[Bug 665405] Re: Bad sector on disk failing mdadm RAID array and marking all disks as spares.
Launchpad Bug Tracker
665405 at bugs.launchpad.net
Sun Mar 12 22:00:19 UTC 2017
Status changed to 'Confirmed' because the bug affects multiple users.
** Changed in: mdadm (Ubuntu)
Status: New => Confirmed
--
You received this bug notification because you are a member of Ubuntu
Foundations Bugs, which is subscribed to mdadm in Ubuntu.
https://bugs.launchpad.net/bugs/665405
Title:
Bad sector on disk failing mdadm RAID array and marking all disks as
spares.
Status in mdadm package in Ubuntu:
Confirmed
Bug description:
Binary package hint: mdadm
Description: Ubuntu 10.04.1 LTS
Release: 10.04
64bit
mdadm - v2.6.7.1 - 15th October 2008
I have a failing hard drive that is in a 3 disk RAID 5 software array.
I am using mdadm to create/manage the array. When a sector goes bad on
the disk it freezes Ubuntu. Upon restart the array becomes degraded
and says that there is not enough devices to start the array. If I
boot to a live CD and install mdadm and check the status of the array.
It has all 3 drives marked as spare drives.
The following is how I correct the problem. While still login to the
live CD I run a badblocks scan (badblocks -nsv) on the failing drive.
At this point I believe that the drive sees the bad block and marks it
as such. I then shutdown and remove the failing drive's SATA cable.
Then I boot back into the live CD install mdadm and force assemble the
array. I then shut back down and reconnect the failing drive's SATA
cable. The array boots degraded then I re-add the failing drive back
into the array to get rebuilt.
What i expected to happen is that Ubuntu marks the sector or drive as failed and does not freeze. If it just marks it as failed then i can run a badblocks scan on the drive and the drive can remap the bad sector. At which point I can re-add the drive back to the array all while Ubuntu is still up and running
To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/mdadm/+bug/665405/+subscriptions
More information about the foundations-bugs
mailing list