Cinnamon random lockups for 10-45 seconds

[spoo333]

Your computers are laughing at our attempts to control them.

it seems that way, doesn't it?

Generally I've found that when low-latency kernels are active, the lock-up issues don't seem to be quite as long and frequent, but this is purely my perception.

I am pretty much out of ideas of what might be happening. I'm sure there is a logical reason somewhere as to what is happening, but I don't have enough knowledge about the inner workings of the kernel to know what else might be a factor at this point.

I think the c-states are now disabled via kernel parameters in grub.

Screenshot from 2021-12-10 08-29-01.png

Over the last 14 hours or so I've only had two lock-ups/freezes, each lasting no longer than 10-15 sec. That's a definite improvement. Thanks for your suggestion re c-states.

Also, I've been trying to keep an eye on how, if at all, those regular updates of Cinnamon that keep coming via the pre-release ppa affect the responsiveness of the system. Before I deactivated the c-states, I thought I could see a difference in regularity and intensity of lock-ups based on the Cinnamon version. Not so sure now if I only imagined it ...
I'll keep observing

[spoo333]

Generally I've found that when low-latency kernels are active, the lock-up issues don't seem to be quite as long and frequent, but this is purely my perception.

Jinxed it. Just had a one minute lock-up with low-latency kernel and c-states deactivated. I'll try MATE for a while. That might provide further clues ...

[SMG]

I think the c-states are now disabled via kernel parameters in grub.
Screenshot from 2021-12-10 08-29-01.png

That appears to be the case. I checked my results and they are 9 so your computer is at the other end of the scale.

I'm starting to feel like there is another common factor we're missing, but since we're missing it, I'm not sure what it might be.

Maybe somehow the regulation of the power supply to the computers might somehow be a factor and because the work you are doing is so sensitive to changes that it is more noticeable? I'm trying to think of what else might be variable since it seems like there does not yet seem to be a pattern to when the freezes happen.

[spoo333]

That appears to be the case. I checked my results and they are 9 so your computer is at the other end of the scale.

Thanks for still being here

Yes I agree - we must be missing something .....

I disabled c-states on all three computers last night. My partner uses the generic kernel laptop for her daily work. Nothing fancy, generic kernel, Crossover Office with MS Office, Libreoffice, Waterfox browser, Evolution e-mail, Nemo, a bit of audio playback of files. Despite c-states being off, she experienced lock-ups today as usual. There is no special audio setting on her laptop, no low-latency setting etc of any kind. Timeshift is on. A couple of nfs4 shares to a NAS in fstab, but that shouldn't really be a problem because I've had countless Linux installations connect to a NAS via nfs4 for almost a decade.

The desktop computer I use is the only one of the three computers that was running Mint before v20. 19.3 was smooth and had the same Ubuntu Studio low latency tweaks activated. No real issues. What I don't remember is if 19.3 was installed on btrfs or ext4, also unsure if Timeshift was on (I think so). Any kind of power supply issue would have had to be there with 19.3 too, right? I don't remember having lock-up issues with Mint 20 or 20.1 - there were other issues but no lock-ups. I first became aware of the lock-up around end of June I think, possibly because I was trying out daily builds via the ppa. My partner started experiencing lock-ups from July/August. Could this have been caused by upgrading from 20.1 to 20.2? Or could it have been a kernel switch, from 5.4 to 5.10/11?

The only thing that all three computers have in common is that they all run off NVMes with btrfs and Timeshift activated.
Both laptops are encrypted and installed as per this guide https://mutschler.eu/linux/install-guid ... rfs-20-04/
Both laptops have swap files, the desktop has a swap partition.

Whilst I do time-critical work (ie record music), a lot of my usage is non time-critical: browsing, editing audio files, the usual office and graphics stuff. I still get those lock-ups with non time-critical work - but 10-45 sec enforced breaks shouldn't really happen, at least not on reasonable well-equipped computers, in particular if those computer(s) was/were running fine with a previous version.

I could try 5.4 again and see what happens.

Appreciate it that you haven't written me off as a lost cause yet

[SMG]

What I don't remember is if 19.3 was installed on btrfs or ext4

This might be the key, but I have no clue how a btrfs Mint install might differ from an EXT4 Mint install or even know how to compare them.

I went to your link regarding encrypting a btrfs install and found that person's Why I (still) like btrfs where it is mentioned, "The installers of Ubuntu, Linux Mint, Pop!_OS or ElementaryOS need manual adustment of configuration files and mount options are usually not set optimally for using btrfs on SSD or NVMe.".

Did you make any adjustments? That person mentions here changes to fstrim timer and tlp which does not seem like something which would lead to what you have noticed. (It also does not seem to fit what the person referenced on the other page; I would have expected more "configuration file adjustments" than that.)

I ran a search on freeze issues with btrfs, but those responses were all hard freezes which required a reboot to clear. And then I found this, Is Btrfs stable? with the short answer of "Maybe".

[spoo333]

This might be the key, but I have no clue how a btrfs Mint install might differ from an EXT4 Mint install or even know how to compare them.

It could be a btrfs v ext4 issue, however I'm pretty confident that I had Mint running on btrfs without lock-ups/freezes (before end of June this year).

Did you make any adjustments? That person mentions here changes to fstrim timer and tlp which does not seem like something which would lead to what you have noticed. (It also does not seem to fit what the person referenced on the other page; I would have expected more "configuration file adjustments" than that.)

Yes I made the suggested modifications. As above, first time I used that installation guide everything worked without freezes up until the end of June (and July/August on my partner's laptop)

I ran a search on freeze issues with btrfs, but those responses were all hard freezes which required a reboot to clear. And then I found this, Is Btrfs stable? with the short answer of "Maybe".

that's not super reassuring ... hoping that link is a little outdated. Years ago I experienced obvious problems with btrfs incl corrupted partitions etc. Nothing of the sort in a while. Plus Synology must have found it good enough to allow it on their NAS systems.

Anyway, on 5.4 now. I don't want to say to you that I haven't had a lock-up yet because whenever I do .....

[spoo333]

Anyway, on 5.4 now. I don't want to say to you that I haven't had a lock-up yet because whenever I do .....

Same lock-ups/temporary freezes with 5.4. Now trying Xfce

[SMG]

Same lock-ups/temporary freezes with 5.4. Now trying Xfce

Your equipment really should be using a newer kernel than 5.4. That kernel was known to result in hard freezes (unrecoverable) for people with CPUs of the era you have.

[spoo333]

Your equipment really should be using a newer kernel than 5.4. That kernel was known to result in hard freezes (unrecoverable) for people with CPUs of the era you have.

I agree, just wanted to eliminate as many possibilities as possible.

Been running Xfce for about 18 hours now. Have thrown everything I can at it. No hiccups of any kind, just smooth, uninterrupted flow - the way Cinnamon used to be in the past. And all that with cstates back on and kernel 5.13.0-22-lowlatency. And btrfs and Timeshift

Screenshot from 2021-12-12 00-37-06.png

BTW, have removed the daily builds ppa and downgraded packages

[SMG]

Been running Xfce for about 18 hours now. Have thrown everything I can at it. No hiccups of any kind, just smooth, uninterrupted flow - the way Cinnamon used to be in the past. And all that with cstates back on and kernel 5.13.0-22-lowlatency. And btrfs and Timeshift

You have jinxed it for sure.

[spoo333]

You have jinxed it for sure.

[spoo333]

Still going strong without any issues whatsoever ...

[spoo333]

All good still. Definitely the longest session without any lock-ups of any kind in a few months ...

[spoo333]

Still no lock-ups in Xfce. Nemo runs smoothly in Xfce too.

I could try yet another DE to see if I get lock-ups there? I I don't, then it seems likely that there is an issue with Cinnamon on my computer(s). If I do, then back to the drawing board ...

[SMG]

Still no lock-ups in Xfce. Nemo runs smoothly in Xfce too.

I could try yet another DE to see if I get lock-ups there? I I don't, then it seems likely that there is an issue with Cinnamon on my computer(s). If I do, then back to the drawing board ...

Did you end up trying MATE? You mentioned the possibility earlier, but I don't see where you mention results with it.

[spoo333]

Did you end up trying MATE? You mentioned the possibility earlier, but I don't see where you mention results with it.

I felt uncomfortable in MATE which is why I removed it again after an hour, but could try again.

[spoo333]

I felt uncomfortable in MATE which is why I removed it again after an hour, but could try again.

As in no technical issues, just didn't like the feel of it ...

[SMG]

I felt uncomfortable in MATE which is why I removed it again after an hour, but could try again.

I was just curious. I have only ever used Cinnamon although I know where the documentation is for the other two desktops and have helped others with issues on those desktops (moreso with Xfce than MATE).

[spoo333]

I was just curious. I have only ever used Cinnamon although I know where the documentation is for the other two desktops and have helped others with issues on those desktops (moreso with Xfce than MATE).

I see no other DE experiments then?

[SMG]

no other DE experiments then?

A computer is very much a tool to me, so unless I'm having issues then I just keep using what I have.

I've seen others do experiments with regards to memory between the three DEs, (those might have included CPU also) but I've not heard before this type of difference with stability. There are differences between the three with regards to windows managers and compositors. Perhaps that difference is showing itself in your testing, but that is just a guess on my part.