Re: git: 2a58b312b62f - main - zfs: merge openzfs/zfs@431083f75

From: Cy Schubert <Cy.Schubert_at_cschubert.com>
Date: Sat, 15 Apr 2023 16:13:47 UTC
On Sat, 15 Apr 2023 18:07:34 +0200
Florian Smeets <flo@smeets.xyz> wrote:

> On 15.04.23 17:51, FreeBSD User wrote:
> > Am Sat, 15 Apr 2023 07:36:25 -0700
> > Cy Schubert <Cy.Schubert@cschubert.com> schrieb:  
> >>
> >> With an up-to-date tree + pjd@'s "Fix data corruption when cloning embedded
> >> blocks. #14739" patch I didn't have any issues, except for email messages
> >> with corruption in my sent directory, nowhere else. I'm still investigating
> >> the email messages issue. IMO one is generally safe to run poudriere on the
> >> latest ZFS with the additional patch.  
> 
> This is also my current observation. I have 2 hosts where I was 
> unfortunate enough to update at the wrong time. I currently *think* that 
> I'm *not* seeing data corruption with head from April 12th and this 
> patch 
> https://github.com/openzfs/zfs/commit/d3a6e5ca3b2f684132238ca968bf0b96f17ec7e1.diff 
> applied.
> 
> One pool has been upgraded with feature@block_cloning and the other hasn't.
> > 
> > FreeBSD 14.0-CURRENT #8 main-n262175-5ee1c90e50ce: Sat Apr 15 07:57:16 CEST 2023 amd64
> > 
> > The box is crashing while trying to update ports with the well known issue:
> > 
> > Panic String: VERIFY(!zil_replaying(zilog, tx)) failed
> >   
> On the pool that has block_cloning enabled I see the above insta panic 
> when poudriere starts building. I found a workaround though:
> 
> --- /usr/local/share/poudriere/include/fs.sh.orig	2023-04-15 
> 18:03:50.090823000 +0200
> +++ /usr/local/share/poudriere/include/fs.sh	2023-04-15 
> 18:04:04.144736000 +0200
> @@ -295,7 +295,6 @@
>   		fi
> 
>   		zfs clone -o mountpoint=${mnt} \
> -			-o sync=disabled \
>   			-o atime=off \
>   			-o compression=off \
>   			${fs}@${snap} \
> 
> With this workaround I was able to build thousands of packages without 
> panics or failures due to data corruption.

Thanks for this. I'll test this next week. A one should be able to test
this by hand to capture a dump.

> 
> Florian



-- 
Cheers,
Cy Schubert <Cy.Schubert@cschubert.com>
FreeBSD UNIX:  <cy@FreeBSD.org>   Web:  https://FreeBSD.org
NTP:           <cy@nwtime.org>    Web:  https://nwtime.org

			e^(i*pi)+1=0