On Thu, 2018-04-12 at 16:32 -0500, Benjamin Marzinski wrote: > On Thu, Apr 12, 2018 at 10:22:19PM +0200, Martin Wilck wrote: > > > > But I suppose I'm still missing your point. > > Any failure in ev_add_path() before the call to domap() could cause > multipath device creation to fail, without triggering this. For > instance > add_map_with_path() or setup_map() can fail for a host of reasons. OK, got it, finally. The semantics of my failed marker were intended to be exactly what they are now: we tried DM_DEVICE_CREATE for this WWID, and failed. Other semantics are certainly possible, but much harder to define cleanly. I believe that catches a large portion of the real-world failures. And in many cases, this failure means that one or more devices are busy. (Oh, it could also mean that we tried to set 'hwhandler "1 alua"' for a device that doesn't support ALUA, but I digress ;-) Martin -- Dr. Martin Wilck <mwilck@xxxxxxxx>, Tel. +49 (0)911 74053 2107 SUSE Linux GmbH, GF: Felix Imendörffer, Jane Smithard, Graham Norton HRB 21284 (AG Nürnberg) -- dm-devel mailing list dm-devel@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/dm-devel
