linux multipath is making me tired.

So, I’ve got a multipath setup that claims to be working:

# multipath -ll
vrp (360060e8010053b90052fb06900000190) dm-8 HITACHI,DF600F
size=70G features=’0′ hwhandler=’0′ wp=rw
`-+- policy=’round-robin 0′ prio=1 status=active
|- 6:0:0:0 sdc 8:32 active ready running
`- 5:0:0:0 sdb 8:16 active ready running

sweet right?

Until you test it. For some reason, the above setup isn’t ’round-robin’, it’s only using one path. If you kill the inactive link (unplug, port disable, whatever), it works fine. Reconnect, and it recovers. If you kill the active link, well, the world blows up on you (read-only file system, etc).

The above is with Oracle Enterprise Linux 5.6 (aka oracle unbreakable linux) and device-mapper-multipath-0.4.9-23.0.8.el5 (which is the latest supported version AFAICT). I’ll update or post again when/if I get this solved.

ETA: Turns out to be a known issue