Good morning all,

       I keep having issues with my CPU and my RAM maxing out randomly to the point where I need to restart the server. It happens almost everyday. I have attached logs. Thank you. 

defiant-diagnostics-20230110-1037.zip

Server load is extremely high, and shfs is one of the main culprits, if possible try to use disk shares instead of user shares for at least some of the dockers and see if that helps.

Can you elaborate further? What is a disk share vs a user share? 

Follow the links.

26 minutes ago, Reddog1400 said:

Can you elaborate further? What is a disk share vs a user share? 

Click the links in your post.

 

Might be more correct to say use disks instead of user shares for some of your dockers, since it has nothing to do with actually sharing disks.

 

Instead of a path in /mnt/user/appdata, for example, you might specify a path in /mnt/cache/appdata. Or maybe /mnt/plex_cache/appdata since I see you also have a pool named plex_cache

 

Not exactly related, but I notice your prefer:cache appdata share has some files on the array. To get these moved to cache you have to disable Docker in Settings and run mover. Mover won't replace files though, so if there are duplicates you will have to clean those up yourself with Dynamix File Manager

ok, thanks, I will give it a try. 

On 1/10/2023 at 11:23 AM, trurl said:

Click the links in your post.

 

Might be more correct to say use disks instead of user shares for some of your dockers, since it has nothing to do with actually sharing disks.

 

Instead of a path in /mnt/user/appdata, for example, you might specify a path in /mnt/cache/appdata. Or maybe /mnt/plex_cache/appdata since I see you also have a pool named plex_cache

 

Not exactly related, but I notice your prefer:cache appdata share has some files on the array. To get these moved to cache you have to disable Docker in Settings and run mover. Mover won't replace files though, so if there are duplicates you will have to clean those up yourself with Dynamix File Manager

I did this with a few and it seemed to work a little. Then it happened again. Did you notice certain dockers being the culprits? Also, is this why the RAM is maxing out too?

Post new diags.

I had to wait for it to happen again. Right now it is pinned and RAM is at 91%

 

 

defiant-diagnostics-20230119-1022.zip

Plex appears to be the main culprit, does it get better if you shut it down?

I haven't tried that by itself. If I stop and restart the array it fixes it. Do you see what I can do to fix the Plex issue. This is all new to me. I mean I have been running a server for a couple of years, but I have just be self teaching myself stuff. 

First confirm if it is Plex by stopping it.

ok, my server ground to a halt so I restarted the array. I will do that the next time this happens. Then I will send new Diags. 

Ok, here are some Diags. Before Plex shutdown, after shutdown, and I restarted Plex after. After I shutdown Plex, the CPUs came down, but the RAM didn't. When I restarted Plex, the RAM cleared. 

defiant-diagnostics-20230120-0956 before Plex shutdown.zip defiant-diagnostics-20230120-1005 restarted Plex.zip defiant-diagnostics-20230120-0959 after Plex shutdown.zip

25 minutes ago, Reddog1400 said:

After I shutdown Plex, the CPUs came down

That suggests a Plex problem then, what Plez release are you using?

linuxserver/plex

 

This is the repository I am using. This is with the GPU. 

I've never used Plex so can't really help with that, suggest visiting the Linuxserver's discord, or maybe someone else can help here.

Post docker run for plex

docker run
  -d
  --name='plex-gpu'
  --net='defiant1'
  --cpuset-cpus='4,5,6,7,12,13,14,15'
  -e TZ="America/Indiana/Indianapolis"
  -e HOST_OS="Unraid"
  -e HOST_HOSTNAME="Defiant"
  -e HOST_CONTAINERNAME="plex-gpu"
  -e 'VERSION'='docker'
  -e 'NVIDIA_VISIBLE_DEVICES'='GPU-22894283-12c9-ee8d-0da8-55e6693131b0'
  -e 'NVIDIA_DRIVER_CAPABILITIES'='all'
  -e 'PUID'='99'
  -e 'PGID'='100'
  -l net.unraid.docker.managed=dockerman
  -l net.unraid.docker.webui='http://[IP]:[PORT:32400]/web'
  -l net.unraid.docker.icon='https://raw.githubusercontent.com/linuxserver/docker-templates/master/linuxserver.io/img/plex-icon.png'
  -p '32400:32400/tcp'
  -p '3005:3005/tcp'
  -p '8324:8324/tcp'
  -p '32469:32469/tcp'
  -p '1901:1900/udp'
  -p '32410:32410/udp'
  -p '32412:32412/udp'
  -p '32413:32413/udp'
  -p '32414:32414/udp'
  -v '/mnt/cache/Movies/':'/movies':'rw'
  -v '/mnt/cache/Television/':'/tv':'rw'
  -v '/mnt/cache/Music/':'/music':'rw'
  -v '/mnt/plex_cache/transcode':'/transcode':'rw'
  -v '/mnt':'/mnt':'rw'
  -v '/mnt/cache/appdata/dizquetv/':'/dizquetv':'rw'
  -v '/mnt/plex_cache/plex_appdata/':'/config':'rw'
  --runtime=nvidia 'linuxserver/plex'

48163179643e14218fa33e7afa3ca232d6d34585639184190ab4e9d6c1131996

The command finished successfully!

@trurl is this what you needed to see?

2 minutes ago, Reddog1400 said:

@trurl is this what you needed to see?

A blast from the past!

 

I notice you are mapping directly to cache or other pool instead of user shares.  I wonder if you didn't take the advice of using disks instead of user shares too far.

 

Post new diagnostics

So, I shouldn't do that? So, what is your advice? 

2 minutes ago, Reddog1400 said:

what is your advice?

55 minutes ago, trurl said:

Post new diagnostics

 

Sorry, missed you asked for that in the original post. 

defiant-diagnostics-20230130-1316.zip

Your appdata has files on the array. That is the main one that you probably should be directly mapping to cache. You need to get that all on cache for that to work across the board.

 

Your other mappings, for example, 

On 1/20/2023 at 6:21 PM, Reddog1400 said:

 -v '/mnt/cache/Movies/':'/movies':'rw'

means that when plex accesses /movies, all it will see are those that are on cache. I assume that isn't what you want, since it looks like your Movies share has files on lots of other disks.