Rootfs 96% full, can't download updates

[jsmj]

Hey friends. I can't figure this one out. My Fix Common Problems plugin is griping that rootfs is getting full. If I understand it right, something is either writing to /tmp or memory, but best I can tell my dockers all write to /mnt/user or /mnt/disk# or /mnt/cache, but I could have overlooked something.

 

Possibly related, I can't download the 6.7 update because I get a i/o failure. From googling this is because the gui saves the update to RAM, and mine is full from the rootfs issue. I've attached my diagnostics from Fix Common Problems. Let me know if I can provide anything else.

 

I'm currently pre-clearing a drive, if that matters. Just trying to include everything I can think of that's going on.

tower-diagnostics-20190321-1617.zip

[trurl]

Some odd things in your diagnostics but maybe just because your version is so old. You are getting wrong csrf tokens from preclear in syslog, but according to your diagnostics you don't actually have the preclear plugin installed.

2 hours ago, jsmj said:

best I can tell my dockers all write to /mnt/user or /mnt/disk#

You have given your docker image 50G, which makes me question if you don't have some issue with your docker setup. But you aren't filling that up, it's just the fact that you have it so large that makes me think you may have been filling it in the past. It is extremely unlikely you would ever need more than 20G, and you currently aren't using anywhere near 20.

 

But filling docker image is not the same as filling rootfs anyway. Filling docker image would mean one of your applications was writing to a path that wasn't mapped. But filling rootfs might mean one of your volume mappings isn't to actual storage.

2 hours ago, jsmj said:

or /mnt/cache

But according to your diagnostics, you don't actually have cache, so if you have that as some path then maybe we found your problem.

[jsmj]

Quote

you don't actually have the preclear plugin installed.

I don't. I used to, but I was under the impression that it's no longer needed and that the server will take care of it by default when the array is spun up with a new drive.

 

Quote

You have given your docker image 50G

It was set at 10G by default and I was nearing that limit. I wasn't sure a sane value to expand it to and 50G sounded good and I had the storage. This was quite awhile back.

 

Quote

But according to your diagnostics, you don't have actually have cache

the only /mnt/cache mapping I can see is Plex /config mapped to /mnt/cache/appdata/Plex. But it's been like this for months. Could it slowly grow to fill rootfs and be finally cropping up? What should it be mapped to if not /mnt/cache/appdata?

[JonathanM]

If you don't have a cache drive, and you are saving your plex config to /mnt/cache which is actually RAM, then if you reboot, all your plex configuration will be lost.

 

Do you care if plex needs to be set up again? If so, you need to move that folder to a physical disk before you reboot.

[jsmj]

alright I moved the /config mapping to /mnt/user/appdata/plex and when this preclear finishes I'll reboot and see where that gets me. I had to set up the Plex libraries again but I take it the'll stick now that they're on a physical drive

 

[jsmj]

Alright, preclear finished over night and I formatted and added the drive to the pool this morning

 

Rebooted and rootfs warning was gone from FCP

 

Successfully updated to 6.6.7, rebooted again (new UI looks legit)

 

FCP is reporting no problems, Plex seems to be working and behaving normally. Best I can tell I'm in the clear. Thanks guys