15 hours ago, foz said:

On Unraid, each share on a ZFS pool is its own dataset, so mapping /mnt/cache as one drive backs up nothing. Map /mnt/cache/<share> instead.

It looks like drives A: and B: don't get mapped/mounted/available for selection in the Backblaze UI. If I've got to map each share on a zpool too (it looked to me like all my zpool shares were getting backed up previously, but I didn't dig too deep there...), I'm unfortunately out of drives to map without A and B! Is there a solution for mapping beyond 26 drives (minus C and Z) in Windows/Backblaze?

edit:

it looks like at least for me, Backblaze is at least looking inside individual directories (the "Archive" directory which is an "exclusive share" on my /mnt/ztank ZFS zpool), which is mapped to drive D

newfstatat(AT_FDCWD</config/wine/drive_c/windows/system32>, "/config/wine/dosdevices/mountpointmanager", 0x7ffffdfff530, 0) = -1 ENOENT (No such file or directory)
--- SIGUSR1 {si_signo=SIGUSR1, si_code=SI_TKILL, si_pid=1896, si_uid=99} ---
--- SIGUSR1 {si_signo=SIGUSR1, si_code=SI_TKILL, si_pid=1896, si_uid=99} ---
newfstatat(AT_FDCWD</config/wine/drive_c/windows/system32>, "/config/wine/dosdevices/d:/Archive/nickdesktop backup/myDir/ectory", {st_mode=S_IFDIR|0777, st_size=4, ...}, 0) = 0
fstatfs(16</drive_d/Archive/nickdesktop backup/myDir/ectory>, {f_type=ZFS_SUPER_MAGIC, f_bsize=131072, f_blocks=258493881, f_bfree=99665724, f_bavail=99665724, f_files=25515974823, f_ffree=25514425426, f_fsid={val=[0xfff2ece9, 0xdbf026]}, f_namelen=255, f_frsize=131072, f_flags=ST_VALID|ST_NOATIME}) = 0

Ah -- I see what's happening. Certain of my top level directories on my ztank zpool are plain directories rather than separate ZFS datasets for whatever reason (probably me moving around files under the hood/at CLI).

Confirmed that the ones which are actually separate datasets don't appear to be backed up, as you said. Note that this appears to apply to single-drive ZFS disks in the array too, which complicates the "use disk shares for Backblaze" plan if you've got ZFS disks in your main array

Edited October 9Oct 9 by nick5429

On Windows, you can mount to an empty folder, but you don't get a drive letter. I don't think you can do that in Wine. If only I had the problem of having too many drives; my 3 14s are it, and one used to be parity 😅

Backblaze is my parity now

-- when I switched to per drive backup I pushed my parity drive into the array. Since if one goes down I can download it from the cloud. I figured since parity is not RAID it's not really doing much if I have full drive backups. I am fortunate enough to have two large usb drives that get backed up to every night so I'm doing the 3:2:1 rule. It would be much harder to figure out what to restore from the usb backup since those are whole shares sent to specific drives.

Edited October 9Oct 9 by rogman

Hey quick question. Is anyone using a tmpfs ramdsik drive as their temp data drive for backblaze? My goal is to save some read/writes on my cache nvme. I have it setup with scripts that dump the contents to my nvme cache drive on array stop and back to the tmfps drive when the array starts back up prior to the container launching. It seems to be working really well aside from one issue, which I'm not sure if its solvable or if i should abandon the idea all together.

I get this error message:
The scanner cannot open 1 directory, so a disk's file list stops there and files after it are not backed up, and bb-doctor names them
bb-doctor is calling out the tmfps drive which obviosly isnt selected as a drive to backup and only used as a scratch disk:
Backup health [ ok ] no stall detected [warn] 1 directory the scanner cannot open, so nothing inside them is backed up: D:\ - Windows error 2 rename the directory, or fix its ownership, and the next scan runs through.

The comand im using to make the tmpfs drive is this:
mount -t tmpfs -o size=6G tmpfs /mnt/disks/backblaze_ramdisk

Keep in mind it is working perfectly fine as a scratch disk as far as i can tell and I dont have it selected as a drive for backups. Backblaze is able to create and flush the temp .dat files so Im not sure why its showing a read error. Maybe i need to use a different comand when creating the tmpfs folder? @rogman Is it possible to suppress this error from showing in the Backblaze64 Monitor health section. I've dismissed it within the docker but its still showing in the plugin.

Thanks for any help

backblaze64-diag-202610171709.zip

The health status pulls directly from the API and what it reads.

It's been running for about 6 hours and it looks like the error cleared itself? Ill keep an eye on it... So i guess back to my original question. Anyone else using a tmpfs ramdisk as your backblaze work folder to save nvme writes? Have you run into any issues?