Re: [PATCH 1/1] expand last partition supports btrfs (inline test)

["Bezdeka, Florian" <florian.bezdeka@siemens.com>]

On Fri, 2022-09-30 at 12:39 +0200, Roberto A. Foglietta wrote:
> Il Ven 30 Set 2022, 12:23 Schmidl, Tobias <tobiasschmidl@siemens.com>
> ha scritto:
> > Hi Jan, Roberto,
> > 
> > Am Freitag, dem 30.09.2022 um 10:56 +0200 schrieb Jan Kiszka:
> > > On 30.09.22 10:00, Roberto A. Foglietta wrote:
> > > 
> > > 
> > >  
> > > https://groups.google.com/d/msgid/isar-users/20220630135531.717696
> > > -2-tobiasschmidl%40siemens.com> is the way forward, but it needs
> > > some more fine-tuning. Tobias, what is
> > > the status there?
> > > 
> > 
> > As I understand it the mentioned patch is dead, since the systemd
> > version
> > in stretch is too old for systemd-resizefs - or did I miss
> > anything?

Could we do that with some kind of version check? If systemd is to old
simply fall back to the previous implementation. So we support stretch
"as is" and care better about "modern" systems.

> 
> Hi Tobias and Jan,
> 
>  as you can see my approach is straight simple: if extending the ext4
> fails because it does not match the filesystem type then try with
> btrfs tool. Obviously, the btrfs tool is added as dependencies among
> others before defined.

First: resize2fs does not only support ext4. It supports all of the ext
variants (ext2-ext4) to my understanding.

Second: There might be more reasons why resize2fs could fail. IMHO, it
doesn't make sense to run a btrfs tool on a ext filesystem that failed
to expand.

Third: The mount point (/tmp/btrfs) is never cleaned up in the Robertos
patch.

> 
>  pro: it supports btrfs, easy to catch
> 
>  contro: not a general approach
> 
>  I tend to be very pragmatic: the current supports ext4 resize only +
> a simple patch it supports btrfs also. This is an improvement and add
> a little of complexity. May be tomorrow, it will arrive the support
> for xfs or dm-crypto. Step by step functionalities will be added and
> at a certain point someone will feel the nessecity to rationalise the
> code but not necessarily generalising (architecture change). 
> 
>  This kind of development is about natural evolution rather than
> intelligent design. Ok, sometimes it is necessary to reorganise,
> rewrite, generalise, the product of the natural evolution but usually
> fits enough. My 2 cents, IMHO.
> 
>  Cheers, R.
>