I have/had a software raid running and sdc drive failed. I got a replacement drive today and installed it. My only experience with set partitions and raids in during initail setup. I could not fdisk the new drive because i guess it wasn't reconized so I rebooted. Now I can reach the drive via fdisk but I have made more problems now (no swap now) and I'm not sure the steps to rebuild. Background: I have a FC3 with a software raid. I have 3 SCSI 18gb hard drives If I recall this how I set it up md0 /boot 100MB raid 1 sda, sdb and sdc as spare md1 /swp 768MB raid 0 sda, sdb, sdc md2 / ext3 33GB raid 5 sda, sdb, sdc
Can someone please help?
Dan Carl writes:
I have/had a software raid running and sdc drive failed. I got a replacement drive today and installed it. My only experience with set partitions and raids in during initail setup. I could not fdisk the new drive because i guess it wasn't reconized so I rebooted. Now I can reach the drive via fdisk but I have made more problems now (no swap now) and I'm not sure the steps to rebuild. Background: I have a FC3 with a software raid. I have 3 SCSI 18gb hard drives If I recall this how I set it up md0 /boot 100MB raid 1 sda, sdb and sdc as spare md1 /swp 768MB raid 0 sda, sdb, sdc md2 / ext3 33GB raid 5 sda, sdb, sdc
Can someone please help?
Your first step is to determine your actual RAID configuration. Unless you know exactly what belongs where, you're not going to get very far.
mdadm -Q -D /dev/md0
mdadm -Q -D /dev/md1
etc…
This will give you the information you need to map out how your existing drives are set up for RAID.
Your first step is to remove any failed devices, if mdadm -Q -D still shows any. Use the mdadm -r option to remove any leftover failed devices.
Once your RAID layout consists of only the working partitions, you then need to figure out how your replacement drive will be added to your RAID configuration. You'll need to figure out how it must be carved up into the different RAID partitions. Use fdisk to partition your new disk, carefully making sure the sizes of the new partitions match EXACTLY the sizes of the corresponding partition in your existing RAID volumes.
Then, once you've double, and triple-checked your new partitions, use "mdadm -a" to add each partition to its appropriate RAID volume. Linux will then automatically resync and rebuild your RAID volumes, one by one.
On Fri, 2006-07-28 at 16:29 -0500, Dan Carl wrote:
I have/had a software raid running and sdc drive failed. I got a replacement drive today and installed it. My only experience with set partitions and raids in during initail setup. I could not fdisk the new drive because i guess it wasn't reconized so I rebooted. Now I can reach the drive via fdisk but I have made more problems now (no swap now) and I'm not sure the steps to rebuild. Background: I have a FC3 with a software raid. I have 3 SCSI 18gb hard drives If I recall this how I set it up md0 /boot 100MB raid 1 sda, sdb and sdc as spare md1 /swp 768MB raid 0 sda, sdb, sdc
This toasted your /swp partition. Raid 0 is striping, and a single failure toasts the entire device.
You would have been ahead with a non-raid swap, and had 3 separate partitions, one on each device, for swap. Failure of one would not have toasted all.
md2 / ext3 33GB raid 5 sda, sdb, sdc
Can someone please help?
I know its a raid 0 is a stripe. Its my swap partition. Why would I need fault tolerance on my swap.
Anyway, I did what Sam suggested. md0 is fine, md1 doesn't exist mdadm -Q -D /dev/md2 it yeilded /dev/md2: Version : 00.90.01 Creation Time : Mon Feb 14 06:42:28 2005 Raid Level : raid5 Array Size : 34812416 (33.20 GiB 35.65 GB) Device Size : 17406208 (16.60 GiB 17.82 GB) Raid Devices : 3 Total Devices : 2 Preferred Minor : 2 Persistence : Superblock is persistent Update Time : Fri Jul 28 17:56:25 2006 State : clean, degraded Active Devices : 2 Working Devices : 2 Failed Devices : 0 Spare Devices : 0 Layout : left-symmetric Chunk Size : 256K Number Major Minor RaidDevice State 0 8 3 0 active sync /dev/sda3 1 8 19 1 active sync /dev/sdb3 2 0 0 -1 removed UUID : b4b161bc:2953b117:9c13c568:47693baa Events : 0.31307539
What if the next step this is my mail server and I really don't have the time to reload it. I have my fstab, partition, mdstat, infomation. I ran this command sfdisk -d > sdb-parts.dump before a added the new drive. Will any of this help?
Like I said before the only raid/partition experience I have is at initial installation.
----- Original Message ----- From: "Jeff Vian" jvian10@charter.net To: "For users of Fedora Core releases" fedora-list@redhat.com Sent: Friday, July 28, 2006 5:44 PM Subject: Re: software raid drive failed, please provide step bu step torebuild
On Fri, 2006-07-28 at 16:29 -0500, Dan Carl wrote:
I have/had a software raid running and sdc drive failed. I got a replacement drive today and installed it. My only experience with set partitions and raids in during initail
setup.
I could not fdisk the new drive because i guess it wasn't reconized so I rebooted. Now I can reach the drive via fdisk but I have made more problems now
(no
swap now) and I'm not sure the steps to rebuild. Background: I have a FC3 with a software raid. I have 3 SCSI 18gb hard drives If I recall this how I set it up md0 /boot 100MB raid 1 sda, sdb and sdc as spare md1 /swp 768MB raid 0 sda, sdb, sdc
This toasted your /swp partition. Raid 0 is striping, and a single failure toasts the entire device.
You would have been ahead with a non-raid swap, and had 3 separate partitions, one on each device, for swap. Failure of one would not have toasted all.
md2 / ext3 33GB raid 5 sda, sdb, sdc
Can someone please help?
-- fedora-list mailing list fedora-list@redhat.com To unsubscribe: https://www.redhat.com/mailman/listinfo/fedora-list
On Fri, 2006-07-28 at 18:03 -0500, Dan Carl wrote:
I know its a raid 0 is a stripe. Its my swap partition. Why would I need fault tolerance on my swap.
From your first post below:
Now I can reach the drive via fdisk but I have made more problems now (no swap now) and I'm not sure the steps to rebuild. That is a good reason to make sure that vital disk partitions are not made critically weak. When striping across 3 drives the failure probability is made 3X as likely and any single failure toasts the entire device.
Since swap can use multiple partitions the likelyhood of failure and total loss of swap space can be reduced by simply defining multiple swap partitions without using striping.
Anyway, I did what Sam suggested. md0 is fine, md1 doesn't exist mdadm -Q -D /dev/md2 it yeilded /dev/md2: Version : 00.90.01 Creation Time : Mon Feb 14 06:42:28 2005 Raid Level : raid5 Array Size : 34812416 (33.20 GiB 35.65 GB) Device Size : 17406208 (16.60 GiB 17.82 GB) Raid Devices : 3 Total Devices : 2 Preferred Minor : 2 Persistence : Superblock is persistent Update Time : Fri Jul 28 17:56:25 2006 State : clean, degraded Active Devices : 2 Working Devices : 2 Failed Devices : 0 Spare Devices : 0 Layout : left-symmetric Chunk Size : 256K Number Major Minor RaidDevice State 0 8 3 0 active sync /dev/sda3 1 8 19 1 active sync /dev/sdb3 2 0 0 -1 removed UUID : b4b161bc:2953b117:9c13c568:47693baa Events : 0.31307539
So mdadm -a needs to be used to add the 3rd device back to md2. Sam's instructions were clear on that. For more information and education use the man page for mdadm.
MANAGE MODE Usage: mdadm device options... devices... This usage will allow individual devices in an array to be failed, removed or added. It is possible to perform multiple operations with one command. For example: mdadm /dev/md0 -f /dev/hda1 -r /dev/hda1 -a /dev/hda1 will firstly mark /dev/hda1 as faulty in /dev/md0 and will then remove it from the array and finally add it back in as a spare. However only one md array can be affected by a single command.
I would do the following that you have not already stated done. 1. create the partition(s) on your new /dev/hdc 2. use mdadm as follows to add it to md2 mdadm /dev/md2 -a /dev/sdc3 note that I assume your partitions are created and numbered as you have already stated.
What if the next step this is my mail server and I really don't have the time to reload it. I have my fstab, partition, mdstat, infomation. I ran this command sfdisk -d > sdb-parts.dump before a added the new drive. Will any of this help?
fdisk -l will list the partition information for each drive including start and end cylinders such as this.
[root@raptor pgsql]# fdisk -l /dev/hda
Disk /dev/hda: 250.0 GB, 250059350016 bytes 255 heads, 63 sectors/track, 30401 cylinders Units = cylinders of 16065 * 512 = 8225280 bytes
Device Boot Start End Blocks Id System /dev/hda1 * 1 21 168651 83 Linux /dev/hda2 22 532 4104607+ 83 Linux /dev/hda3 533 721 1518142+ 82 Linux swap /dev/hda4 722 30401 238404600 5 Extended /dev/hda5 722 1359 5124703+ 83 Linux ....
From that you can get not only the size of each partition, but the
actual cylinders used and can recreate the table on the new drive appropriately with fdisk.
Like I said before the only raid/partition experience I have is at initial installation.
----- Original Message ----- From: "Jeff Vian" jvian10@charter.net To: "For users of Fedora Core releases" fedora-list@redhat.com Sent: Friday, July 28, 2006 5:44 PM Subject: Re: software raid drive failed, please provide step bu step torebuild
On Fri, 2006-07-28 at 16:29 -0500, Dan Carl wrote:
I have/had a software raid running and sdc drive failed. I got a replacement drive today and installed it. My only experience with set partitions and raids in during initail
setup.
I could not fdisk the new drive because i guess it wasn't reconized so I rebooted. Now I can reach the drive via fdisk but I have made more problems now
(no
swap now) and I'm not sure the steps to rebuild. Background: I have a FC3 with a software raid. I have 3 SCSI 18gb hard drives If I recall this how I set it up md0 /boot 100MB raid 1 sda, sdb and sdc as spare md1 /swp 768MB raid 0 sda, sdb, sdc
This toasted your /swp partition. Raid 0 is striping, and a single failure toasts the entire device.
You would have been ahead with a non-raid swap, and had 3 separate partitions, one on each device, for swap. Failure of one would not have toasted all.
md2 / ext3 33GB raid 5 sda, sdb, sdc
Can someone please help?
-- fedora-list mailing list fedora-list@redhat.com To unsubscribe: https://www.redhat.com/mailman/listinfo/fedora-list
Ok I totally see your point about striping my swap partition. I read it some howto a few years back.
I did just what you said I paritioned my new drive the exact same as the other two. I issued the command mdadm /dev/md2 -a /dev/sdc3 and the raid md2 is rebuilding as I write this.
So how do I get setup my swap as you described? Now I have three drives w/ partition 265072 sda2 265072 sdb2 265072 sdc2
Thank you soooo much Its 100 degrees, its Friday and now I have go outside and mow my lawn. Look forward to reading your reply while sipping a cold one!
----- Original Message ----- From: "Jeff Vian" jvian10@charter.net To: "For users of Fedora Core releases" fedora-list@redhat.com Sent: Friday, July 28, 2006 6:48 PM Subject: Re: software raid drive failed, please provide step bu steptorebuild
On Fri, 2006-07-28 at 18:03 -0500, Dan Carl wrote:
I know its a raid 0 is a stripe. Its my swap partition. Why would I need fault tolerance on my swap.
From your first post below:
Now I can reach the drive via fdisk but I have made more problems now (no swap now) and I'm not sure the steps to rebuild.That is a good reason to make sure that vital disk partitions are not made critically weak. When striping across 3 drives the failure probability is made 3X as likely and any single failure toasts the entire device.
Since swap can use multiple partitions the likelyhood of failure and total loss of swap space can be reduced by simply defining multiple swap partitions without using striping.
Anyway, I did what Sam suggested. md0 is fine, md1 doesn't exist mdadm -Q -D /dev/md2 it yeilded /dev/md2: Version : 00.90.01 Creation Time : Mon Feb 14 06:42:28 2005 Raid Level : raid5 Array Size : 34812416 (33.20 GiB 35.65 GB) Device Size : 17406208 (16.60 GiB 17.82 GB) Raid Devices : 3 Total Devices : 2 Preferred Minor : 2 Persistence : Superblock is persistent Update Time : Fri Jul 28 17:56:25 2006 State : clean, degraded Active Devices : 2 Working Devices : 2 Failed Devices : 0 Spare Devices : 0 Layout : left-symmetric Chunk Size : 256K Number Major Minor RaidDevice State 0 8 3 0 active sync /dev/sda3 1 8 19 1 active sync /dev/sdb3 2 0 0 -1 removed UUID : b4b161bc:2953b117:9c13c568:47693baa Events : 0.31307539
So mdadm -a needs to be used to add the 3rd device back to md2. Sam's instructions were clear on that. For more information and education use the man page for mdadm.
MANAGE MODE Usage: mdadm device options... devices... This usage will allow individual devices in an array to be failed, removed or added. It is possible to perform multiple operations with one command. For example: mdadm /dev/md0 -f /dev/hda1 -r /dev/hda1 -a /dev/hda1 will firstly mark /dev/hda1 as faulty in /dev/md0 and will then remove it from the array and finally add it back in as a spare. However only one md array can be affected by a single command.
I would do the following that you have not already stated done.
- create the partition(s) on your new /dev/hdc
- use mdadm as follows to add it to md2 mdadm /dev/md2 -a /dev/sdc3
note that I assume your partitions are created and numbered as you have already stated.
What if the next step this is my mail server and I really don't have the time to reload it. I have my fstab, partition, mdstat, infomation. I ran this command sfdisk -d > sdb-parts.dump before a added the new drive. Will any of this help?
fdisk -l will list the partition information for each drive including start and end cylinders such as this.
[root@raptor pgsql]# fdisk -l /dev/hda Disk /dev/hda: 250.0 GB, 250059350016 bytes 255 heads, 63 sectors/track, 30401 cylinders Units = cylinders of 16065 * 512 = 8225280 bytes Device Boot Start End Blocks Id System /dev/hda1 * 1 21 168651 83 Linux /dev/hda2 22 532 4104607+ 83 Linux /dev/hda3 533 721 1518142+ 82 Linux swap /dev/hda4 722 30401 238404600 5 Extended /dev/hda5 722 1359 5124703+ 83 Linux ....From that you can get not only the size of each partition, but the
actual cylinders used and can recreate the table on the new drive appropriately with fdisk.
Like I said before the only raid/partition experience I have is at initial installation.
----- Original Message ----- From: "Jeff Vian" jvian10@charter.net To: "For users of Fedora Core releases" fedora-list@redhat.com Sent: Friday, July 28, 2006 5:44 PM Subject: Re: software raid drive failed, please provide step bu step torebuild
On Fri, 2006-07-28 at 16:29 -0500, Dan Carl wrote:
I have/had a software raid running and sdc drive failed. I got a replacement drive today and installed it. My only experience with set partitions and raids in during initail
setup.
I could not fdisk the new drive because i guess it wasn't reconized so I rebooted. Now I can reach the drive via fdisk but I have made more problems now
(no
swap now) and I'm not sure the steps to rebuild. Background: I have a FC3 with a software raid. I have 3 SCSI 18gb hard drives If I recall this how I set it up md0 /boot 100MB raid 1 sda, sdb and sdc as spare md1 /swp 768MB raid 0 sda, sdb, sdc
This toasted your /swp partition. Raid 0 is striping, and a single failure toasts the entire device.
You would have been ahead with a non-raid swap, and had 3 separate partitions, one on each device, for swap. Failure of one would not have toasted all.
md2 / ext3 33GB raid 5 sda, sdb, sdc
Can someone please help?
-- fedora-list mailing list fedora-list@redhat.com To unsubscribe: https://www.redhat.com/mailman/listinfo/fedora-list
-- fedora-list mailing list fedora-list@redhat.com To unsubscribe: https://www.redhat.com/mailman/listinfo/fedora-list
On Fri, 2006-07-28 at 19:37 -0500, Dan Carl wrote:
Ok I totally see your point about striping my swap partition. I read it some howto a few years back.
I did just what you said I paritioned my new drive the exact same as the other two. I issued the command mdadm /dev/md2 -a /dev/sdc3 and the raid md2 is rebuilding as I write this.
So how do I get setup my swap as you described? Now I have three drives w/ partition 265072 sda2 265072 sdb2 265072 sdc2
First format the swap partitions. Note that fdisk should list these as type 82 Linux swap.
mkswap /dev/sda2 repeat for sdb2 and sdc2
then put into fstab the following /dev/sda2 swap swap defaults 0 0 /dev/sdb2 swap swap defaults 0 0 /dev/sdc2 swap swap defaults 0 0
Finally just issue the "swapon -a" command to enable swap as defined.
Now if one fails the remaining ones will continue to be used.
Thank you soooo much Its 100 degrees, its Friday and now I have go outside and mow my lawn. Look forward to reading your reply while sipping a cold one!
----- Original Message ----- From: "Jeff Vian" jvian10@charter.net To: "For users of Fedora Core releases" fedora-list@redhat.com Sent: Friday, July 28, 2006 6:48 PM Subject: Re: software raid drive failed, please provide step bu steptorebuild
On Fri, 2006-07-28 at 18:03 -0500, Dan Carl wrote:
I know its a raid 0 is a stripe. Its my swap partition. Why would I need fault tolerance on my swap.
From your first post below:
Now I can reach the drive via fdisk but I have made more problems now (no swap now) and I'm not sure the steps to rebuild.That is a good reason to make sure that vital disk partitions are not made critically weak. When striping across 3 drives the failure probability is made 3X as likely and any single failure toasts the entire device.
Since swap can use multiple partitions the likelyhood of failure and total loss of swap space can be reduced by simply defining multiple swap partitions without using striping.
Anyway, I did what Sam suggested. md0 is fine, md1 doesn't exist mdadm -Q -D /dev/md2 it yeilded /dev/md2: Version : 00.90.01 Creation Time : Mon Feb 14 06:42:28 2005 Raid Level : raid5 Array Size : 34812416 (33.20 GiB 35.65 GB) Device Size : 17406208 (16.60 GiB 17.82 GB) Raid Devices : 3 Total Devices : 2 Preferred Minor : 2 Persistence : Superblock is persistent Update Time : Fri Jul 28 17:56:25 2006 State : clean, degraded Active Devices : 2 Working Devices : 2 Failed Devices : 0 Spare Devices : 0 Layout : left-symmetric Chunk Size : 256K Number Major Minor RaidDevice State 0 8 3 0 active sync /dev/sda3 1 8 19 1 active sync /dev/sdb3 2 0 0 -1 removed UUID : b4b161bc:2953b117:9c13c568:47693baa Events : 0.31307539
So mdadm -a needs to be used to add the 3rd device back to md2. Sam's instructions were clear on that. For more information and education use the man page for mdadm.
MANAGE MODE Usage: mdadm device options... devices... This usage will allow individual devices in an array to be failed, removed or added. It is possible to perform multiple operations with one command. For example: mdadm /dev/md0 -f /dev/hda1 -r /dev/hda1 -a /dev/hda1 will firstly mark /dev/hda1 as faulty in /dev/md0 and will then remove it from the array and finally add it back in as a spare. However only one md array can be affected by a single command.
I would do the following that you have not already stated done.
- create the partition(s) on your new /dev/hdc
- use mdadm as follows to add it to md2 mdadm /dev/md2 -a /dev/sdc3
note that I assume your partitions are created and numbered as you have already stated.
What if the next step this is my mail server and I really don't have the time to reload it. I have my fstab, partition, mdstat, infomation. I ran this command sfdisk -d > sdb-parts.dump before a added the new drive. Will any of this help?
fdisk -l will list the partition information for each drive including start and end cylinders such as this.
[root@raptor pgsql]# fdisk -l /dev/hda Disk /dev/hda: 250.0 GB, 250059350016 bytes 255 heads, 63 sectors/track, 30401 cylinders Units = cylinders of 16065 * 512 = 8225280 bytes Device Boot Start End Blocks Id System /dev/hda1 * 1 21 168651 83 Linux /dev/hda2 22 532 4104607+ 83 Linux /dev/hda3 533 721 1518142+ 82 Linux swap /dev/hda4 722 30401 238404600 5 Extended /dev/hda5 722 1359 5124703+ 83 Linux ....From that you can get not only the size of each partition, but the
actual cylinders used and can recreate the table on the new drive appropriately with fdisk.
Like I said before the only raid/partition experience I have is at initial installation.
----- Original Message ----- From: "Jeff Vian" jvian10@charter.net To: "For users of Fedora Core releases" fedora-list@redhat.com Sent: Friday, July 28, 2006 5:44 PM Subject: Re: software raid drive failed, please provide step bu step torebuild
On Fri, 2006-07-28 at 16:29 -0500, Dan Carl wrote:
I have/had a software raid running and sdc drive failed. I got a replacement drive today and installed it. My only experience with set partitions and raids in during initail
setup.
I could not fdisk the new drive because i guess it wasn't reconized so I rebooted. Now I can reach the drive via fdisk but I have made more problems now
(no
swap now) and I'm not sure the steps to rebuild. Background: I have a FC3 with a software raid. I have 3 SCSI 18gb hard drives If I recall this how I set it up md0 /boot 100MB raid 1 sda, sdb and sdc as spare md1 /swp 768MB raid 0 sda, sdb, sdc
This toasted your /swp partition. Raid 0 is striping, and a single failure toasts the entire device.
You would have been ahead with a non-raid swap, and had 3 separate partitions, one on each device, for swap. Failure of one would not have toasted all.
md2 / ext3 33GB raid 5 sda, sdb, sdc
Can someone please help?
-- fedora-list mailing list fedora-list@redhat.com To unsubscribe: https://www.redhat.com/mailman/listinfo/fedora-list
-- fedora-list mailing list fedora-list@redhat.com To unsubscribe: https://www.redhat.com/mailman/listinfo/fedora-list
On Fri, 2006-07-28 at 17:44 -0500, Jeff Vian wrote:
This toasted your /swp partition. Raid 0 is striping, and a single failure toasts the entire device.
You would have been ahead with a non-raid swap, and had 3 separate partitions, one on each device, for swap. Failure of one would not have toasted all.
Yes, but losing swap will kill any processes that happened to be using it. The results of continuing to run with unpredictable things happening might be worse than crashing immediately.
-- Les Mikesell lesmikesell@gmail.com
On Mon, 2006-07-31 at 10:54 -0500, Les Mikesell wrote:
On Fri, 2006-07-28 at 17:44 -0500, Jeff Vian wrote:
This toasted your /swp partition. Raid 0 is striping, and a single failure toasts the entire device.
You would have been ahead with a non-raid swap, and had 3 separate partitions, one on each device, for swap. Failure of one would not have toasted all.
Yes, but losing swap will kill any processes that happened to be using it. The results of continuing to run with unpredictable things happening might be worse than crashing immediately.
I never said loss of a swap partition would not have consequences. I said that by using multiple partitions for swap that losing one would not totally lose the swap. A reboot with one failed will continue to use the remaining parts and at least part of the swap will be available.
-- Les Mikesell lesmikesell@gmail.com
-
Dan Carl -
Jeff Vian -
Les Mikesell -
Sam Varshavchik