Hi -
Having a problem removing a failed disk from a software raid5 array on a
server here. There was a hot spare configured and the raid has already
rebuilt onto it. The contents of /proc/mdstat:
Personalities : [linear] [raid0] [raid1] [raid5]
read_ahead 128 sectors
md0 : active raid5 sde1[1] sdd1[2] sdc1[1](F) sdb1[0] 8377728 blocks level
5, 32k chunk, algorithm 0 [3/3] [UUU]
...
I would have expected the unit numbers to change once the rebuild was
complete. (?) The log does show the rebuild completing and the new device
list show a new device order, but this doesn't reflect in mdstat.
Raidhotremove now claims that the disk is busy. Rebooting or stopping
the md set is not an option, but the server is due for a phased switchover
to a new system soonish.
Versions:
RH Linux 5.1
2.0.35 with raid patches current around November 9th, 1998.
Hardware is Compaq Proliant 6500R, disks are housed in the hot-swap bay,
SCSI is NCR 53c875.
I'd appreciate any suggestions. Please CC: my address if possible.
Regards
-------------------------------------------------------------------------
Louis Mandelstam Technical Manager
Infoline (Pty.) Ltd.
email: [EMAIL PROTECTED] http://www.infoline.web.za
