Issue The swap usage threshold has been exceeded, every morning at 6:35

[PeopleInside]

Server operating system version

Ubuntu 22.04.3 LTS

Plesk version and microupdate number

Version 18.0.55 Update #2

My server was having no downtime issue for weeks, now on this days every morning at 6:35 I get the alert of swap usage threshold has been exceeded.


Looking at the Plesk Monitor I see an increase of the RAM use at 00:00

Seems this intense use persist until the downtime at the morning happen then the usage seems return to normal.

Disk usage:


Do you have any idea on how can understand what is wrong and how to avoid this downtime?

free -h


Thank you

 

[PeopleInside]

If I run the command:

cat /proc/sys/vm/swappiness

I get as result 60, I'm asking if I should try to set this value to 10.
I read this in a discussion here on Plesk and in a external article.

I'm unable to understand why memory usage seems high from the midnight to the morning 6:35. I see this in the graphic above.
Looks strange.

 

[PeopleInside]

Another strange thing that I'm unable to understand is: staying at Plesk Monitor graphic seems my RAM reached the 100 % or more then 7 GB when total RAM is 8 GB. Looking at hosting resource monitoring I don't find this high usage of the RAM.

Near the 6 of morning seems there is an high usage of the CPU and RAM but not since 00:00 as I can see in Plesk Monitor.


It would be nice if I can get some help to understand why all morning near the 6 and 35 I get this high usage and short downtime.
Maybe I set some settings in Plesk that help services to restart so the downtime is only two minutes instead to be permanent until I manually restart.

 

[PeopleInside]

In the Plesk log I can see this:

Can be this related? I don't know what can be the issue

 

[PeopleInside]

 

[PeopleInside]

The issue at the morning around 6 and 35 can be caused by the Plesk Daily Task

https://support.plesk.com/hc/en-us/articles/12377592733207

I'm not sure but checking on my server the task is around 6:25 AM

 

[mow]

It would be nice if I can get some help to understand why all morning near the 6 and 35 I get this high usage and short downtime.
Maybe I set some settings in Plesk that help services to restart so the downtime is only two minutes instead to be permanent until I manually restart.

As you can see in /etc/crontab, at 6:25 each script in /etc/cron.daily is started, which does updates, backup, and other plesk tasks.
You seem to also have a task like update-slocate running at 0:00 which does some disk I/O and fills the cache. That is not a problem, it's supposed to use otherwise free ram for cache as the cache can just be discarded when the ram is needed again. And it's probably why the hosting monitoring doesn't report anything.
When the 6:25 tasks eat up application memory, the cache is discarded. But it is not enough, it goes deeply into swap, and eventually starts killing processes when none is left.
Interesting here is however that according to the graphs it doesn't even reach the limits at 6:35, evident in that you still have cache after the event. Possible reasons:

• some program wants to allocate a huge amount of memory, most likely a bug or really bad programming
• the host itself is running out of memory because a lot of plesk instances all want to run their daily tasks at the same time

 

[PeopleInside]

As you can see in /etc/crontab, at 6:25 each script in /etc/cron.daily is started, which does updates, backup, and other plesk tasks.
You seem to also have a task like update-slocate running at 0:00 which does some disk I/O and fills the cache. That is not a problem, it's supposed to use otherwise free ram for cache as the cache can just be discarded when the ram is needed again. And it's probably why the hosting monitoring doesn't report anything.
When the 6:25 tasks eat up application memory, the cache is discarded. But it is not enough, it goes deeply into swap, and eventually starts killing processes when none is left.
Interesting here is however that according to the graphs it doesn't even reach the limits at 6:35, evident in that you still have cache after the event. Possible reasons:

• some program wants to allocate a huge amount of memory, most likely a bug or really bad programming
• the host itself is running out of memory because a lot of plesk instances all want to run their daily tasks at the same time

Thank you for the reply.