On Thu, 2018-04-12 at 22:16 +0200, Martin Wilck wrote:
> On Thu, 2018-04-12 at 13:33 -0500, Benjamin Marzinski wrote:
> > On Wed, Apr 04, 2018 at 06:16:17PM +0200, Martin Wilck wrote:
> > > dm_addmap_create() is where we actually try to set up a new
> > > multipath map. Depending on the result, mark the wwid as
> > > failed (or not), and re-trigger an uevent if necessary.
> > > If a path changes from multipath to non-multipath, use an "add"
> > > event to make sure LVM2 rules pick it up. Increase log level
> > > of this event to 3.
> > > 
> > 
> > By only looking at domap, we will miss instances of multipathd
> > failing
> > to create maps earlier in the process. This isn't necessarily
> > wrong.
> > It
> > just means that we can't rely on checking
> > /dev/shm/multipath/failed_wwids to definitively tell us whether
> > multipathd has tried and failed to create the device.
> Sorry, I can't follow you. Where else except in the 
> domap->dm_addmap_create->dm_addmap() code path do we create maps?
> I'm feeling stupid, I really can't see it.

If you were referring to other instances of multipathd which have
already terminated (e.g. multipathd which ran in initramfs): these
leave the failed markers under /dev/shm when they quit. That's the
whole point of this patch. A failed marker will only be removed if a)
the system is rebooted, or b) another attempt to create the map
succeeds, or c) a user removes the marker manually.

But I suppose I'm still missing your point.


Dr. Martin Wilck <mwi...@suse.com>, Tel. +49 (0)911 74053 2107
SUSE Linux GmbH, GF: Felix Imend├Ârffer, Jane Smithard, Graham Norton
HRB 21284 (AG N├╝rnberg)

dm-devel mailing list

Reply via email to