Unraid 6.6.4 Broke Cache/appdata, unsure how to fix

[Dartanis]

Hi there,

 

Read that it was possible to transition my 2 SSDs from a RAID1 config to a RAID0 and thought it fit my use cases as just a temp holding location a lot more since I have Mover run hourly anyways. Used "-dconvert=raid0 -mconvert=raid1" in the Balance field of my cache window, saw it going to work and thought great, now I'll have a 500GB cache instead of mirrored 250GB. Saw the value not change for total space in cache but looked at the btrfs filesystem df which listed:

 

Data, RAID1: total=2.00GiB, used=1.99GiB Data, RAID0: total=18.00GiB, used=3.86GiB System, RAID1: total=32.00MiB, used=16.00KiB Metadata, RAID1: total=1.00GiB, used=322.72MiB GlobalReserve, single: total=19.66MiB, used=64.00KiB

 

So I thought it had created a bit of both, rebooted the server and saw the value had not changed; despite being a very bad idea in retrospect I tried spinning down the typically mirror drive and seeing how the cache behaved. At this point a user for my Plex server informed me they lost connection. I tried it on my local network, server reports offline, try to access webgui through ip:32400, refused despite still being forwarded. Checked my nextcloud docker, unable to access it as well, same Refused to Connect error. Annd that is where I am currently stuck, tried reloading the Plex docker but getting the same error, though in bridge mode it seems to time out instead of giving me the error.

 

I decided it would be best to remove my cache drives from the pool, reboot and then re-add after running -dconvert=raid1 -mconvert=raid1 to try and revert my changes. But the filesystem df still shows the RAID0 data on there too. So this is the point where I am stuck now, basically trying to reset my cache and forcibly reloading my appdata but unsure of how to proceed with it

[JorgeB]

Please post your diagnostics: Tools -> Diagnostics

[Dartanis]

Attached diagnostics

pantheon-diagnostics-20181124-0105.zip

[JorgeB]

There's a problem with the cache filesystem, that's why the balance didn't complete, best way forward is to backup any important cache data, re-format (either raid1 or raid0) and restore data.

[Dartanis]

Nothing important lived on the cache drives, how exactly do I go about making sure the cache is set to RAID0 and reformatting it? Not sure where I'm supposed to initiate this action from.

[JorgeB]

Stop, array, clear both SSDs with:

blkdiscard /dev/sdX

start array, format cache pool, convert to raid0

[Squid]

After this is all sorted out, you need to upgrade to 6.6.5.  6.6.4 was only available for ~1 day and has a big issue with schedules not running which may or may not have contributed to the issue.