On lun, 2005-06-13 at 14:36 -0500, Mike Christie wrote: > Christophe Varoqui wrote: > > > > rport-0:0-0: blocked FC remote port time out: removing target > > rport-0:0-1: blocked FC remote port time out: removing target > > I think if we end up doing this to all the devices paths we are going to > be screwed. If all the paths suddenly fail and then those devices are > freed when trying to readd/create devices when the ports come back is > going to take a lot of GFP_KERNELs. Right. One thing I don't understand in the current scheme is why io are not rerouted by the DM multipath target between the LOOP_DOWN and the rport target removal ... If io could get rerouted within a few seconds, we could set a target removal multi-minutes timer, which should cover most reasonable SAN failure scenarii. Regards, -- christophe varoqui <christophe.varoqui@xxxxxxx>