Re: zfs snapshot corruption when using encryption

> 8 nov. 2024 kl. 20:20 skrev void <void@f-m.fm>:
> 
> On Fri, Nov 08, 2024 at 03:29:22PM +0100, Palle Girgensohn wrote:
> 
>> I cannot run `zfs send -I fs@previous_snap fs@problematic_snap`, I get
>> warning: cannot send 's@problematic_snap': Input/output error`
>> Removing the snapshot fixes the problem.
> 
> What an odd problem. Like, it can write but not read. What zfs version?
> On 13.3 I'm using:
> zfs-2.1.14-FreeBSD_gd99134be8
> zfs-kmod-2.1.14-FreeBSD_gd99134be8
> 
> on 14-stable:
> zfs-2.2.6-FreeBSD_g33174af15
> zfs-kmod-2.2.6-FreeBSD_g33174af15
> 
> on 15-current:
> zfs-2.3.99-31-FreeBSD_gb2f6de7b5
> zfs-kmod-2.3.99-31-FreeBSD_gb2f6de7b5
> 

$ freebsd-version -ku
14.0-RELEASE-p10
14.0-RELEASE-p10

dunno how to present in the above fashion? Binary install, not from source.

> Is the encryption you're using is the GELI
> based whole-disk one (which has been around iirc for a few years) or
> the relatively recent zfs encryption that works per-filesystem. This is why I'm asking the zfs version. I thought the latter is relatively quite new, and I've never heard of it working on 14.0.

The latter, zfs per-filesystem zfs encryption. I haven't heard any horror stories about it, albeit it's quite fresh code.


> The way I'd go about trying to address the issue would be to start from the lowest layer and work up. On a raidz2 array (the card in JBOD mode), 

yes, JBOD

> I can identify all the drives as da0-7, so can use smartctl -x to query them all directly, looking for, in particular:
> 
> Reallocated_Sector_Ct
> Reported_Uncorrect
> Current_Pending_Sector
> Offline_Uncorrectable
> 
> If your disk arrangement is connected via something like HP Smart Array
> or similar, you'll need to look at the man page for smartctl for the exact
> syntax to query the disks behind the card.

Na, it's JBOD alright. I see nothing weird with the disks.

Palle