On 7/9/25 15:41, Kirill A. Korinsky wrote:
On Wed, 09 Jul 2025 23:28:03 +0200,
Stuart Henderson<stu.li...@spacehopper.org> wrote:
On 2025-07-09, Kirill A Korinsky<kir...@korins.ky> wrote:
On Wed, 09 Jul 2025 22:54:50 +0200,
Stuart Henderson<stu.li...@spacehopper.org> wrote:
- restore (and archive browsing) on OpenBSD is more annoying than it could
be if we had a fuse implementation that was a bit more compatible with
current versions on other os.
It has a fork with name rustic which has interactive mode to navigate inside
snapshots, and webdav.
not a fork, it's different software written in a different language though
using a compatible repo format.
"rustic currently is in beta state and misses regression tests. It is
not recommended to use it for production backups, yet."
https://rustic.cli.rs/docs/comparison-restic.html restic development: "conservative with
changes" (which I agree with from
experience) is a good state for backup software.
rustic: "moving fast, add new features early" ...
I wouldn't use it for backups yet though maybe worth a look for use as a
restore tool. There is alsohttps://github.com/emuell/restic-browser.
Yeah, "fork" defently a bad word here.
I'd like to add this explanation about "beta status":
https://github.com/rustic-rs/rustic/discussions/1459#discussioncomment-13071763
Brian that's great you use OpenBSD as a NAS! Crystal you make a great point
about using a simple shell script to verify file integrity with checksums.
I certainly want to detect a failing drive for my source data and replace it
before corrupted data is backed up from it.
Does anyone have examples of such shell scripts, such as with chsum(1) or
md5(1)?
Do you trust them with giant files of several gigabytes or even terabyte-sized
VM
or database files?
If cryptographic hashes would be better, would features of LibreSSL, OpenSSH, or
another OpenBSD base system tool be suitable for a fileserver/NAS?
Looking at the manpages, I don't think softraid(4) or bioctl(8) can contribute
to repairing or replacing an individual corrupted file with a known good copy,
or
that ffs has a copies-equals-two option, except for the superblock described in
fs(5) (search for the word 'copies').
I imagine some backup systems, dump(8), pax(1), or ports such as borgbackup can
detect intentional modification versus unintentional corrumption due to bit
flips
when a backup is run.
I also understand it's easy enough to restore a single file or selection of
files
from backup.
But I would rather have the system hosting my source data find and fix data
corruption and alert me to drive failures in the first place and only have
verified-correct data be backed up to a different system in the first place.
If the periodic checksum script discovers an error, what are options to correct
it on the same system without restoring from backup?
FreeBSD with it's deep integration with ZFS is the easy answer for some. As
might
be NetBSD with ZFS, DragonflyBSD with Hammer (I have not looked into its
features),
or even OmniOS if with supported hardware.
But I want to consider OpenBSD for my NAS and see how others such as Brian have
succeeded at it.
Specifically I'm looking either for base system tools, or ports that are up to
date. That would sadly exclude exciting tools such as Parchive, which is
actively developed with a 1.0 release in April this year, but with an
out-of-date
OpenBSD port from 2017. (Yes I could learn to become a port maintainer and
update
it myself.)
https://parchive.github.io
https://github.com/Parchive
