@grahamperrin@bsd.cafeKDE Plasma: ZFS mount points as devices in the sidebar of Dolphin on FreeBSD-RELEASE:
— is this a new thing, has anyone else seen it?
I'm accustomed to seeing rpool and bpool misrepresented as devices in Kubuntu (a known issue).
I don't recall seeing a comparable bug in FreeBSD before yesterday.
#FreeBSD
@JSkier@social.linux.pizzaMe rebooting my #archlinux #zfs #nas box this morning after updating the kernel.
Both zpools came back up, and all containers 😀
@grifferz@social.bitfolk.comNew ZFS AnyRAID feature would probably get me to use ZFS at home instead of btrfs.
I have a lot of different-sized old SATA enterprise SSDs retired from @… and a cheap 8 bay eSATA enclosure. The performance is good enough. I can't really justify buying 8 new, matching drives.
AnyMirror would be good enough for my purposes.
@nelson@tech.lgbtCopying 4TB of files a second time because I put them in the wrong ZFS dataset the first time. And ZFS doesn't seem to have a zero copy way to move files from dataset to dataset, even within the same pool and on the same disk.
@kcase@mastodon.social“zpool import -FX” is not my favorite command. But when the hosting system panics and leaves the virtual devices backing a FreeBSD VM’s zfs filesystem in an inconsistent state, I'm sure glad it exists.
(I do have hourly zfs snapshots mirrored to a different system that I could restore from if necessary, which comforts me in situations like this. But it's nice not to have to roll back by even a few minutes, other than for the few affected files identified by “zpool scrub” checks…
@grumpybozo@toad.socialIn case you were wondering: no, it is NOT safe to do an unconstrained zfs send of the root filesystem to a vmhgfs-mounted filesystem.
#Sysadminnery
@arXiv_quantph_bot@mastoxiv.pageOvertone Rabi oscillation of optically polarized triplet electron spins and nuclear hyperpolarization in powder
Koichiro Miyanishi, Takuya F. Segawa, Makoto Negoro, Akinori Kagawa, Kazuyuki Takeda
https://arxiv.org/abs/2506.10498
@timfoster@mastodon.socialI finally had a go at doing the thing to our build system that I've been thinking about for ages. Result: proof-of-concept reduces build times by at least 20 minutes, probably more when we're done.
I now just have to get from POC to production. Uh, how hard can it be?
(timf winces)
#fast-git-clone-using-zfs-reference-repos
@JSkier@social.linux.pizza @grahamperrin@bsd.cafePictured: Kubuntu shutting down gracefully – without forcing off the computer – following an insane zpool-scrub(8) command.
For the insanity:
https://github.com/openzfs/zfs/issues/17527
― Gracefully reject an attempt to scrub a read-only pool
@grumpybozo@toad.socialPersonal virt host fell over in the middle of the night. Took out the ZFS pool for my main vm. Rebuilding from backups. Ugh.
@rasterweb@mastodon.social @grifferz@social.bitfolk.comNVMe RAID controllers… I'm skeptical
The selling points here appear to be:
- Each device still has its own channel so you're not trying to hang 4 NVMe off one slot
- The processor on the card can still do the parity
Personally I'm not that interested in parity RAID on NVMe. Either it's mirrors or I'm wanting to use something more advanced like ZFS or Ceph or whatever which pretty much requires JBOD.
Personally I'm more concerned about havin…
@grahamperrin@bsd.cafe@… for convenience:
https://github.com/openzfs/zfs/issues/17185
@grifferz@social.bitfolk.comYear of our lord 2025 and you can buy a 122TB NVMe drive for $12,400 each, but you'll have to use UEFI to boot it and EFI still doesn't understand any software RAID, zfs, etc.
https://wiki.debian.org/UEFI#RAID_for_the_EFI_System_Partition
@grahamperrin@bsd.cafe@… please see the preceding post, it's root-on-ZFS.
@grahamperrin@bsd.cafe@… I don't want to manually create ZFS datasets.
@grahamperrin@bsd.cafeI switched to Kubuntu with root-on-ZFS:
― https://www.reddit.com/r/freebsd/comments/1lr4ayi/switched/
FreeBSD is not entirely abandoned:
―
