Trying to install Centos6 on my system (it's been running C5 for years). I've got a pair of drives, so I'm trying to use RAID-1.
Previously I used a procedure where each partition was part of a separate RAID device, but this time I'm trying the HOWTO from the Centos WIKI on making a partitionable RAID pair.
I've been having a lot of problems, but have also been working thru them.
but t his latest one has me stumped, I'd appreciate some advice.
Booting encounters a panic when trying to mount the RAID subsystem. you can see an image of it at users.rcn.com/fredricksmith/panic.JPG.
Looks as if it's trying to boot from md127. there ain't no such critter, where'd it get the idea it could boot from there?
or more importantly, how can I solve it?
thanks!
I may have to revert to the other method, but I figured this one was worth a try.
Fred
On Sat, Feb 23, 2013 at 9:46 AM, Fred fred.fredex@gmail.com wrote:
Previously I used a procedure where each partition was part of a separate RAID device, but this time I'm trying the HOWTO from the Centos WIKI on making a partitionable RAID pair.
Although the instructions work, there are pitfalls when one of the disk fails.
I sought help in this mailing list way back in June/July 2012 time frame. I would suggest you search the list archives of that period.
I did update dracut to the latest version but removing and restoring a failed disk would not boot.
Posted the problem in the madm mailing list as well with no resolution.
or more importantly, how can I solve it?
See below.
I may have to revert to the other method, but I figured this one was worth a try.
I second the above option, if you have the luxury. In the "other" method, the system at least boots with one disk gone.
-- Arun Khan
On Sat, Feb 23, 2013 at 2:16 AM, Arun Khan knura9@gmail.com wrote:
On Sat, Feb 23, 2013 at 9:46 AM, Fred fred.fredex@gmail.com wrote:
Previously I used a procedure where each partition was part of a separate RAID device, but this time I'm trying the HOWTO from the Centos WIKI on making a partitionable RAID pair.
Although the instructions work, there are pitfalls when one of the disk fails.
I sought help in this mailing list way back in June/July 2012 time frame. I would suggest you search the list archives of that period.
I did update dracut to the latest version but removing and restoring a failed disk would not boot.
Posted the problem in the madm mailing list as well with no resolution.
or more importantly, how can I solve it?
See below.
I may have to revert to the other method, but I figured this one was
worth
a try.
I second the above option, if you have the luxury. In the "other" method, the system at least boots with one disk gone.
Ah, thanks for the tip. Guess I'll quit banging my head on the wall and
use the same method I did before.
Fred
On Sat, Feb 23, 2013 at 8:45 AM, Fred fred.fredex@gmail.com wrote:
On Sat, Feb 23, 2013 at 2:16 AM, Arun Khan knura9@gmail.com wrote:
On Sat, Feb 23, 2013 at 9:46 AM, Fred fred.fredex@gmail.com wrote:
Previously I used a procedure where each partition was part of a
separate
RAID device, but this time I'm trying the HOWTO from the Centos WIKI on making a partitionable RAID pair.
I use LVM on top of my softraid arrays. boot is generally the only physical partition other than the LVM physical volume.
Although the instructions work, there are pitfalls when one of the disk fails.
I sought help in this mailing list way back in June/July 2012 time frame. I would suggest you search the list archives of that period.
I did update dracut to the latest version but removing and restoring a failed disk would not boot.
Posted the problem in the madm mailing list as well with no resolution.
or more importantly, how can I solve it?
The name of the softraid device is not what your initial ramdisk expects.
I noticed that one of my MDs was detected as md127 instead of md2 on a recent install of 6.3. But my softraid array was not the root, so I stopped it, re-assembled it and updated mdadm.conf.
Check your mdadm.conf config and compare it to the running softraid. You'll likely have to boot to a rescue cd, assemble the raid array(s), update your mdadm.conf, change root your CentOS install, and rebuild your initial ramdisk (man mkinitrd).
See below.
I may have to revert to the other method, but I figured this one was
worth
a try.
I second the above option, if you have the luxury. In the "other" method, the system at least boots with one disk gone.
Ah, thanks for the tip. Guess I'll quit banging my head on the wall and
use the same method I did before.
Fred _______________________________________________ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
-
Arun Khan -
Fred -
SilverTip257