Greetings I have been running a raid 10 array for almost 3 years. Last September on a reboot (forced by Firefox and kernel memory interaction problems) came up with one of the drives being listed as DOA. The raid array has continued to run on 3 drives although on reboots there is much complaining from whatever in the boot up process. I had the replacement drive sent to me and today I installed the drive. I was under the assumption that the raid array would rebuild itself upon startup with a new drive (4th out of 4). This did not happen. I am running Debian Jessie (testing) and have the whole time in question. I can mount the array and it is visible. I'm looking at backing up the array (on blu-ray discs) but as I'm now to circa 45 GB of data and I was at about 22 GB when I last did a backup this is going to be a momentous event. How do I ask mdadm to include this new drive into the array? (Without borking everything!! I have found lots of instructions on how to create but none in an hour of trying different search phrases on how to rebuild or how to cause the array to rebuild itself. I have only been able to find instructions for when the array is totally sick and I'm not there (yet - - - grin!) but I don't want to wait until another drive craps out to get things going.) TIA Dee
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 First, you should have backups. Why not get an external USB drive? You can get those quite inexpensive. Heck, even a 64GB thumb drive shouldn't be too expensive, though flash is not the most reliable. No matter what, I would start with a good backup. Playing with mdadm is relatively safe, but if it's your first time playing with it, you can get yourself into trouble. Hardware RAID controllers often auto-use replacement drives to rebuild a degraded array, but mdadm does not. How exactly you go about repairing the array depends on your specific configuration and which drive failed. You will need to start by partitioning the replacement drive such that it matches the existing drives (each array has to have a partition equal to or larger than the same partitions on the other drives). Once partitioned (and rebooted, if needed), you will add the new partition to the degraded array. Once added, mdadm will start re-assembling the array. Note that if one of the arrays has /boot, you'll probably need to run grub's install on the raw disk so that it can be booted from. The command to add the new partition to an existing array will look something like this: mdadm --manage /dev/md0 --add /dev/sda1 Obviously, the 'mdX' and 'sdYZ' will need to be adjusted for your configuration. digimer On 17/02/15 04:42 PM, o1bigtenor wrote:
Greetings
I have been running a raid 10 array for almost 3 years.
Last September on a reboot (forced by Firefox and kernel memory interaction problems) came up with one of the drives being listed as DOA. The raid array has continued to run on 3 drives although on reboots there is much complaining from whatever in the boot up process. I had the replacement drive sent to me and today I installed the drive.
I was under the assumption that the raid array would rebuild itself upon startup with a new drive (4th out of 4).
This did not happen.
I am running Debian Jessie (testing) and have the whole time in question. I can mount the array and it is visible. I'm looking at backing up the array (on blu-ray discs) but as I'm now to circa 45 GB of data and I was at about 22 GB when I last did a backup this is going to be a momentous event.
How do I ask mdadm to include this new drive into the array? (Without borking everything!! I have found lots of instructions on how to create but none in an hour of trying different search phrases on how to rebuild or how to cause the array to rebuild itself. I have only been able to find instructions for when the array is totally sick and I'm not there (yet - - - grin!) but I don't want to wait until another drive craps out to get things going.)
TIA
Dee
--- Talk Mailing List talk@gtalug.org http://gtalug.org/mailman/listinfo/talk
- -- Digimer Papers and Projects: https://alteeve.ca/w/ What if the cure for cancer is trapped in the mind of a person without access to education? -----BEGIN PGP SIGNATURE----- Version: GnuPG v1 iQIcBAEBAgAGBQJU47mYAAoJECChztQA3mh0hM8P/0pQLC3io+BWsDV1NzUObQGp F7TjarGw6zHo4p0dn2oTPt1JoiOp6GgmZKw3qVvnPkdnfR8loI4Ikuv6S2S2UwPi 5eGgdCZIrKw8iQfuyAd1NkukuLiQeUQ45DpaCpisCoDNAnMlCECg9CZELNB2sBnF WxvSp3i4I+tMutpRri/X+mUVwkzOPWAPE8qoPUCeYFs+X4fcF0L4+LfF+0oJIIIC 4j+aRykitL9AYLZwy/tgi+W7U8wZuwxr12f6AgRLn3VyArWnE6f5Jd6TM2/qmJy0 SSBLQ/1jjWXFlHkTMs/130kRzBEmqJjkh7//lvgXdnExNXU9lQZyaW7oly08zB7V 6i8whwkrcE6cEfu/RJZi+IUj/Y4Ja2KCVb2ltTDEbHzWQ2qvq6NEPzdtceASsYvm BGkSMQqcVT6HgCLRpnbZI+Wl3DXiC9TJg9KwL7wUlxjhB0MCV3X9dFV2RR+eWQEm zFYMvuFGSVuHZ1DEywvkHSLfgfcUtNGBAYcyauIh9pDbffNFBNxkBTa87qLcJGnP ZUVBi9EDIysbAHesmV0Ku9PxA05P3RwYgNmYDU+d2mpqXCxUt1CbQ9xUR4kkDa00 71AFSl2HhROgcwYhpKHj+aScHxTiMabs2hofcz11GAMVxoS+eJazVD9Si/OE/nC5 4FAKSOEgPUa5vc5YAjLy =zt65 -----END PGP SIGNATURE-----
On Tue, Feb 17, 2015 at 3:58 PM, Digimer <lists@alteeve.ca> wrote:
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
First, you should have backups. Why not get an external USB drive? You can get those quite inexpensive. Heck, even a 64GB thumb drive shouldn't be too expensive, though flash is not the most reliable.
I have about 45 GB on the array at present. I had decided to use Blu-ray discs for backup when I bought the system. There is no easy to back up method. (I'm no rsync expert!! just a noob.)
No matter what, I would start with a good backup. Playing with mdadm is relatively safe, but if it's your first time playing with it, you can get yourself into trouble.
Hardware RAID controllers often auto-use replacement drives to rebuild a degraded array, but mdadm does not. How exactly you go about repairing the array depends on your specific configuration and which drive failed. You will need to start by partitioning the replacement drive such that it matches the existing drives (each array has to have a partition equal to or larger than the same partitions on the other drives).
Once partitioned (and rebooted, if needed), you will add the new partition to the degraded array. Once added, mdadm will start re-assembling the array. Note that if one of the arrays has /boot, you'll probably need to run grub's install on the raw disk so that it can be booted from.
I have a separate drive for systems. The array is for storage. So partitioning the drive is necessary. OK - - - I hadn't thought of that!
The command to add the new partition to an existing array will look something like this:
mdadm --manage /dev/md0 --add /dev/sda1
Obviously, the 'mdX' and 'sdYZ' will need to be adjusted for your configuration.
You have helped me start the process! D
On Tue, Feb 17, 2015 at 03:42:07PM -0600, o1bigtenor wrote:
Greetings
I have been running a raid 10 array for almost 3 years.
Last September on a reboot (forced by Firefox and kernel memory interaction problems) came up with one of the drives being listed as DOA. The raid array has continued to run on 3 drives although on reboots there is much complaining from whatever in the boot up process. I had the replacement drive sent to me and today I installed the drive.
I was under the assumption that the raid array would rebuild itself upon startup with a new drive (4th out of 4).
This did not happen.
I am running Debian Jessie (testing) and have the whole time in question. I can mount the array and it is visible. I'm looking at backing up the array (on blu-ray discs) but as I'm now to circa 45 GB of data and I was at about 22 GB when I last did a backup this is going to be a momentous event.
So what does 'cat /proc/mdstat' show?
How do I ask mdadm to include this new drive into the array? (Without borking everything!! I have found lots of instructions on how to create but none in an hour of trying different search phrases on how to rebuild or how to cause the array to rebuild itself. I have only been able to find instructions for when the array is totally sick and I'm not there (yet - - - grin!) but I don't want to wait until another drive craps out to get things going.)
For software raid in linux you have to tell it what to do. So if you want it to add that disk to the raid, you have to use mdadm to tell it do do that. If it was partitioned before, you have to do that before adding the partition to the raid. -- Len Sorensen
On Tue, Feb 17, 2015 at 4:19 PM, Lennart Sorensen < lsorense@csclub.uwaterloo.ca> wrote:
Greetings
I have been running a raid 10 array for almost 3 years.
Last September on a reboot (forced by Firefox and kernel memory interaction problems) came up with one of the drives being listed as DOA. The raid array has continued to run on 3 drives although on reboots there is much complaining from whatever in the boot up process. I had the replacement drive sent to me and today I installed the drive.
I was under the assumption that the raid array would rebuild itself upon startup with a new drive (4th out of 4).
This did not happen.
I am running Debian Jessie (testing) and have the whole time in question. I can mount the array and it is visible. I'm looking at backing up the array (on blu-ray discs) but as I'm now to circa 45 GB of data and I was at about 22 GB when I last did a backup
On Tue, Feb 17, 2015 at 03:42:07PM -0600, o1bigtenor wrote: this
is going to be a momentous event.
So what does 'cat /proc/mdstat' show?
- grin!) but I don't want to wait until another drive craps out to get things going.)
How do I ask mdadm to include this new drive into the array? (Without borking everything!! I have found lots of instructions on how to create but none in an hour of trying different search phrases on how to rebuild or how to cause the array to rebuild itself. I have only been able to find instructions for when the array is totally sick and I'm not there (yet -
For software raid in linux you have to tell it what to do.
So if you want it to add that disk to the raid, you have to use mdadm to tell it do do that.
If it was partitioned before, you have to do that before adding the partition to the raid.
Now I used gdisk to set things up so likely that is where I will need to start - - - hmmmmmmmmmmm. Has anyone done anything like this in the somewhat recent past? (Wondering if when disk is partitioned that mdadm will happily absorb that disk into the array. Likely will need another reboot too!) Dee
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On 17/02/15 06:04 PM, o1bigtenor wrote:
Now I used gdisk to set things up so likely that is where I will need to start - - - hmmmmmmmmmmm.
Has anyone done anything like this in the somewhat recent past?
(Wondering if when disk is partitioned that mdadm will happily absorb that disk into the array. Likely will need another reboot too!)
Dee
I use parted or fdisk, I am not familiar with other tools. In the end, what matters is that the partition(s) on the new drive match (equal size or larger) the existing partitions. You will then have to tell mdadm to use the new partitions. It will not "just use them". As Lennart asked; Show us the output of 'cat /proc/mdstat'. Also, please share the output of: parted /dev/sda print free parted /dev/sdb print free parted /dev/sdc print free parted /dev/sdd print free digimer - -- Digimer Papers and Projects: https://alteeve.ca/w/ What if the cure for cancer is trapped in the mind of a person without access to education? -----BEGIN PGP SIGNATURE----- Version: GnuPG v1 iQIcBAEBAgAGBQJU48nAAAoJECChztQA3mh0FjsP/j0/HQhoe0yI/WBL3xBCVaiE rejYxagfzsmpxkYgOdRHOnuE0wNTUSYT9oTIj1cupj8SiwhtJF6EtJInVWYPvkit LuGQVj1KgtXORDu3rsvES9ZWluX8OQSFcXhxkzC4f51qkWMxMUANkUSuO/idOTCl 3oR/iaBxnG8MggRl0568SCyn9jLXHkcaDDRGd33gggO7wQz6SoPcR30W5LM8Ptnj NkRSoVJZ1gP69YnfWtcCl4uiVuYlrjbgPvidqjNrWb5Gu+uHBxwgqYppJECqyNeq EEZ9KZ9BiGNo6DV/0XgHyJTLssLAenkzs+JyP7Kqg5E97vIhi1jgvcVfTVvzD4d0 ITDvxk1Lx4o6znUsJYuACpP+cbJRQcpOT417jXiprpoJcz/C7orXzi+cBO3KOPWj wvoEkdXwIXa2UiCpmEkZaKuU+sR6RPKLW/OJaQFLorIGJzeKQVp37X2RWvOZXZKA sernbZAI7XhzvErh04Fq99WdsnABllgCWiwTJ4GUvlZdx3EFPjtiU6ZIT1PO2qyX h2QCItZ4ev7QW7sNwb2eW2pcX9yZJzAkA7IMdRbf3TLm7fMnDD9THknC0lu+Qtnh MdZrgM9Mcd8GZcLSggIwLOXS36ISQeA2mWJ3tjhL4W/qADCtX3DQq16rmTU+2HiH sVV4MBxN4BzuAfEp3cuV =68rh -----END PGP SIGNATURE-----
Hi On 17/02/15 06:04 PM, o1bigtenor wrote:
Now I used gdisk to set things up so likely that is where I will need to start - - - hmmmmmmmmmmm.
Has anyone done anything like this in the somewhat recent past?
Yes, I faced your problem late last year. You need to prepare your new drive to match existing drive. If the other 3 are not partitioned, leave the new one unpartitioned. Then you have to force it into the array using mdadm tool. I know, I spent says avoiding force until Lennart advised to go that way. Search the tlug old mail for that conversation. If you can't find it, I can look around for precise steps I used to get back up. On transit at the moment so just relying on memory. William
(Wondering if when disk is partitioned that mdadm will happily absorb that disk into the array. Likely will need another reboot too!)
Dee
I use parted or fdisk, I am not familiar with other tools. In the end, what matters is that the partition(s) on the new drive match (equal size or larger) the existing partitions. You will then have to tell mdadm to use the new partitions. It will not "just use them". As Lennart asked; Show us the output of 'cat /proc/mdstat'. Also, please share the output of: parted /dev/sda print free parted /dev/sdb print free parted /dev/sdc print free parted /dev/sdd print free digimer - -- Digimer Papers and Projects: https://alteeve.ca/w/ What if the cure for cancer is trapped in the mind of a person without access to education? -----BEGIN PGP SIGNATURE----- Version: GnuPG v1 iQIcBAEBAgAGBQJU48nAAAoJECChztQA3mh0FjsP/j0/HQhoe0yI/WBL3xBCVaiE rejYxagfzsmpxkYgOdRHOnuE0wNTUSYT9oTIj1cupj8SiwhtJF6EtJInVWYPvkit LuGQVj1KgtXORDu3rsvES9ZWluX8OQSFcXhxkzC4f51qkWMxMUANkUSuO/idOTCl 3oR/iaBxnG8MggRl0568SCyn9jLXHkcaDDRGd33gggO7wQz6SoPcR30W5LM8Ptnj NkRSoVJZ1gP69YnfWtcCl4uiVuYlrjbgPvidqjNrWb5Gu+uHBxwgqYppJECqyNeq EEZ9KZ9BiGNo6DV/0XgHyJTLssLAenkzs+JyP7Kqg5E97vIhi1jgvcVfTVvzD4d0 ITDvxk1Lx4o6znUsJYuACpP+cbJRQcpOT417jXiprpoJcz/C7orXzi+cBO3KOPWj wvoEkdXwIXa2UiCpmEkZaKuU+sR6RPKLW/OJaQFLorIGJzeKQVp37X2RWvOZXZKA sernbZAI7XhzvErh04Fq99WdsnABllgCWiwTJ4GUvlZdx3EFPjtiU6ZIT1PO2qyX h2QCItZ4ev7QW7sNwb2eW2pcX9yZJzAkA7IMdRbf3TLm7fMnDD9THknC0lu+Qtnh MdZrgM9Mcd8GZcLSggIwLOXS36ISQeA2mWJ3tjhL4W/qADCtX3DQq16rmTU+2HiH sVV4MBxN4BzuAfEp3cuV =68rh -----END PGP SIGNATURE----- --- Talk Mailing List talk@gtalug.org http://gtalug.org/mailman/listinfo/talk
Sorry for a slow response but life intervened! On Tue, Feb 17, 2015 at 5:07 PM, Digimer <lists@alteeve.ca> wrote:
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
On 17/02/15 06:04 PM, o1bigtenor wrote:
Now I used gdisk to set things up so likely that is where I will need to start - - - hmmmmmmmmmmm.
Has anyone done anything like this in the somewhat recent past?
(Wondering if when disk is partitioned that mdadm will happily absorb that disk into the array. Likely will need another reboot too!)
Dee
I use parted or fdisk, I am not familiar with other tools. In the end, what matters is that the partition(s) on the new drive match (equal size or larger) the existing partitions.
The drive has not yet been partitioned. The disks (in the array) were partitioned to be just one large partition (whole disk).
You will then have to tell mdadm to use the new partitions. It will not "just use them".
As Lennart asked; Show us the output of 'cat /proc/mdstat'.
# cat /proc/mdstat Personalities : [raid10] md0 : active raid10 sdc1[0] sda1[3] sde1[1] 1953518592 blocks super 1.2 512K chunks 2 near-copies [4/3] [UU_U] unused devices: <none>
Also, please share the output of:
parted /dev/sda print free
# parted /dev/sda print free Model: ATA ST31000524AS (scsi) Disk /dev/sda: 1000GB Sector size (logical/physical): 512B/512B Partition Table: msdos Number Start End Size Type File system Flags 32.3kB 1049kB 1016kB Free Space 1 1049kB 1000GB 1000GB primary raid 1000GB 1000GB 745kB Free Space
parted /dev/sdb print free
sdb is my system drive and is not part of the array
parted /dev/sdc print free
# parted /dev/sdc print free Model: ATA ST31000524AS (scsi) Disk /dev/sdc: 1000GB Sector size (logical/physical): 512B/512B Partition Table: msdos Number Start End Size Type File system Flags 32.3kB 1049kB 1016kB Free Space 1 1049kB 1000GB 1000GB primary raid 1000GB 1000GB 745kB Free Space
parted /dev/sdd print free
sdd is my RAM drive sde # parted /dev/sde print free Model: ATA ST31000524AS (scsi) Disk /dev/sde: 1000GB Sector size (logical/physical): 512B/512B Partition Table: msdos Number Start End Size Type File system Flags 32.3kB 1049kB 1016kB Free Space 1 1049kB 1000GB 1000GB primary raid 1000GB 1000GB 745kB Free Space sdf # parted /dev/sdf print free Error: /dev/sdf: unrecognised disk label When the system was setup I had the array as disks sda, sdb, sdc and sdd with sde being my system disk and sdf being the RAM drive. In the process of installing the new hard drive I did not label the wires from the existing setup when I was removing the wires. (I know better just didn't think first!) The new number schema doesn't seem to be causing any difficulties. Should I be trying to change things back to the sda - d raid and e/f being for the system drive and RAM drive? Thanking you for your sharing and insights!! Dee
On Thu, Feb 19, 2015 at 06:17:57AM -0600, o1bigtenor wrote:
Sorry for a slow response but life intervened!
On Tue, Feb 17, 2015 at 5:07 PM, Digimer <lists@alteeve.ca> wrote:
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
On 17/02/15 06:04 PM, o1bigtenor wrote:
Now I used gdisk to set things up so likely that is where I will need to start - - - hmmmmmmmmmmm.
Has anyone done anything like this in the somewhat recent past?
(Wondering if when disk is partitioned that mdadm will happily absorb that disk into the array. Likely will need another reboot too!)
Dee
I use parted or fdisk, I am not familiar with other tools. In the end, what matters is that the partition(s) on the new drive match (equal size or larger) the existing partitions.
The drive has not yet been partitioned. The disks (in the array) were partitioned to be just one large partition (whole disk).
You will then have to tell mdadm to use the new partitions. It will not "just use them".
As Lennart asked; Show us the output of 'cat /proc/mdstat'.
# cat /proc/mdstat Personalities : [raid10] md0 : active raid10 sdc1[0] sda1[3] sde1[1] 1953518592 blocks super 1.2 512K chunks 2 near-copies [4/3] [UU_U]
unused devices: <none>
Well once you have partitioned the new disk you should be able to just do: mdadm --add /dev/md0 /dev/sdX1 (whatever X is for the new drive). Then /proc/mdstat should show it rebuilding the raid. -- Len Sorensen
snip
Well once you have partitioned the new disk you should be able to just do:
mdadm --add /dev/md0 /dev/sdX1 (whatever X is for the new drive).
Then /proc/mdstat should show it rebuilding the raid.
Result: 1. partitioned the drive using gparted (in the running system) to be sdf1 I did a boot into a rescue disk to check but it had worked on the running system. This was not listed as possible but the command worked!) 2. read and read (a few times) the man page for mdadm I was looking specifically at the commands -D -v --add as suggested by William found the total lack of examples frustrating (a normal man page I think!) 3. took the command above, suggested by Len, and changed it somewhat to # mdadm -v --add /dev/md0 /dev/sdf1 4. checked the rebuilding by using # vi /proc/mdstat Thank you to those who contributed!! Dee
On Tue, Feb 17, 2015 at 05:04:16PM -0600, o1bigtenor wrote:
Now I used gdisk to set things up so likely that is where I will need to start - - - hmmmmmmmmmmm.
Has anyone done anything like this in the somewhat recent past?
(Wondering if when disk is partitioned that mdadm will happily absorb that disk into the array. Likely will need another reboot too!)
There should be zero reboots required for this. Most partition tools do a partition table reread automatically when you exit, or you can use 'hdparm -z /dev/sdX' to do it. If /proc/partitions shows the new sizes, you are good to go. The partitions have to be at least the size of the old partitions or the raid won't be able to use them. -- Len Sorensen
On Thu, Feb 19, 2015 at 9:30 AM, Lennart Sorensen < lsorense@csclub.uwaterloo.ca> wrote:
Now I used gdisk to set things up so likely that is where I will need to start - - - hmmmmmmmmmmm.
Has anyone done anything like this in the somewhat recent past?
(Wondering if when disk is partitioned that mdadm will happily absorb
On Tue, Feb 17, 2015 at 05:04:16PM -0600, o1bigtenor wrote: that
disk into the array. Likely will need another reboot too!)
There should be zero reboots required for this.
Most partition tools do a partition table reread automatically when you exit, or you can use 'hdparm -z /dev/sdX' to do it. If /proc/partitions shows the new sizes, you are good to go.
Is there a way to partition the drive in a running system? The only way that I know of is to reboot onto a rescue disk and use gparted there to partition the drive. I have not been able to find anything using duckduck nor ms google. A complete back is just finishing as I write this. Thanking those that have contributed to the conversation for their assistance!! Dee
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On 19/02/15 05:14 PM, o1bigtenor wrote:
On Thu, Feb 19, 2015 at 9:30 AM, Lennart Sorensen <lsorense@csclub.uwaterloo.ca <mailto:lsorense@csclub.uwaterloo.ca>> wrote:
On Tue, Feb 17, 2015 at 05:04:16PM -0600, o1bigtenor wrote:
Now I used gdisk to set things up so likely that is where I will need to start - - - hmmmmmmmmmmm.
Has anyone done anything like this in the somewhat recent past?
(Wondering if when disk is partitioned that mdadm will happily absorb that disk into the array. Likely will need another reboot too!)
There should be zero reboots required for this.
Most partition tools do a partition table reread automatically when you exit, or you can use 'hdparm -z /dev/sdX' to do it. If /proc/partitions shows the new sizes, you are good to go.
Is there a way to partition the drive in a running system?
The only way that I know of is to reboot onto a rescue disk and use gparted there to partition the drive.
I have not been able to find anything using duckduck nor ms google.
A complete back is just finishing as I write this.
Thanking those that have contributed to the conversation for their assistance!!
Dee
You can use parted on running systems just fine. On older systems, the drive with the OS will not update the partition table until you reboot, but repartitioning other drives should be updated fine. Parted will warn you if a reboot is needed. On newer systems, even the drive under the running OS can be updated without a reboot. - -- Digimer Papers and Projects: https://alteeve.ca/w/ What if the cure for cancer is trapped in the mind of a person without access to education? -----BEGIN PGP SIGNATURE----- Version: GnuPG v1 iQIcBAEBAgAGBQJU5mwUAAoJECChztQA3mh0T34P/RQ8PWnOXxeqKMhBkwLCN+A+ 1BkwLiJs6rRF/dbpCOtJh48SL+RzFQ4IhFmM7JsILSe6ueKxyHUXHohgAF6nl5TI K4XBorvz5QKO2d3Qc88o6X0Dreaxt9JLMCIe/fZH/jsYzMzZZp0jBYzAhevdSiWG W33BD25262iy/8TKnlhYa8Xx9xNkicdrHHqitbXPpj1Y4kXgcYIyCMXSioxaeBmu 4+DZp0ZWRn94yYay1dUlHObd7DjHCWr7CxQTlHxRB/UDHPf3Fat1SSiN6bFN7ows e0bOrzGFEFLSsJM6sVMt22zHWAo+zHgD41OlmoMXHc84QKq48kx9uVQHlflF846a KLTZLKRKODsAjBDpk+4+7NJf2jvvuG4XjxJE6LMS4cxkVOJ69nvz7yANrMQ1PvuA 3iIJjOcR4ZSrTS7upY+/Mowb23yK5hD6DBvLhSoTGkGJGWN1zWgR/RVh83+1ZGL9 jDJaY46WrWdc1mVGCI9e8AZU+dROzF19QuYcMCfye3Flo3UMAA8I0F/hQJYy2okt MgnYZjwKhFDniAzF0iGktUweuYYR031/7lYOv6M8aeL+FXeeb+Ed/XESi2TBiwlh m/rCk2ErRwLaJAV8hAFAVEBn1pQssN/Epn5HlkvxWooZGMxPIId2t/1BU81bmNTQ 13qxo0y8zubvwcKBB5sp =ypgx -----END PGP SIGNATURE-----
snip
Is there a way to partition the drive in a running system?
The only way that I know of is to reboot onto a rescue disk and use gparted there to partition the drive.
snip
You can use parted on running systems just fine. On older systems, the drive with the OS will not update the partition table until you reboot, but repartitioning other drives should be updated fine. Parted will warn you if a reboot is needed.
On newer systems, even the drive under the running OS can be updated without a reboot.
I installed gparted using CLI. I can find a preferences file for gparted (that won't open) but I cannot run the program. I tried using a number of techniques to start gparted but so far nothing is working. I am not worried about doing a reboot to a rescue disk and then doing the work and then rebooting back into Debian testing but am also trying to use this as a learning opportunity (hopefully without that being learning how to redo the system as a whole which I have had to do at least twice since the original install). Thanks for your advice! Dee
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On 19/02/15 06:26 PM, o1bigtenor wrote:
snip
Is there a way to partition the drive in a running system?
The only way that I know of is to reboot onto a rescue disk and use gparted there to partition the drive.
snip
You can use parted on running systems just fine. On older systems, the drive with the OS will not update the partition table until you reboot, but repartitioning other drives should be updated fine. Parted will warn you if a reboot is needed.
On newer systems, even the drive under the running OS can be updated without a reboot.
I installed gparted using CLI. I can find a preferences file for gparted (that won't open) but I cannot run the program. I tried using a number of techniques to start gparted but so far nothing is working. I am not worried about doing a reboot to a rescue disk and then doing the work and then rebooting back into Debian testing but am also trying to use this as a learning opportunity (hopefully without that being learning how to redo the system as a whole which I have had to do at least twice since the original install).
Thanks for your advice!
Dee
I don't use Debian or gparted, so I can't provide advice. - -- Digimer Papers and Projects: https://alteeve.ca/w/ What if the cure for cancer is trapped in the mind of a person without access to education? -----BEGIN PGP SIGNATURE----- Version: GnuPG v1 iQIcBAEBAgAGBQJU5nt+AAoJECChztQA3mh0vQkP/3YHnRAYuAGP36YsBHYAJ+FQ soIFFTyq2OVKGCgH04IoQm30GsrSju5zu3xOCHxbZNiKACgwcabub0mxuZV0DpQV V7XlfToVW0c5q3KK04AKyrjQBWHYX4n/BJFjR+HU0h6FAR+pWsq3282YhgmM1mz9 KLciBx/1FI6P1wjNMPpB5SGE7XDLARtp1G2m1v48q6sWDyRmbUlRXvYX/Vpzacmg NZOPTaGr6ls+HiPs5S/6rPbITpMLZW3EzIURqpr+5OWyJQWCjb12PEAbxdL8WTOX 9cvoWlWitQQppL1FfCuQNAfxMVr1kC9x/Yb25vScUxiKlk2s6Q7PO0sImqe2ok/r Rkw31IiwVL3ULG8ZrNhjSXlM9Nw1ScY9ObfmY1pecpD5n1hC3Aze/Lzf9Y9+y6pJ s82oFzLrti5OIi8uHyQkt2SjRDBj4IoIvHSKgD87ZEZiyoB7rNX3U4LyFHqS18kz KwlTJHPACtE3VmVn07DAp9hCikY1otyXCttQpe8JL63dXY36FplcXpB1K+xVyAAJ ZcOspB3qkvyFD7rkv+P7uZcsvgLpNMR0cHOgmSpZ8hlnoUafXzJWZhZRcvmQ7nOy f/fAGPQwrHjkR6QYATGpN+HPRqkysOCVH5JAFwEv2lD2qrMLbc1/tjylHxQu6Yja zBmgziWlw4Hwj86Xhs9P =StVm -----END PGP SIGNATURE-----
On Thu, Feb 19, 2015 at 06:04:52PM -0500, Digimer wrote:
You can use parted on running systems just fine. On older systems, the drive with the OS will not update the partition table until you reboot, but repartitioning other drives should be updated fine. Parted will warn you if a reboot is needed.
On newer systems, even the drive under the running OS can be updated without a reboot.
In my experience, linux will reread the partition table as long as NOTHING is using that drive (so nothing is mounted). If it is in use, you need to reboot to get the partition change to take. -- Len Sorensen
On Fri, Feb 20, 2015 at 2:42 PM, William Park <opengeometry@yahoo.ca> wrote:
On Fri, Feb 20, 2015 at 02:23:58PM -0500, Lennart Sorensen wrote:
If it is in use, you need to reboot to get the partition change to take.
Or, partprobe /dev/sdx at least on my Slackware.
Have been looking at what I can find on this. Will respond with what I did after the fact! Thanking those that have responded to my query! Dee
On Thu, Feb 19, 2015 at 04:14:04PM -0600, o1bigtenor wrote:
On Thu, Feb 19, 2015 at 9:30 AM, Lennart Sorensen < lsorense@csclub.uwaterloo.ca> wrote:
Now I used gdisk to set things up so likely that is where I will need to start - - - hmmmmmmmmmmm.
Has anyone done anything like this in the somewhat recent past?
(Wondering if when disk is partitioned that mdadm will happily absorb
On Tue, Feb 17, 2015 at 05:04:16PM -0600, o1bigtenor wrote: that
disk into the array. Likely will need another reboot too!)
There should be zero reboots required for this.
Most partition tools do a partition table reread automatically when you exit, or you can use 'hdparm -z /dev/sdX' to do it. If /proc/partitions shows the new sizes, you are good to go.
Is there a way to partition the drive in a running system?
The only way that I know of is to reboot onto a rescue disk and use gparted there to partition the drive.
I have not been able to find anything using duckduck nor ms google.
A complete back is just finishing as I write this.
Thanking those that have contributed to the conversation for their assistance!!
Well cfdisk, parted, fdisk, gdisk, etc. Somewhat depends if you need GPT or MBR style partition table. -- Len Sorensen
On Fri, Feb 20, 2015 at 1:22 PM, Lennart Sorensen < lsorense@csclub.uwaterloo.ca> wrote:
On Thu, Feb 19, 2015 at 04:14:04PM -0600, o1bigtenor wrote:
On Thu, Feb 19, 2015 at 9:30 AM, Lennart Sorensen < lsorense@csclub.uwaterloo.ca> wrote:
Now I used gdisk to set things up so likely that is where I will need to start - - - hmmmmmmmmmmm.
Has anyone done anything like this in the somewhat recent past?
(Wondering if when disk is partitioned that mdadm will happily absorb
On Tue, Feb 17, 2015 at 05:04:16PM -0600, o1bigtenor wrote: that
disk into the array. Likely will need another reboot too!)
There should be zero reboots required for this.
Most partition tools do a partition table reread automatically when you exit, or you can use 'hdparm -z /dev/sdX' to do it. If /proc/partitions shows the new sizes, you are good to go.
Is there a way to partition the drive in a running system?
The only way that I know of is to reboot onto a rescue disk and use gparted there to partition the drive.
I have not been able to find anything using duckduck nor ms google.
A complete back is just finishing as I write this.
Thanking those that have contributed to the conversation for their assistance!!
Well cfdisk, parted, fdisk, gdisk, etc.
Somewhat depends if you need GPT or MBR style partition table.
Needs to be a GPT style partition table. Its on a UEFI system and that was the only way I could get things working. I got quite used to fdisk and gdisk is just different enough that I am still a little careful. I don't really enjoy those long teaching moments anymore! TIA Dee
Disk failed last September, and you're replacing it now? Boy, you got balls! You are better off to use whatever utility you used in the first place. Manual typing would go something like man mdadm cat /proc/mdstat mdadm -D -v /dev/md0 mdadm /dev/md0 --add /dev/sdx where /dev/sdx is the new disk. Here, the new device should be the "same" as the old. If the old was a partition (eg. /dev/sda1), then the new should be a partition. If the old was entire device (eg. /dev/sda), then the new should be entire device also. -- William On Tue, Feb 17, 2015 at 03:42:07PM -0600, o1bigtenor wrote:
Greetings
I have been running a raid 10 array for almost 3 years.
Last September on a reboot (forced by Firefox and kernel memory interaction problems) came up with one of the drives being listed as DOA. The raid array has continued to run on 3 drives although on reboots there is much complaining from whatever in the boot up process. I had the replacement drive sent to me and today I installed the drive.
I was under the assumption that the raid array would rebuild itself upon startup with a new drive (4th out of 4).
This did not happen.
I am running Debian Jessie (testing) and have the whole time in question. I can mount the array and it is visible. I'm looking at backing up the array (on blu-ray discs) but as I'm now to circa 45 GB of data and I was at about 22 GB when I last did a backup this is going to be a momentous event.
How do I ask mdadm to include this new drive into the array? (Without borking everything!! I have found lots of instructions on how to create but none in an hour of trying different search phrases on how to rebuild or how to cause the array to rebuild itself. I have only been able to find instructions for when the array is totally sick and I'm not there (yet - - - grin!) but I don't want to wait until another drive craps out to get things going.)
TIA
Dee
--- Talk Mailing List talk@gtalug.org http://gtalug.org/mailman/listinfo/talk
On Tue, Feb 17, 2015 at 9:19 PM, William Park <opengeometry@yahoo.ca> wrote:
Disk failed last September, and you're replacing it now? Boy, you got balls!
You are better off to use whatever utility you used in the first place. Manual typing would go something like man mdadm cat /proc/mdstat
understand above commands
mdadm -D -v /dev/md0
what does this command do/mean?
mdadm /dev/md0 --add /dev/sdx
(I understand I will need a different letter instead of the x) what does this command do/mean?
where /dev/sdx is the new disk. Here, the new device should be the "same" as the old. If the old was a partition (eg. /dev/sda1), then the new should be a partition. If the old was entire device (eg. /dev/sda), then the new should be entire device also. --
I am not asking the questions to be confrontational just have learned that typing in commands the results of which you do not know or the effect you do not know/understand can have large effects. (In the past I have too often totally borked my system and I really don't want to take two weeks from my life and spend all day trying to reestablish things.) Thanking you for your interest and assistance! Dee
William
Greetings
I have been running a raid 10 array for almost 3 years.
Last September on a reboot (forced by Firefox and kernel memory interaction problems) came up with one of the drives being listed as DOA. The raid array has continued to run on 3 drives although on reboots there is much complaining from whatever in the boot up process. I had the replacement drive sent to me and today I installed the drive.
I was under the assumption that the raid array would rebuild itself upon startup with a new drive (4th out of 4).
This did not happen.
I am running Debian Jessie (testing) and have the whole time in question. I can mount the array and it is visible. I'm looking at backing up the array (on blu-ray discs) but as I'm now to circa 45 GB of data and I was at about 22 GB when I last did a backup
is going to be a momentous event.
How do I ask mdadm to include this new drive into the array? (Without borking everything!! I have found lots of instructions on how to create but none in an hour of trying different search phrases on how to rebuild or how to cause the array to rebuild itself. I have only been able to find instructions for when the array is totally sick and I'm not there (yet -
On Tue, Feb 17, 2015 at 03:42:07PM -0600, o1bigtenor wrote: this -
- grin!) but I don't want to wait until another drive craps out to get things going.)
TIA
Dee
--- Talk Mailing List talk@gtalug.org http://gtalug.org/mailman/listinfo/talk
--- Talk Mailing List talk@gtalug.org http://gtalug.org/mailman/listinfo/talk
You should read manpage of 'mdadm', paying particular attention to its arguments. So, man mdadm and search for options -D -v --add -- William On Thu, Feb 19, 2015 at 06:38:29AM -0600, o1bigtenor wrote:
I am not asking the questions to be confrontational just have learned that typing in commands the results of which you do not know or the effect you do not know/understand can have large effects.
(In the past I have too often totally borked my system and I really don't want to take two weeks from my life and spend all day trying to reestablish things.)
Thanking you for your interest and assistance!
Dee
On Thu, Feb 19, 2015 at 10:54 PM, William Park <opengeometry@yahoo.ca> wrote:
You should read manpage of 'mdadm', paying particular attention to its arguments. So, man mdadm and search for options -D -v --add
This was one of the first things I did BEFORE I installed the new drive. Thank you for the exact command. Would you be able to verify that I can only use gparted on a non-running system? <rant on> Man pages are largely incomprehensible to the average user. I think that there are almost totally designed for someone who is hugely conversant with what is being done and just needs a reminder of which command does what. Now for someone who doesn't program a man page provides mostly no examples (ones that will work or ones that won't). There is no listing of a group of commands to do things (as you have done above) telling you that this string will achieve this result. Even better would be if they would then tell you that if you don't get this result (with an example) but you get x - - then this went wrong which you fix like this or if you get y - - - that went wrong which you fix like that. When I have done something 20 times the man page becomes an OK tool (often far too terse but still useable) but when I have done something once - - well a man page most often makes the problem worse. (Then I get told to "just google it" and the first 20 hits are at least 6 if not 12 to 15 years old - - - so what the net pukes out just isn't useful. Makes it quite difficult to fix problems that one hasn't run into before. The fact that fixing some of my problems has had my wife telling me that I'm wasting too much time and that I could hire someone to do this for me if I just were running MS Win - - - I have been able to resist this but logically she has a point - - - my computers are tools not time sinks!) <rant off> Thanking you for your assistance AND consideration!! Dee
participants (5)
-
Digimer -
Lennart Sorensen -
o1bigtenor -
William Muriithi -
William Park