Koozali.org: home of the SME Server
Obsolete Releases => SME Server 7.x => Topic started by: bcdaus on June 11, 2006, 02:22:22 PM
-
Hi,
I am just installing SME7rc3 on a Dell Precision series with 2 250Gb SATA drives, when one of the drive had a hardware failure and dropped its bundle. It is currently Sunday night of a holdiay weekend here, and I need to have some semblence of a server running for Tuesday morning. Dell will send out a new drive (arrival on Wednesday morning !!!), but what I want to do is setup SME7 as a RAID install with one drive degraded.
I tried leaving the faulty drive in, but the installer will hang with a stack of I/O errors.
Any advice on how to 'fake' the installer into using a RAID setup with only one drive ??
Thanks,
Bill
-
Answering my own questions here....
Looks like when I drop the replacement second drive in it will be added to the existing MD1 raid array. This is based on what I see in the Disk Redundancy Status page of the console.
Can someone please confirm ?
Thanks,
Bill.
-
It worked for me.
Thats the best I can offer. I didnt disect the process, I only did it the once with RC2 and I havent done it again nor followed it through. I certainly havent needed to revisit it sevral months later.
HTH
P
-
with only 1 drive a raid install happens automatically
-
Got my second drive and stuck it in today. Found a how-to on how to add and re-assemble a RAID1 array which worked. All semmed to go well, but upon reboot (for something else) we are back to running a single drive with no available devices in the Disk Redundancy pane.
Is it possible to GROW a single drive install to be a 2 drive RAID 1 setup without doing a backup / clean install and a restore ?? If I do how can I make it persistant ?
With Thanks,
Bill.
-
Found a how-to on how to add and re-assemble a RAID1 array which worked...
You'll probably have to tell us what HOWTO you followed. If you followed a howto for v6, it probably won't work on v7. You shouldn't have to add a raid, it should already be there. This indicates you followed the wrong instructions.
Is it possible to GROW...
yes. The raids should already be there in degraded mode. You can get an overview of the process here:
http://no.longer.valid/phpwiki/index.php/lvm
In short, you have to:
-make sure the new disk is exactly the same as the existing disk
-partition the new disk
-add the new disk to the existing raids.
You'll have to come back with:
-what HOWTO you followed
-the output of: cat /proc/mdstat
-
Your RAID should already be there... by rebuilding it using the RAID controller it would have implemented RAID for you...
I suggest you do the following;
Reboot your Server
CTRL+A (or what ever it is) into your RAID controller interface
Check the RAID configuration and ensure it is "Optimal"
If it is then it should all be fine.
You won't see any "Spare Disks" in SME because SME will see the RAID as a single volume.
I think that your server RAID is most likely perfectly healthy
-
he has a software raid, not a hardware raid
-
Sorry for the delay in getting back. Its a very busy week at the moment.
The how-to was one that Darrell May had written regarding adding additional RAID devices to v4.1.2 found down at the bottom of http://forums.contribs.org/index.php?topic=31847.0. I took this as an outline of the process and made use of mdadm --help to get me to :
mdadm /dev/md2 --add /dev/sdb2
mdadm /dev/md1 --add /dev/sdb1
after reading the partition table from SDA and using fdisk to create the same layout on SDB.
The array re-built as it should, problem is that on reboot my logs show that the Partition table on SDB (second drive) is not recognized.
Extract from DMESG:
scsi1 : ata_piix
Vendor: ATA Model: Maxtor 7Y250M0 Rev: YAR5
Type: Direct-Access ANSI SCSI revision: 05
SCSI device sda: 488281250 512-byte hdwr sectors (250000 MB)
SCSI device sda: drive cache: write back
SCSI device sda: 488281250 512-byte hdwr sectors (250000 MB)
SCSI device sda: drive cache: write back
sda: sda1 sda2
Attached scsi disk sda at scsi0, channel 0, id 0, lun 0
Vendor: ATA Model: WDC WD2500JS-75N Rev: 10.0
Type: Direct-Access ANSI SCSI revision: 05
SCSI device sdb: 488281250 512-byte hdwr sectors (250000 MB)
SCSI device sdb: drive cache: write back
SCSI device sdb: 488281250 512-byte hdwr sectors (250000 MB)
SCSI device sdb: drive cache: write back
sdb: unknown partition table
Attached scsi disk sdb at scsi0, channel 0, id 1, lun 0
device-mapper: 4.5.0-ioctl (2005-10-04) initialised: dm-devel@redhat.com
md: raid1 personality registered as nr 3
md: Autodetecting RAID arrays.
md: autorun ...
md: considering sda2 ...
md: adding sda2 ...
md: sda1 has different UUID to sda2
md: created md2
md: bind<sda2>
md: running: <sda2>
raid1: raid set md2 active with 1 out of 2 mirrors
md: considering sda1 ...
md: adding sda1 ...
md: created md1
md: bind<sda1>
md: running: <sda1>
raid1: raid set md1 active with 1 out of 2 mirrors
md: ... autorun DONE.
md: Autodetecting RAID arrays.
md: autorun ...
md: ... autorun DONE.
I ran fdisk again and was able to read the partition table from the disk and then write it back. I then used the mdadm commands as above to rebuild the array.
/proc/mdstat from before re-building the array was :
[root@betty /]# cat /proc/mdstat
Personalities : [raid1]
md2 : active raid1 sda2[0]
244035264 blocks [1/2] [U_]
md1 : active raid1 sda1[0]
104320 blocks [1/2] [U_]
unused devices: <none>
[root@betty /]#
I am concerened about the partition table on SDB is not sticking after a reboot and the fact that /proc/mdstat doesn't recognise /dev/sdb as a potential unused device.
Thanks,
Bill.
-
Should I log this in Bug Tracker ? Looks like the partition table of any additional drives is not being recognised. Just noticed it also happened to a external USB drive we were using for backup. When I used fdisk to read the table then write it back we got the drive back running.
Very Odd.
When I get some time early next week I will do a 911 backup, then clean re-install with both of the new drives in software RAID config, rather than replacing the Maxtors one at a time and rebuilding, as I don't trust it not to bork at the worst possible time..
Bill.
-
Bill
If you are not sure that something is working as it should, log a bug (after checking that a similar bug doesn't already exist).
Better that your suspected bugs logged and ruled out rather than leave it an let a problem get into the final code that could have been sorted.
(I never seem to give answers, just point people to the bug tracker :( )
Dave