Re: changes to the zfs boot (was: Re: git: 72a1cb05cd23 - main - rc(8): Add a zpoolupgrade rc.d script)

From: Alexander Leidinger <Alexander_at_leidinger.net>
Date: Wed, 09 Nov 2022 20:15:23 UTC
Quoting "Patrick M. Hausen" <pmh@hausen.com> (from Wed, 9 Nov 2022  
21:02:52 +0100):

> Hi,
>
>> Am 09.11.2022 um 20:58 schrieb Alexander Leidinger  
>> <Alexander@leidinger.net>:
>>
>> Quoting "Patrick M. Hausen" <pmh@hausen.com> (from Wed, 9 Nov 2022  
>> 20:49:37 +0100):
>>
>>> Hi,
>>>
>>>> Am 09.11.2022 um 20:45 schrieb Alexander Leidinger  
>>>> <Alexander@leidinger.net>:
>>>> But "zpool set feature@edonr=enabled rpool" (or any other feature  
>>>> not in the list we talk about) would render it unbootable.
>>>
>>> Sorry, just to be sure. So an active change of e.g. checksum or  
>>> compression algorithm
>>> might render the system unbootable but a zpool upgrade never will?  
>>> At least not intentionally? ;-)
>>
>> If you mean "zpool upgrade", then no (modulo bugs). OpenZFS uses  
>> the feature flags instead of zpool upgrade.
>
> I know about feature flags and all my pools are recent enough to have them.
>
> Yet, I made it a habit to whenever I see this message:
>
> -----------
> status: Some supported features are not enabled on the pool. The pool can
> 	still be used, but some features are unavailable.
> action: Enable all features using 'zpool upgrade'. Once this is done,
> 	the pool may no longer be accessible by software that does not support
> 	the features. See zpool-features(7) for details.
> -----------
>
> to do a "zpool upgrade" after some time of burn in followed by an  
> update of the
> boot loader.
>
> I desire to know if that is in fact dangerous.

Ugh. This changed. It is indeed dangerous now. I just tested it with a  
non-root pool which didn't had all flags enabled. "zpool upgrade  
<pool>" will now enable all features.

I remember that it wasn't in the past and I had to enable the feature  
flags by hand. I don't know if a pool with bootfs set is behaving  
differently, but I consider testing this with a real rpool to be  
dangerous.

Bye,
Alexander.

-- 
http://www.Leidinger.net Alexander@Leidinger.net: PGP 0x8F31830F9F2772BF
http://www.FreeBSD.org    netchild@FreeBSD.org  : PGP 0x8F31830F9F2772BF