I've had a software RAID1 array in production for over a year which has LVM partitions on top of /dev/md0. I rebooted over the weekend to apply some kernel patches and now the array won't come up. Getting the "Continue to wait; or Press S to skip mounting or M for manual recovery" on boot. I hit M, login as root, and the RAID array is up, but none of the LVM partitions are available. It's like everything is gone. I stopped the array and brought it up on a single disk (it's RAID1) with --run. Ok, the lvm stuff is there now. So I added a new disk and add it to the degraded array. It starts rebuilding. I do an fdisk of the new disk I just added and there's a brand new partition there of type 'Linux LVM'. I did not add that partition. What's going on? I'm not even using partitions, I'm just using the raw devices.
Answer
Only way I could get the software raid array stable was to leave LVM off and hard-partition. I tried using both raw devices, as well as type 0xFD partitions and as soon as I tried using LVM across /dev/md0, the partition types would automatically change from 0xFD to "LVM" on all the raid members. Very, very strange. I've been using LVM over Linux software RAID for nearly a decade and have never seen this problem before. I'm buying an Areca card.
No comments:
Post a Comment