Hi,
At the second minute of hours 21,22,23,00 (±2), my pc completely freezes for about 5 seconds. This shitty behavior has caused me plenty of rage, and sometimes data and time loss when doing sensitive tasks that don’t handle freezes. It’s been going on for months.
Can someone suggest a good way of finding out what is causing it, and how (full ram, cpu usage)?
Manjaro Cinnamon (5.17.1-3-MANJARO).
Run memx86 test? Reseat the memory / cpu with repaste. What are you running / doing when on the system when it happens?
Usually under windows I’d have suggested updates and a reboot, but I’m not sure if the same mantra applies to linux.
Also have you run the disk diags i.e. seatools for segate drives / samsung magician for samung ssds.
Check to see if you have any backups, snapshots, updates, etc happening at that time. Sorry, I don’t know Manjaro enough to give anything more specific.
For Windows users I have seen where the File history function (copy of files to a different drive) slows the machine to a crawl even with separate SSD drives for the source/target.
I manually install updates, so that’s a no-go (snaps do update automatically, but I highly doubt that the few I have update that frequently).
I do have daily snapshots. They are done through timeshift, but they are btrfs snapshots, so they are instantaneous, and after they are done once in the day, so they don’t repeat again till tomorrow.
What’s annoying about this issue is that it keeps repeating like its actually doing something every time it freezes up.
I’m going to contribute absolutely nothing here, but ask this question:
What time continuum are you from? That line triggers no comprehension sensors in my mush-brain.
Best I could interpret was the second labeled only “2” after the hour. Ex: 00:00:02, 20:00:02, …
But this is a case of bad English; both in the sense that,
the author could have written it better (bad English)
the language could be better (Bad, English!)
I would be tempted to say, “the third second of the first minute”, but if you are unsure if I am counting from zero, that could sound like: _:01:03, _:00:03, or _:01:02; when what I am trying to convey is _:00:02.
Regardless, @gzgzone an example of a particular time would be appreciated.
Regarding the issue though,
if it is occurring that close to the top of the hour, my guess would be that some sort of scheduled process is at fault; have you checked cron or SystemD or other utilities that could be scheduling processes to ein at particular times?
Also, you could watch what processes are running when this occurs, using a commandline tool like top or htop, or a GUI System/Activity/Task Monitor/Guard/Manager utility (on Cinnamon this is probably System Monitor).
There is also dmesg, which gives you error messages from the Linux kernel itself. If this is a scheduled process issue like I suspect, this will probably show nothing, but it is good to check regardless.
A privacy consideration
dmesg could contain serial numbers, MAC addresses, and IP addresses that you may not want to share publicly; if you need assistance interpreting the output, I would wait for a freeze to occur, then run dmesg and only copy the last few lines.
There is also a flag (-w I think, check the man page) that will keep dmesg running and show you errors/warnings/notices as they appear.
My bad. I meant the second minute (example 21:02:??).
I’ve already checked all cron entries, and none are scheduled for that time. Dmesg wasn’t helpful either. Htop freezes, as well, so that’s a nogo. I’ve tried reproducing the problem by manually changing system time, but to no avail. I’ll try top or something else next time before this problem comes up.
Are you running this from a graphical desktop, or a TTY? If the GUI is what is freezing, there is a chance that watching a TTY while this happens could be informative.
Alternatively, you could set the sample rate on htop or top very high (be aware the configuration refers to this setting as delay not sample time). If you have columns sorted by thread state or CPU, then as it freezes it might let you see what happened right before. The default time between samples is 3.0 seconds; maybe try .5 or .1?
Also, turn on System Monitor and leave open the Resources tab, so that after the freeze you can check the CPU and memory graphs for any spikes. Be aware that it only contains a minute of graph history, so check on it soon after the freeze.