resources.
The time to apply an update is generally a function of your storage device's speed, so there's probably not much to be done here. It's possible that load on your storage device interacts with the sleep subsystem but as you say there's nothing strong linking the two here.
BTW, I do consider that an inappropriate method of stopping auto updates because it also prevents any manual updates. That is bad architecture. Disabling auto updates should not prevent manual updates. That was a bad idea.
If you opt out of auto-updates, you're responsible for updating. The policy setting (registry value) is for administrators who are maintaining machines in some other way -- say, by re-imaging them on some cadence.
Well, no, because you cannot update. There are only two settings in prefs
- Automatic updates ~ which I found unsatisfactory. I choose when to install my updates and sometimes I am on a worse connection and choose to wait for a better one.
- Download updates, but let me choose when to install them. ~ which is also unsatisfactory because this results in a constant nag to install the update it has downloaded.
There should be a third setting, to update manually, but there is none.
There is a manual update button in about but you have to leave the constant nag on the screen with the second setting to use it. The only way to get rid of the nag is to disable updates in the registry which incorrectly disables manual updates on the about page. That's what's wrong architecture here and that should be fixed.
The registry setting should not disable manual updates if that's the best we can get, but a better solution is a third update setting in prefs to update manually. A user ought not have to use regedit to get rid of a constant nag. In fact, the nag should be dismissed and should not return, maybe ever or after the next update. The way this is, well, it's just asinine. Poor architecture. Sending people constant nags is not OK. Forcing automatic updates is not OK. That the only recourse is an administrator setting is not OK.
I see the log. It was in April, 68.7.0. If it were copyable I would have pasted it but it's not. The update log should be copyable if you want to past it here.
|
||
Comment 7•5 years ago
|
||
Hi Steve,
Thank you for your input. I'm sorry to have to report that we no longer support manual updates in the way you describe. It is simply too easy and too dangerous for a user to request manual updates and then forget to ever update. We want to support users in their choices, but we also have an obligation to keep our users safe. This includes users that don't fully understand the extensive ramifications of disabling browser updates. A slightly more detailed explanation of this decision exists here.
And while the enterprise policies can disable update, they are intended for enterprise environments. In an enterprise environment, if the administrator deems that we should not update, we can't provide an update button.
I'm really sorry that the available options do not quite meet your needs. I've filed Bug 1653430 to consider adding an enterprise policy more like what you have in mind, but I cannot guarantee it will be added.
Oh, I know, I have users I put on auto updates without hesitation. But I am a power user so I don't do it for myself. I found it too annoying. The system would update when it was inconvenient for me. Firefox would do that, update on startup and it would make me sit there and wait.
I also find the incessant nags too annoying, which is what I originally complained about.
Why anyone would think subjecting users to constant nags is productive, I can't imagine.
But then, the whole idea of Mozilla was always to empower the user.
If I wanted someone else to make all my decisions for me I would just use Microsoft products.
|
||
Comment 9•5 years ago
|
||
I'm closing this bug; there's not enough information here to go on, and the discussion became unproductive. Please feel free to comment again if the same issue reoccurs.
Description
•