Hi,
Use a "raidhotadd" to add in the partition on the newly replaced failed
drive.
For example: md0 is made from sda1 and sdb1. The sdb drive fails, so md0
comes up in degraded mode running off of sda. When you add in a new sdb
drive the raid driver does not bother with it, but continues to run in
degraded mode. To have it recognize the new working sdb drive, say
"raidhotadd /dev/md0 /dev/sdb1" and it will background reconstruct, then
come up running in non-degraded mode.
I don't like the "mkraid --force-resync" because I never know which drive
will end up being the master. I'd really hate to have my data drive
"resynced" to the binary garbage of a new replacement drive.
I also recommend upgrading to the newest kernel raid driver.
- David Harris
Principal Engineer, DRH Internet Services
-----Original Message-----
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED]] On Behalf Of Brian Leeper
Sent: Monday, December 14, 1998 10:19 AM
To: [EMAIL PROTECTED]
Subject: Restoring a mirrored array?
What is the recommended procedure for restoring a mirror after one of the
drives fails? I'm using the latest patches for 2.0.35,
raid0145-19981110-2.0.35 and raidtools-19981201-0.90.tar.gz
I've used mkraid --force-resync, but when this is done I run e2fsck on the
filesystem and I get a lot of errors, with some stuff left in lost+found.
/etc/raidtab is correct when this is done--was the same file used to
create the mirror.
Is there some other way of doing this?
Brian
