seamon

@johner Another Issue. I have a use case where I just want to use Tier 2 and not Tier 1. In this scenario, it seems to be working but Dashboard is useless and doesn't show anything.

There seems to be a bug with Tier 2. Upon a restart, Tier 2 is skipped initially [WARN] Disk swap mount (/mnt/swap) not available yet. Skipping Tier 2.You have to go to ZRAM Settings and click Apply & Save to enable it.

@johner When you get back, is it possible to have: Custom Swappiness for the Tier 2 (Max Value Clamped to Tier 1 Swappiness). How much data is in Tier 1 vs how much data is in Tier 2. Also what is priority? It's set to 100 for Tier 1 and 10 to Tier 2. How does that differ from Swappiness? Can we customize that as well?

--runtime=nvidia in place of --gpus=all is the fix for the docker issue I believe.

I am not able to set the Auto Size to 75% nor make the size customizable. It always creates 16G. Edit: NVM figured it out -> You have to save settings first.

I did some preliminary searching but couldn't find a definitive answer. Is the ZFS Arc shared across all ZFS Pools in Unraid? I like to assign a lot of RAM to Arc and was wondering if all ZFS Pool share it. Special Vdevs specifically L2Arc I suppose is per ZFS Pool? Other ZFS Properties like zfs_arc_max, l2arc_headroom, zil_nocacheflush, etc. are applied to all ZFS Pools with no way to configure them manually per ZFS Pool?

I improved the scripts a little bit more: On Array Startup ls /mnt/cache if [ $? -ne 0 ] then echo "ERROR: Unsafe Shutdown detected!" zpool import cache sleep 15s zpool remove cache /dev/nvme0n1p1 zpool remove cache /dev/nvme0n1p2 shutdown -r now else echo "All operations normal!" zfs set compression=lz4 cache zpool add cache log /dev/nvme0n1p1 zpool add cache cache /dev/nvme0n1p2 fi On Array shutdown zpool remove cache /dev/nvme0n1p1 zpool remove cache /dev/nvme0n1p2 Also, note that each time a log device is removed from a zpool, it leaves a hole vdev in its place. This can be viewed with 'zdb -C'. This apparently is normal behavior for ZFS and completely harmless. It's just tracking vdev removals.

I feel like that's Unraid in a nutshell with regard to a lot of other things over the years.

But my solution works!

Okay I figured out a dirty way to do this. Follow this guide to partition your Optane Drive into 2 Partition (I did 10GB for SLOG and 100GB for L2ARC). SLOG is /dev/nvme0n1p1 and L2ARC is /dev/nvme0n1p2 You can't start the array with SLOG and L2ARC enabled, they have to be added after array is started so I created a User Script that runs at Array Startup. zfs set compression=lz4 cache #zfs compression gets sets to on instead of lz4, this step is optional zpool add cache log /dev/nvme0n1p1 zpool add cache cache /dev/nvme0n1p2 At Array Shutdown, the vdevs have to be removed. zpool import cache #if unraid was improperly shutdown, this script will instead import the zpool, remove and support vdevs and then you have to restart unraid again to make this work. zpool remove cache /dev/nvme0n1p1 zpool remove cache /dev/nvme0n1p2 Any help in refining this further would be appreciated.

Is it somehow possible in Unraid to put ZFS SLOG and L2ARC on the same SSD (Optane P1600x 118GB). I am using it as just a SLOG rn and it's barely using any space. I believe this is possible on TrueNas but not sure if possible to do this on Unraid. My workload involves a lot of Databases so need the SLOG. Adding the SLOG makes almost a 10x difference. I have also maxed out my RAM for the ARC.

Both will work without issues.

If you want to run Minio in Single Node Multi Drive config, then you'll have to change the Post Arguments. It prolly looks something like "server /data ..." for you right now. If you want more drives, you need to change it to "server /data1 /data2 ..." or "server /data {1-3}" (for more than 2 drives). and ofc /data1 /data2 ... should map to the mount points of where you want to store the data. Also, you cannot convert an existing Single Node Single Drive to Multi Drive, it has to be a fresh installation.