This is a discussion on Software RAID 5 - a long story within the Linux Administration forums, part of the Linux Forums category; Hello -- I'll try to keep this as brief as I can. A while back I created an LVM array ...
|
|||||||
| FAQ | Members List | Calendar | Search | Today's Posts | Mark Forums Read |
|
|||
|
Hello --
I'll try to keep this as brief as I can. A while back I created an LVM array from 4 1TB Maxtor externals. This LVM was to serve as a temporary data dump, while two of my internal HDs, were RMA'd. During the waiting period, one of the Maxtors failed, and I lost my data. Having learned my lesson, I pulled the Maxtor for RMA, and resconstructed the three remaining as a RAID5 array, then created an LVM volume with it. Lovely. When the failed Maxtor's replacement returned, I decided to grow the RAID5 array with it. Everything was going fine until one of the other Maxtors decided to die on me as well. Needless to say I was unimpressed. The newly-failed unit has been pulled and RMA'd , but I notice that whenever the just-replaced unit falls asleep, it "falls off" the SCSI chain and disappears. Usually, I reboot and then re-assemble the array (this is only temporary after all), but this time I decided to re-add the device after switching it off then back on again. Its designation changed from /dev/sdn1 to /dev/sdq1. However, mdadm added it as a spare device, instead of a "normal" raid component, and of course updated its superblock accordingly. Now, I can't start the array at all. mdadm --examine /dev/sdn1 gives: /dev/sdn1: Magic : a92b4efc Version : 00.90.00 UUID : 37e70d06:2fd37f7b:2b41430b:610aafc7 Creation Time : Sun Aug 19 00:33:28 2007 Raid Level : raid5 Used Dev Size : 976767936 (931.52 GiB 1000.21 GB) Array Size : 2930303808 (2794.56 GiB 3000.63 GB) Raid Devices : 4 Total Devices : 4 Preferred Minor : 5 Update Time : Sun Nov 4 21:44:14 2007 State : clean Active Devices : 2 Working Devices : 3 Failed Devices : 1 Spare Devices : 1 Checksum : 6575e3e1 - correct Events : 0.705392 Layout : left-symmetric Chunk Size : 32K Number Major Minor RaidDevice State this 4 65 1 4 spare 0 0 8 241 0 active sync /dev/sdp1 1 1 0 0 1 faulty removed 2 2 8 225 2 active sync /dev/sdo1 3 3 0 0 3 faulty removed 4 4 65 1 4 spare "Total Devices" should be 3 not 4, since no spares were specified at creation time. The drive contents have not been modified, so what I need to do is change the major/minor number of the drive in the MD superblock. Is this possible? The "real" major and minor numbers are: /dev/sdn(8,208) /dev/sdn1(8,209) All help gratefully received, Rohan Beckles rohan.beckles@virgin.net |
|
|||
|
Rohan Beckles wrote:
> Hello -- > > I'll try to keep this as brief as I can. > > A while back I created an LVM array from 4 1TB Maxtor externals. This LVM > was to serve as a temporary data dump, while two of my internal HDs, were > RMA'd. > > During the waiting period, one of the Maxtors failed, and I lost my data. > Having learned my lesson, I pulled the Maxtor for RMA, and resconstructed > the three remaining as a RAID5 array, then created an LVM volume with it. > Lovely. > > When the failed Maxtor's replacement returned, I decided to grow the RAID5 > array with it. Everything was going fine until one of the other Maxtors > decided to die on me as well. Needless to say I was unimpressed. > > The newly-failed unit has been pulled and RMA'd , but I notice that > whenever the just-replaced unit falls asleep, it "falls off" the SCSI > chain and disappears. > > Usually, I reboot and then re-assemble the array (this is only temporary > after all), but this time I decided to re-add the device after switching > it off then back on again. Its designation changed from /dev/sdn1 > to /dev/sdq1. However, mdadm added it as a spare device, instead of > a "normal" raid component, and of course updated its superblock > accordingly. > > Now, I can't start the array at all. mdadm --examine /dev/sdn1 gives: > > /dev/sdn1: > Magic : a92b4efc > Version : 00.90.00 > UUID : 37e70d06:2fd37f7b:2b41430b:610aafc7 > Creation Time : Sun Aug 19 00:33:28 2007 > Raid Level : raid5 > Used Dev Size : 976767936 (931.52 GiB 1000.21 GB) > Array Size : 2930303808 (2794.56 GiB 3000.63 GB) > Raid Devices : 4 > Total Devices : 4 > Preferred Minor : 5 > > Update Time : Sun Nov 4 21:44:14 2007 > State : clean > Active Devices : 2 > Working Devices : 3 > Failed Devices : 1 > Spare Devices : 1 > Checksum : 6575e3e1 - correct > Events : 0.705392 > > Layout : left-symmetric > Chunk Size : 32K > > Number Major Minor RaidDevice State > this 4 65 1 4 spare > > 0 0 8 241 0 active sync /dev/sdp1 > 1 1 0 0 1 faulty removed > 2 2 8 225 2 active sync /dev/sdo1 > 3 3 0 0 3 faulty removed > 4 4 65 1 4 spare > > > "Total Devices" should be 3 not 4, since no spares were specified at > creation time. The drive contents have not been modified, so what I need > to do is change the major/minor number of the drive in the MD superblock. > Is this possible? The "real" major and minor numbers are: > /dev/sdn(8,208) /dev/sdn1(8,209) > > > > All help gratefully received, > I think, if you created a lvm, you'd want to examine the lvm uuid of the drive as well (and set it again, according to the /etc/lvm/backup/<volume_group> file)? Btw., which checksum should the partition have according to mdadm.conf? -- vista policy violation: Microsoft optical mouse found penguin patterns on mousepad. Partition scan in progress to remove offending incompatible products. Reactivate MS software. Linux 2.6.17mm,Xorg7.2/nvidia [LinuxCounter#295241,ICQ#4918962] |
![]() |
| Thread Tools | |
| Display Modes | |
|
|