dopeytree

I'm thinking to remove one of the cache pool mirrors (total 2x drives) Then wipe the other one format as ZFS. Copy data back to NEW ZFS cache pool single. Confirm data. Then wipe the other BTrFS cache, then add it to the ZFS pool (requiring a format of the BTrFS drive). Does this sound like the right steps for converting a cache pool (mirror) to ZFS mirror?

-... ..- - -..-. ..-. .. .-. ... - -..-. ..- ..-. --- .----. ... -..-. .... .- ...- . -... . . -. -..-. -.. .. ... -.-. --- ...- . .-. . -.. -..-. ... - . .- .-.. .. -. --. -.-- --- ..- .-. ... --- -.-. -.- ... -..-.

Release note here explain: https://docs.unraid.net/unraid-os/release-notes/6.12.0 ZFS is essentially another file format like XFS & BTrFS.. The core benefit of ZFS is data integrity & repair built in. The array still exists & you can use ZFS single drives in the array (for data intregiry benefits). You can also use all the old formats too. ZFS speed benefits only exist in a pool much like the current BTFS raid options available in pools. More detailed ZFS article here: https://arstechnica.com/information-technology/2020/05/zfs-101-understanding-zfs-storage-and-performance/

Is this all good to use with 6.12? the last update was v2021-04-21

6.12 upgrade a breeze. Dying PSU priceless. Only thing I noticed was s3 sleep plugin activated sleep during a parity check but suspect that's a plugin or my config issue. Now to do some copying around to make room for ZFS drives

Updated to 6.12 Then my PSU gave up.. it even made that electric zapping noise as some dodgy Chinese component must have finally given up. (it had been randomly rebooting for a while). Have swapped to a temporary MAG PSU while a new 1u flex psu arrives tomorrow. Powered up started a parity check. Then went for lunch. Came back an hour later and s3 sleep had activated during parity check. Have disable sleep until any further comment from others.

Thanks for new plugin. I noticed config transfer had picked mnt/cache/appdata & mnt/user/ Is this ok? It wont cause a double backup?

Amazing thanks devs

The default 50000 memory is that Kb / Mb or MB??

Sorry I missed your reply in the bug post. Thanks I still think it's a bug having storage percentages shown under a Memory heading? Anyway it is how the devs have decided it is.

Could someone help me understand what the memory GUI is referencing. A mod said that the docker % isn't to do with ram usage but refers to the amount of space used by docker which has a default of 20GB. It's under a heading Memory so I assumed it was a breakdown of the RAM by usage. Are these subheadings actually Storage % ? Thanks

It is under the section heading Memory.. Please can we add your info to the manual? https://docs.unraid.net/unraid-os/release-notes/6.12.0#dashboard

in 6.12 - Are we able to display memory usage in GB as-well as percentage? & also maybe explain or simply what RAM, FLASH, LOG & DOCKER numbers correlate to? i.e Docker is not using 44% of 128GB. Docker is using 44% of total usage which is 6%. Double percentages are weird... 🤓 6% of 128 - 7.68 so 44% of 7.68 = 3.3792 so docker is using 3.37GB

See screenshot of unassigned plugin settings

I think unassigned disks are under /remote have a little dig around & you'll find it.

/plugins/unassigned.devices/UnassignedDevices.php: wrong csrf_token Is this csrf_token of concern? Or is it simply because I have refreshed the page rather than close and open a new window to login? CSRF is to do with login security.

Mover would run daily. You'd have to define which is more important (maybe per share.) For the 'data downloads share' priority is to keep 65% (or whatever number you choose) on the SSD. The idea of SSD's should be to use them not to just dump data to them and then move it to the array or another ZFS pool. which then leaves the SSD blank. Mover tuning can do some basic customisations like only move certain files or files over certain days old but there's still nothing on a per share basis. However the author is going to write some custom scripts. ------------------------------ Essentially I am saying I would like 'ALL new folders' to be 'cached on SSD's' (because its fast) then anything that is not my data downloads share to move to the array daily/weekly (insert time point) for parity protection whilst my 'data downloads share' keeps itself around 65% full. Maybe ZFS L2arc could play a part in all this in the future...

Extra models can be downloaded here: https://huggingface.co/TheBloke and copied to appdata/serge/weights/

It's now in the app store

Perhaps a script to do this.. e.g 'when share on cache/s (primary) reach 75% capacity' = move 'anything over 90days old' to array (secondary). { but never move: APPDATA, DOMAINS, SYSTEM }

It sounds like the incoming 6.12 only has primary & secondary storage options for mover so thanks this will be much appreciated. If it makes it any easier I run downloads in a separate cache pool. Everything else happens on the normal cache pool. To make dates work for newly copied over folders on normal cache (non downloads) I guess I'd want those to use dates that refer to when the item was copied rather than its original creation date which happened on another PC. Anyway e.g 'when share on cache/s (primary) reaches 75% capacity' = move 'anything over 90days old' to array (secondary). { but never move: APPDATA, DOMAINS, SYSTEM }

I would love to be able to keep ONLY my recent DOWNLOADS (data share) cached while at 75 / 85%. Would still like other normal new share folders to get moved regularly to the array.

It did about 6 sudden resets last night before I removed those plugins & turned sleep off. Removed usb stick and ran in bios for 40mins to see if it occurred again. Seems to be stable since. Is running a parity check 14hours in with 2 more to go.

Cool so out of curiosity how do you guys have your downloads & mover set up?