Unable to login to the webui, I get an error of "Warning: session_write_close(): write failed: No space left on device (28) in /usr/local/emhttp/login.php on line 33". There are two others errors but didn't manage to copy them.

I have to reboot via SSH, but after I do, a bunch of my dockers lose all of their settings.

The array has over 7TB free so it definitely is not full. RAID 6 and no cache drive.

Edit:Pihole keeps its settings, located at "/mnt/cache/appdata/pihole/pihole/". Plex loses settings, located at "/mnt/user/appdata/PlexMediaServer".

Edit 2: Under shares, appdata, system, and domains are set to prefer cache, appdata has a yellow triangle stating "Some or all files unprotected"

Edit 3: https://imgur.com/a/ivlCeBf it shows 357mb on cache and 0b free, but I don't even have a cache drive set up?

Edit 4: I'll turn one of the smaller drives (600GB, 15k rpm) into a cache drive if that'll help?

6 minutes ago, Katsumaaji said:

RAID 6

Unraid IS NOT RAID so I don't know what you mean here.

 

8 minutes ago, Katsumaaji said:

no cache drive...Pihole keeps its settings, located at "/mnt/cache/appdata/pihole/pihole/"

If you have no cache drive the path /mnt/cache should not exist. Unless of course you create that path by specifying it in the mappings for that docker. In which case, it would be a path in RAM. But that wouldn't survive reboot.

 

My guess is you are filling RAM due to an incorrect host path on Pihole.

 

If possible before rebooting and preferably with the array started
Go to Tools - Diagnostics and attach the complete Diagnostics ZIP file to your NEXT post in this thread.

Also, post your docker run for pihole as explained at this very first link in the Docker FAQ:

 

tower-diagnostics-20201216-1553.zip

3 minutes ago, trurl said:

Unraid IS NOT RAID so I don't know what you mean here.

 

If you have no cache drive the path /mnt/cache should not exist. Unless of course you create that path by specifying it in the mappings for that docker. In which case, it would be a path in RAM. But that wouldn't survive reboot.

 

My guess is you are filling RAM due to an incorrect host path on Pihole.

 

Yeah my mistake.  Double party always just sticks out as RAID6 in my head.  Pihole was just an example.  Several other dockers keep settings while others don't.  I can list off each one?

Edited December 16, 20205 yr by Katsumaaji

4 minutes ago, trurl said:

Unraid IS NOT RAID so I don't know what you mean here.

 

If you have no cache drive the path /mnt/cache should not exist. Unless of course you create that path by specifying it in the mappings for that docker. In which case, it would be a path in RAM. But that wouldn't survive reboot.

 

My guess is you are filling RAM due to an incorrect host path on Pihole.

 

The dockers with settings that don't survive reboot have settings located in /mnt/user/appdata/

15 minutes ago, trurl said:

post your docker run for pihole as explained at this very first link in the Docker FAQ

 

so I looked at /mnt/cache and sure enough all of the appdata folders are there, even though I don't have a cache drive.  And since Plex is there, I'm sure metadata is what's filling the 'cache' up.  What's the best way to fix this?  Edit each docker setting, or convert a data drive into a cache drive?

Edited December 16, 20205 yr by Katsumaaji

If you don't have a cache drive, then none of the host paths for any of your dockers should refer to /mnt/cache, since that is how you are accidentally creating a "cache" that only exists in RAM.

 

And of course, none of this has anything at all to do with updating to 6.9rc1.

48 minutes ago, trurl said:

If you don't have a cache drive, then none of the host paths for any of your dockers should refer to /mnt/cache, since that is how you are accidentally creating a "cache" that only exists in RAM.

 

And of course, none of this has anything at all to do with updating to 6.9rc1.

While that makes sense, I rolled back to 6.8.3 and everything is fine now.  Dockers survive reboots.

 

I think what I'll do is try and figure out to take one of the smaller data drives (600gb, 15k rpm) and make it a cache drive.

Edited December 17, 20205 yr by Katsumaaji

53 minutes ago, Katsumaaji said:

I rolled back to 6.8.3 and everything is fine now

Do you still see /mnt/cache?

32 minutes ago, trurl said:

Do you still see /mnt/cache?

I do, but it's empty.

You must have something creating it then, most likely a docker mapping as I said, since you don't actually have a mounted cache device; that path is in RAM waiting for one of your dockers to fill it up again.

 

Go to the Dockers page, set the slider at upper right to Advanced View, and post a screenshot. 

2 hours ago, Katsumaaji said:

Dockers survive reboots.

It might just be that your dockers are restarting from scratch at each boot. The dockers executable code is in docker.img, which you have specified to be in /mnt/user/system and so is on one of your disks. The docker templates that set them up again is on the flash drive. But if you have appdata for some of them in /mnt/cache, then that is going to be in RAM, so while the dockers will exist upon reboot, the applications themselves won't remember anything because their appdata is starting over.

Should have been more specific  The dockers survive reboots with appdata intact

You still haven't figured out how you are creating /mnt/cache. I am just speculating based on the limited information you've given. 

 

In any case there would be several advantages if you actually had a cache drive. Quite a bit of work to do to get a disk out of the array and make it cache though. And more work to get things that need to be on cache moved there. 

 

Is that what you want to do? 

7 hours ago, trurl said:

You still haven't figured out how you are creating /mnt/cache. I am just speculating based on the limited information you've given. 

 

In any case there would be several advantages if you actually had a cache drive. Quite a bit of work to do to get a disk out of the array and make it cache though. And more work to get things that need to be on cache moved there. 

 

Is that what you want to do? 

I think so.  I've tripled the size of the array, so I think I can sacrifice one slot for a cache drive.

 

Is your disk controller(s) using some sort of RAID mode? The disks aren't being identified by their serial numbers. You may have more to work through than I realized at first. Using RAID with Unraid can cause several problems.

55 minutes ago, trurl said:

Is your disk controller(s) using some sort of RAID mode? The disks aren't being identified by their serial numbers. You may have more to work through than I realized at first. Using RAID with Unraid can cause several problems.

Perc H310.  12 drive array with two parity.  I have the Disk Location plugin so I can accurately ID them.

Edit: Why would you think they're not being IDd correctly?
Edit 2:  Oh right.  The attached diags

 

Edited December 17, 20205 yr by Katsumaaji

Do a google search for this:

 

Quote

perc h310 raid mode site:forums.unraid.net

 

to see the sorts of problems you can have running RAID and advice on how you can fix that.

2 hours ago, trurl said:

Do a google search for this:

 

 

to see the sorts of problems you can have running RAID and advice on how you can fix that.

It's already been flashed into IT mode.  That was the first thing I did.  Disks weren't even recognized until it was flashed.

Edited December 18, 20205 yr by Katsumaaji

1 hour ago, Katsumaaji said:

been flashed into IT mode

OK

 

On 12/16/2020 at 9:20 PM, Katsumaaji said:

take one of the smaller data drives (600gb, 15k rpm) and make it a cache drive

To get that disk assigned as cache, first go to Settings and disable both Docker and VM Manager.

 

Then simplest thing would be to New Config with the disks assigned as you want including cache then rebuild parity. You must be very sure you don't assign any data disk to any parity slot or it will be overwritten with parity.

 

After that we can work on getting things moved to/from cache as needed before enabling Dockers/VMs again.