Yeah I dunno how you got that to happen, but you can't really force it to do otherwise. Maybe MD1 being your /tmp it was so unused when the drive got reconnected it actually looked like the newer copy.
Despite what you posted above maybe your reading it backwards? On mine for example, despite the drives being listed as hde, hdg for both raids I still get this from looking at /proc/mdstat
md1 : active raid1 hde5 hdg5
1999744 blocks [2/2] [UU]
md0 : active raid1 hdg2 hde2
76150720 blocks [2/2] [UU]
the hde, and hdg are reversed for each raid and I don't really know why.
Last edited by KnightHawk; 12-07-2005 at 08:00 PM.