Screen saver wakes during night

[mellobob]

I've found that the screen saver is occasionally waking up during the middle of the night on my computers.

At first this was happening on my main computer which was running cinnamon. I originally thought it might be due to backintime running, but stopped that and it continued. So, then I figured it might be plex-media-server. Stopped that as well ... but the computer kept waking up in the night. So, I changed from cinnamon to mate and all has been fine since.

But, the last few days has another computer in my house (the one my wife uses) doing the same thing. And this one has always run Mate. It dutifully goes to sleep around 11PM when the last check of facebook is done. But, it seems to wake up around 1 or 2am (don't know exactly since I'm supposed to be sleeping as well).

On this computer the only things running are a grive update at 1:30am and backintime at 3am. I'm pretty sure the wake up is happening around 2am, but not sure. And these cron items have been running for months without issue.

Oh, and no cats to walk over keyboards either

Anyone else with similar issues? Anyway to track what is happening to cause the wake?

I'm running Linux Mate 18.2 with kernel 4.10.0-35.39~16.04.1-generic 4.10.17.

Thanks.

[Spearmint2]

go into BIOS and turn off WOL (wake on LAN). If that doesn't and you have pets, then also turn off the other "wake on" functions.

[mellobob]

Thanks for the suggestion. Never consider the WOL as a culprit.

I turned it off yesterday on the misbehaving machine and it did sleep the night. We'll monitor for a bit longer before we give the problem a <solved> label.

But, question: Is this a bug in the BIOS, kernel or what? I've run machines for years with WOL enabled (even though I have never used the function) and never had this issue before. And even if the port does get a magic packet telling it to wake, should it not go back to sleep after the set timeout? Inquiring minds want to know

[Spearmint2]

Thanks for the suggestion. Never consider the WOL as a culprit.

I turned it off yesterday on the misbehaving machine and it did sleep the night. We'll monitor for a bit longer before we give the problem a <solved> label.

But, question: Is this a bug in the BIOS, kernel or what? I've run machines for years with WOL enabled (even though I have never used the function) and never had this issue before. And even if the port does get a magic packet telling it to wake, should it not go back to sleep after the set timeout? Inquiring minds want to know

To have it go back to sleep after a while you need to set that in the power settings, which you probably have set to NEVER. Windows co-opts the ACPI from the BIOS, but not sure if Linux does the same. Good questions!

[mellobob]

I usually have the <put computer to sleep> to NEVER and the <put display to sleep> to 30 minutes.

So, what you are saying is that WOL will wake the computer (which mine wasn't) and as a result of that the display gets woken as well? Perhaps the display waking is a side effect of the computer waking (even if it isn't sleeping). And then the computer doesn't know to turn the screen off again?

All very confusing. Plus, where is that darned WOL magic coming from? Maybe Facebook is sending this out just because

[Spearmint2]

That's it. If you change power settings to sleep after 30-40 minutes, you should be OK in both situations.

[all41]

fwiw:
I use WOL on my network and have never had a rude or unexpected awakening.
WOL will turn remote computers on, wake them from suspend, or wake them from hibernate.
WOL does not effect computers when their screensaver is running in any manner

[mellobob]

That's it. If you change power settings to sleep after 30-40 minutes, you should be OK in both situations.

Yes, but putting the computer to sleep ... doesn't that stop things like cron and other midnight tasks from running?

[all41]

Suspend (to ram) and Hibernate (suspend to disk) differ from Screensaver.
So this may just be terminology.
Screensaver only blanks the screen, or substitutes alternate graphics, but the entire system is still operational under full power.
Suspend copies the entire system to ram for later recall and retains only enough power to keep those ram memories intact.
Hibernate copies the entire system to the HD and shuts the computer entirely off.
In your bios settings under Power settings is your system perhaps set to resume after power failure?

[trytip]

why not just turn off the computer before you bed. no sense in it sleeping and consuming power unless you want to keep somethings in sync.

[sgtor]

In the summer time when I had a fan running very far away from the computer during sleep it would actually move the mouse just enough to wake the computer. And I mean a miniscule amount so be sure to check things like that if it keeps happening, even a breeze from an open window.

I started turning the mouse off, problem went away.

[Spearmint2]

In the summer time when I had a fan running very far away from the computer during sleep it would actually move the mouse just enough to wake the computer. And I mean a miniscule amount so be sure to check things like that if it keeps happening, even a breeze from an open window.

I started turning the mouse off, problem went away.

That can happen if the battery in mouse is low too, because it pops up a warning and can wake computer up.

[mellobob]

Just a bit of summary:

- mouse is not moving. The room has no windows or even a heat vent.
- I too had WOL enabled for years. This is a new problem on 2 different computers.
- computer 1 solution was to switch from Cinnamon to Mate
- computer 2 solution was to disable WOL
- both computers have blank screen set for 20 minutes and "sleep" never.
- turn off the computer at night? Really? People do that?

Anyway, it seems to be okay now. If the little guy starts to wake up again I'll have a serious chat with it and let you folks know.

[Spearmint2]

Speaking of nightime. Puts a button at end of URL toolbar, only needs a clk to change between day and night mode. Works in FF57.

https://addons.mozilla.org/en-US/firefo ... e/?src=api

what it looks like on this forum.

[sgtor]

In the summer time when I had a fan running very far away from the computer during sleep it would actually move the mouse just enough to wake the computer. And I mean a miniscule amount so be sure to check things like that if it keeps happening, even a breeze from an open window.

I started turning the mouse off, problem went away.

That can happen if the battery in mouse is low too, because it pops up a warning and can wake computer up.

I actually never would of thought of that, thanks for the tip.

[mellobob]

So, learned one thing: It's not a problem with WOL. The miscreant woke up last night.

Next thing to try: I've installed (well, plugged in!) a different keyboard. We'll see how this manages.

BTW, I've got no fans blowing, no batteries in my keyboard or mouse. The computer is plugged into UPS.

Is there a logging I could enable or examine which lists time and event for wake?

[sgtor]

So, learned one thing: It's not a problem with WOL. The miscreant woke up last night.

Next thing to try: I've installed (well, plugged in!) a different keyboard. We'll see how this manages.

BTW, I've got no fans blowing, no batteries in my keyboard or mouse. The computer is plugged into UPS.

Is there a logging I could enable or examine which lists time and event for wake?

Actually give this a try.

Code: Select all

journalctl | grep -i wake | less

I'm not sure how informative it will be but it's all I can think of off the top of my head.

The crazy thing is I used to have this problem in windows and every time I would turn off the wake settings for the culprit program or device it would get set to wake again after an update etc, etc.

With some research I found a command to run in dos that showed anything that could wake the computer and if it was set to wake or not.
There must be some way to do that in Linux but I don't know how since I've never had to research it since I've switched to linux.
Maybe someone else knows how.

Edit:
Actually now that I think of it you should run dmesg. that should give you some good info too if you run it as soon as you see the computer has woken up.

[mellobob]

Okay ... it's not the keyboard.

I'm just going to leave this for a few more days until I get updated to 18.3. Hopefully that will help

One thing I have discovered is that the timing seems to be connected to the running of backintime. Looking at the syslog I see that it does a:

backintime (val/1): INFO: Inhibit Suspend started. Reason: take snapshot

at the start of a run and

mellownet backintime (val/1): INFO: Release inhibit Suspend

at the end. Makes sense to inhibit suspend while doing a backup, and I don't know if there is a connection between releasing this and X screen displays.