Hi all, I'm getting out of memory errors on my server, it says it's killing off processes but I'm not sure why. I have 16GB of RAM on my server with only a handful of dockers and 1 VM. I followed instructions on Tips and Tweaks and reduced Disk Cache 'vm.dirty_background_ratio' (%): to 1 and Disk Cache 'vm.dirty_ratio' (%): to 2

 

Diagnostics are attached. Thanks all for the help.

 

 

unraid-diagnostics-20180129-1550.zip

My guess is a docker has misconfigured volume mappings and so is writing into RAM instead of to an actual storage device. Only /mnt/user, /mnt/disk#, and /mnt/cache are actual storage. /boot is the flash drive, and any other path is in the RAMfs.

3 minutes ago, trurl said:

My guess is a docker has misconfigured volume mappings and so is writing into RAM instead of to an actual storage device. Only /mnt/user, /mnt/disk#, and /mnt/cache are actual storage. /boot is the flash drive, and any other path is in the RAMfs.

 

Thanks for that, I went over my dockers and there's nothing that isn't pointing to /mnt/user

 

See screenshot

 

12 hours ago, yitzi said:

 

Thanks for that, I went over my dockers and there's nothing that isn't pointing to /mnt/user

 

See screenshot

 

 

Do you get out of memory errors when using Plex? If that is the case it might be related to transcoding and how the folder mapping is set for the /transcode folder

3 hours ago, Leifgg said:

 

Do you get out of memory errors when using Plex? If that is the case it might be related to transcoding and how the folder mapping is set for the /transcode folder

 

Default is for Plex to transcode to a folder inside appdata. It is possible to change that in the Plex application, but since he hasn't mapped a folder for transcoding, even if he told Plex to use something other than default, that unmapped folder would be inside docker.img, not in RAM.

4 hours ago, Leifgg said:

 

Do you get out of memory errors when using Plex? If that is the case it might be related to transcoding and how the folder mapping is set for the /transcode folder

1 hour ago, trurl said:

 

Default is for Plex to transcode to a folder inside appdata. It is possible to change that in the Plex application, but since he hasn't mapped a folder for transcoding, even if he told Plex to use something other than default, that unmapped folder would be inside docker.img, not in RAM.

 

Thanks for the responses. I tackled this by using Open Files plugin and seeing a process for "mono" consuming lots of RAM. I shutdown each docker one at a time whole monitoring RAM usage. Looks like Jackett was using about 60% of my 16GB. No idea why. 

 

I'll reach out on LSIO support page for this. 

 

Thanks again, this is solved.