[dm-devel] multipath-tools 0.7.4 failure to remove device

[Martin Wilck]

On Thu, 2018-01-18 at 09:22 -0600, Benjamin Marzinski wrote:
> On Thu, Jan 18, 2018 at 09:11:02AM +0100, Martin Wilck wrote:
> > On Wed, 2018-01-17 at 21:50 -0600, Benjamin Marzinski wrote:
> > 
> > > Now, I don't see how this could cause a boot to fail since this
> > > is a
> > > new
> > > device that isn't getting set up, but I really never test this,
> > > because
> > > when we create the initramfs in RedHat based distros, we edit the
> > > multipath.conf file in it to blacklist all devices except those
> > > needed
> > > by the initramfs.
> > 
> > I'm a bit puzzled. It's against my personcal conviction that the
> > blacklist must be the same between initrd and root FS. How do you
> > prevent LVM or other layers grabbing blacklisted devices during
> > initramfs processing, so that they can't be multipathed any more
> > later?
> 
> lvm only assembles on the devices necessary to boot in the initramfs
> also.

And MD? And other layers such as dm-crypt? I haven't thought it all
through, but I feel there's a lot more stuff that might grab a block
device once it appears. But hey, if that's not the case, that's good
news. The set of dracut modules is finite, so I guess it can actually
be reviewed, at least in a given distribution framework.

IIUC this requires hardcoding the WWIDs to be activated in the initrd,
which also has some disadvantages.

Anyway, making the initramfs and booted system play together nicely is
not an upstream issue for multipath-tools. I guess this will remain an
area where distributions differentiate.

Btw, I've been considering whether the dracut multipath module should
rather be maintained in multipath-tools than in dracut. It would make
certain changes easier. Opinions welcome.

Regards,
Martin

-- 
Dr. Martin Wilck <mwilck at suse.com>, Tel. +49 (0)911 74053 2107
SUSE Linux GmbH, GF: Felix Imendörffer, Jane Smithard, Graham Norton
HRB 21284 (AG Nürnberg)