Michael_P

Looks like on the 2nd and 18th of June Plex media scanner hit its memory limit and was killed. You can check the container logs to see if there's a file it doesn't like. Reboot to clear the log so FCP stops warning you

If it does it again, run ps -auxf > ps.txt from the command line and that should show you what's using it. Navigate to a folder share first so you can grab the ps.txt that it generates

Only if they ship it DDP, else it's on the IOR

and fees - MPF alone will be $103 and change, not to mention the carrier's fees to broker it for you (and tariffs if it's made in China).

Kate's a text editor so probably not related to watching plex - if it happens again you can run ps -auxf from the command line to see what process is kicking it off (krusader was running at the time, may have been in there)

Looks like on May 01 at 0915 and 0916 kate decided to use an absolute crap ton of RAM, so if you were doing something at that time then that's the issue. You can reboot to clear the log so FCP stops warning you, and if it happens again you can investigate further

Ok, you can use it to find out what's spawning the gunicorn and ruby processes, and either limit the RAM to that container or reduce your VM's allocation. Since the reaper is killing the VM each time (it's the largest consumer at the time the host starts to OOM), that would indicate that you just don't have enough RAM to run everything, so you can either add more (in this economy!?!) or reconfigure your services. Once you've done all that, you can re-boot to clear the log so FCP stops warning you

If I had to guess based on the processes running at the time, I'd say it was immich. Dump the result of ps -auxf to a file and post it, taking care to remove any passwords if they are included in your container's startup commands ps -auxf > ps.txt

Back on APR 12 one of your containers (8c565a95c675) hit its assigned limit and was killed. Looks like a one-off occurrence so you can reboot to clear the log so FCP stops warning you. If it happens again, you can check the container's config or assign it more memory

It's not actually running out of memory, it's a driver issue - update the driver and/or the motherboard BIOS

It's usually just a driver/BIOS issue, not something physically wrong with the card

It's actually your nvidia card falling over, FCP is reporting based on finding 'out of memory' in the syslog. Try updating the drivers and/or the motherboard BIOS Apr 11 03:08:00 nomad kernel: NVRM: nvCheckOkFailedNoLog: Check failed: Out of memory [NV_ERR_NO_MEMORY] (0x00000051) returned from _memdescAllocInternal(pMemDesc) @ mem_desc.c:1359

Looks like back on the 23 March whatever container matches this ID 7f7c77b7b0989 ran over its assigned limit and was repeatedly killed, then on the 24th it looks like you were running something that had ffmpeg doing work which tipped the host OOM so it stopped the highest consumer which at the time was your VM. Doesn't look like you need to do anything, so just reboot to clear the log and FCP will stop warning you

That looks right for the server, just make sure /temp is set up the same way for the tdarr_node container

Yes, both the server and node containers should look something like this

Yep, and that's why it was killed, it blew past the limit Mar 31 02:09:14 Orion kernel: Memory cgroup out of memory: Killed process 1611155 (Tdarr_Node) total-vm:24564080kB, anon-rss:12980236kB, file-rss:13308kB, shmem-rss:0kB, UID:0 pgtables:316432kB oom_score_adj:0 Mar 31 02:09:17 Orion kernel: oom_reaper: reaped process 1611155 (Tdarr_Node), now anon-rss:76kB, file-rss:2432kB, shmem-rss:0kB You need to map /temp to a disk share in the tdarr_node container, or preferably an old SSD you don't mind abusing.

Maybe, but it was tdarr that hit its limit and ran OOM and was killed. Change the config so it's work area is on a disk. Reboot to clear the log so FCP stops warning you

Yeah, stress ran the host OOM, reboot to clear the log so FCP stops warning you

Try re-uploading your diagnostics, there's nothing in it

Looks like frigate was transcoding a bunch of streams back on the 28th of Feb and ran the host OOM. You can re-boot to clear the log so FCP stops warning you, and if it happens again you can fix the config so that it's transcoding to disk instead

Didn't read the wall of text, but it's clear you're over thinking it :) Just use NginxProxyManager, it'll let you host your services and provide the SSL certs if you need them (you probably don't). The only time you REALLY need to use a VPN is to access the local network. Make sure your services stay up-to-date, and you won't have much to be worried about.

Update the nvidia driver and/or your motherboard bios

With 16G of RAM you shouldn't be using it as cache, it's actually recommended to use an old SSD for cache as the file sizes can be pretty large if you're working with blurays and such. Map the cache location to a disk.

While power issues are a concern, I would direct your attention to the controller falling over and the critical btrfs errors @JorgeB what do you think?