Hi, I've been having the "Out Of Memory errors detected on your server" on and off for almost as long as I've had my server. Haven't been able to figure it out. Does get solved by restarting, but was hoping the community could help figure out the cause. As far as I can tell no services or other issues occur because of it. Thanks for your time.

 

See attached diagnostics.

mercury-diagnostics-20240714-1001.zip

I am having a similar "intermittent" issue. I'm not seeing any problems with the server, just the warning from "Fix Common Problems" plugin. When I first noticed it last year, I tried to upgrade my RAM to 48GB, but that didn't help. My Ram usage seems within reason, and cache fully utilizing the available RAM.

 

I can tell from my sys logs that the "used" dips, are not related to my OOM-reaper killing any tasks.

 

It seems to me that the cache isn't being culled properly or is being over-utilized, not leaving space for system usage when a demand comes. Is there a way to modify how much "free" RAM remains, rather than utilizing it for "cache"?

 

Wish I could be more help, Velourfogg. But maybe we can get a discussion started here?

 

JP

1 hour ago, velourfogg said:

Hi, I've been having the "Out Of Memory errors detected on your server" on and off for almost as long as I've had my server. Haven't been able to figure it out. Does get solved by restarting, but was hoping the community could help figure out the cause. As far as I can tell no services or other issues occur because of it. Thanks for your time.

 

See attached diagnostics.

mercury-diagnostics-20240714-1001.zip 211.96 kB · 2 downloads

 

 

Looks like back in April it may have been Frigate, along with whatever was spawning all of those Apache processes which killed off whatever  VM you were running

 

In may, Apache again spawning a bunch but also appears to be Frigate

 

Again in June, Frigate and a bunch of curl cron jobs?

 

Doesn't appear to be any since June, tho it will keep complaining until you restart and they're no longer in the logs

 

 

31 minutes ago, capnJiP said:

But maybe we can get a discussion started here?

You should start your own thread and add your diagnostics

1 hour ago, Michael_P said:

 

 

Looks like back in April it may have been Frigate, along with whatever was spawning all of those Apache processes which killed off whatever  VM you were running

 

In may, Apache again spawning a bunch but also appears to be Frigate

 

Again in June, Frigate and a bunch of curl cron jobs?

 

Doesn't appear to be any since June, tho it will keep complaining until you restart and they're no longer in the logs

 

 

Okay thanks very much. I appreciate the help. I'll keep an eye on Frigate, though it's been stable for a while. The VM runs Home Assistant.

1 hour ago, velourfogg said:

I'll keep an eye on Frigate

I don't run it personally, but quite a few users have issues with it running the host OOM

So I went digging around about frigate and out of memory issues. On a proxmox forum they seemed to conclude there is an issue with python3 used in frigate which is causing these out of memory issues. That forum has postings as recently as last week.

31 minutes ago, capnJiP said:

So I went digging around about frigate and out of memory issues. On a proxmox forum they seemed to conclude there is an issue with python3 used in frigate which is causing these out of memory issues. That forum has postings as recently as last week.

oh good to know, maybe i'll look into an alternative.

It's hard to beat frigate. I've tried several other solutions, and nothing comes close to the usefulness of the AI identification. I'm keeping frigate. I'll just hope that someone smarter than me fixes this upstream.

Does anyone know if this message about "swap" is a problem?

8 hours ago, capnJiP said:

Does anyone know if this message about "swap" is a problem?

 

It's safe to ignore

Hey @velourfogg, question about your setup: Are you saving your frigate footage to a drive under "Unassigned Devices"?

if so:

Do you auto-update or manually update your plugins?

Are your crashes within a few hours of an update corresponding to "unassigned devices" plugin being updated? Last update for Unassigned Devices was released 14Apr.

 

I'm still collecting data, but this looks to be a strong lead...