[Bug 316670] Re: Raid10 array not assembled at boot, all disks are marked as spare

[Steve Beattie]

Riasing the priority and nominating for jaunty.

** Description changed:

  I have just installed jaunty with a separate boot partition and / on a
  raid10 array (/dev/md1).
  
  However when I boot, I'm dropped in a initramfs shell and I have to
  manually assemble the array as all disks are marked as spare. There are
  also segfaults in dmesg. mdadm[1370]: segfault at 0 ip 000000000040839b
  sp 00007fffb2af2620 error 4 in mdadm[400000+2a000]
+ 
+ The segfault is easily reproducable; steps to do so:
+ 
+ 1) create 4 empty loopback disks:
+ 
+   for i in $(seq 1 4) ; do
+       dd if=/dev/zero of=disk${i} bs=1024 count=512000
+       sudo losetup /dev/loop${i} disk${i}
+   done
+ 
+ 2) create raid10 array
+ 
+   sudo mdadm -C /dev/md0 -l raid10 -n 4 /dev/loop{1,2,3,4}
+ 
+ 3) stop the array:
+ 
+   sudo mdadm --stop /dev/md0
+ 
+ 4) incrementally assemble the first loop device; this is how udev
+ invokes mdadm in /lib/udev/rules.d/85-mdadm.rules.
+ 
+   sudo mdadm --incremental /dev/loop1
+ 
+ et voila, segv.

** Description changed:

  I have just installed jaunty with a separate boot partition and / on a
  raid10 array (/dev/md1).
  
  However when I boot, I'm dropped in a initramfs shell and I have to
  manually assemble the array as all disks are marked as spare. There are
  also segfaults in dmesg. mdadm[1370]: segfault at 0 ip 000000000040839b
  sp 00007fffb2af2620 error 4 in mdadm[400000+2a000]
  
- The segfault is easily reproducable; steps to do so:
+ The segfault is easily reproducible on jaunty without needing to do a
+ full install on a raid10 setup; steps to do so:
  
  1) create 4 empty loopback disks:
  
    for i in $(seq 1 4) ; do
        dd if=/dev/zero of=disk${i} bs=1024 count=512000
        sudo losetup /dev/loop${i} disk${i}
    done
  
  2) create raid10 array
  
    sudo mdadm -C /dev/md0 -l raid10 -n 4 /dev/loop{1,2,3,4}
  
  3) stop the array:
  
    sudo mdadm --stop /dev/md0
  
  4) incrementally assemble the first loop device; this is how udev
  invokes mdadm in /lib/udev/rules.d/85-mdadm.rules.
  
    sudo mdadm --incremental /dev/loop1
  
  et voila, segv.

** Changed in: mdadm (Ubuntu)
   Importance: Medium => High

-- 
Raid10 array not assembled at boot, all disks are marked as spare
https://bugs.launchpad.net/bugs/316670
You received this bug notification because you are a member of Kernel
Bugs, which is subscribed to Linux.