I have searched the web for a solution and have found nothing on this topic. The error I get is sda1 and sdb1 have the same UUID. I would rather not remove a drive from raid and unplug it to do the install.
I must be missing something.
Any help appreciated,
David
Am 01.05.2023 um 18:12 schrieb David Woodyard dwoody5654@gmail.com:
I have searched the web for a solution and have found nothing on this topic. The error I get is sda1 and sdb1 have the same UUID. I would rather not remove a drive from raid and unplug it to do the install.
I must be missing something.
It’s a well proven procedure. What do you want to install? Workstation, Server, one of the Spins?
-- Peter Boy https://fedoraproject.org/wiki/User:Pboy pboy@fedoraproject.org
Timezone: CET (UTC+1) / CEST (UTC+2)
Fedora Server Edition Working Group member Fedora docs team contributor Java developer and enthusiast
the server is what I need to install.
David
On Mon, May 1, 2023 at 11:28 AM Peter Boy pboy@uni-bremen.de wrote:
Am 01.05.2023 um 18:12 schrieb David Woodyard dwoody5654@gmail.com:
I have searched the web for a solution and have found nothing on this
topic.
The error I get is sda1 and sdb1 have the same UUID. I would rather not
remove a drive from raid and unplug it to do the install.
I must be missing something.
It’s a well proven procedure. What do you want to install? Workstation, Server, one of the Spins?
-- Peter Boy https://fedoraproject.org/wiki/User:Pboy pboy@fedoraproject.org
Timezone: CET (UTC+1) / CEST (UTC+2)
Fedora Server Edition Working Group member Fedora docs team contributor Java developer and enthusiast
users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-leave@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
Am 01.05.2023 um 18:33 schrieb David Woodyard dwoody5654@gmail.com:
the server is what I need to install.
OK, o I guess Server will be the only Operating System on the device. And because you ‚see‘ sda and sdb, you don’t have a hardware raid but use a software raid.
Second bunch of questions:
(a) Do you need to keep data? If yes, how they are stored? On its own partition?
If not, is there anything against completely restructuring the hard disks? Then you wouldn't have to worry about the UUID at all.
(b) how are the disk currently formatted, what is the output of
lsblk
(c)
Do you know, if it is a BIOSboot machine or an UEFI machine?
-- Peter Boy https://fedoraproject.org/wiki/User:Pboy pboy@fedoraproject.org
Timezone: CET (UTC+1) / CEST (UTC+2)
Fedora Server Edition Working Group member Fedora docs team contributor Java developer and enthusiast
On Mon, May 1, 2023 at 12:06 PM Peter Boy pboy@uni-bremen.de wrote:
Am 01.05.2023 um 18:33 schrieb David Woodyard dwoody5654@gmail.com:
the server is what I need to install.
OK, o I guess Server will be the only Operating System on the device. And because you ‚see‘ sda and sdb, you don’t have a hardware raid but use a software raid.
Fedora is the only OS.
Second bunch of questions:
(a) Do you need to keep data? If yes, how they are stored? On its own partition?
Both / and /home has raid1. / of course will be over written. /home needs to be preserved.
If not, is there anything against completely restructuring the hard disks? Then you wouldn't have to worry about the UUID at all.
(b) how are the disk currently formatted, what is the output of
Formatted as ext4 for everything.
lsblk
NAME MAJ:MIN RM SIZE RO TYPE MOUNTPOINTS
sda 8:0 1 931.5G 0 disk ├─sda1 8:1 1 14.7G 0 part └─sda2 8:2 1 916.9G 0 part └─md127 9:127 0 916.7G 0 raid1 /home sdb 8:16 1 931.5G 0 disk └─sdb2 8:18 1 916.9G 0 part └─md127 9:127 0 916.7G 0 raid1 /home sdc 8:32 0 111.8G 0 disk └─sdc1 8:33 0 111.8G 0 part └─md0 9:0 0 111.7G 0 raid1 sdd 8:48 0 111.8G 0 disk └─sdd1 8:49 0 111.8G 0 part └─md0 9:0 0 111.7G 0 raid1 sde 8:64 1 28.6G 0 disk ├─sde1 8:65 1 1M 0 part └─sde2 8:66 1 28.6G 0 part / zram0 252:0 0 8G 0 disk [SWAP]
The sde drive is a usb that I am using to boot with until I get the hdd fixed.
(c)
Do you know, if it is a BIOSboot machine or an UEFI machine?
The MB is from 2009. So it is bios, does not have BIOSboot partition.
Thanks for your help,
David
-- Peter Boy https://fedoraproject.org/wiki/User:Pboy pboy@fedoraproject.org
Timezone: CET (UTC+1) / CEST (UTC+2)
Fedora Server Edition Working Group member Fedora docs team contributor Java developer and enthusiast
users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-leave@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
Am 01.05.2023 um 19:39 schrieb David Woodyard dwoody5654@gmail.com:
... Fedora is the only OS.
That’s good. So it is quite easy.
Second bunch of questions:
(a) Do you need to keep data? If yes, how they are stored? On its own partition? Both / and /home has raid1. / of course will be over written. /home needs to be preserved.
That’s not so good, but perfectly doable. Keeping sda2/sdb2 and mounting as /home may cause some headaches in the long run, in organizational and administrative terms. But does not affect the reliability.
Maybe, you study https://docs.fedoraproject.org/en-US/fedora-server/installation/#_storage_or... and https://docs.fedoraproject.org/en-US/fedora-server/installation/interactive-...
If you haven’t already done so.
If not, is there anything against completely restructuring the hard disks? Then you wouldn't have to worry about the UUID at all.
(b) how are the disk currently formatted, what is the output of
Formatted as ext4 for everything.
lsblk
NAME MAJ:MIN RM SIZE RO TYPE MOUNTPOINTS sda 8:0 1 931.5G 0 disk ├─sda1 8:1 1 14.7G 0 part └─sda2 8:2 1 916.9G 0 part └─md127 9:127 0 916.7G 0 raid1 /home sdb 8:16 1 931.5G 0 disk └─sdb2 8:18 1 916.9G 0 part └─md127 9:127 0 916.7G 0 raid1 /home sdc 8:32 0 111.8G 0 disk └─sdc1 8:33 0 111.8G 0 part └─md0 9:0 0 111.7G 0 raid1 sdd 8:48 0 111.8G 0 disk └─sdd1 8:49 0 111.8G 0 part └─md0 9:0 0 111.7G 0 raid1 sde 8:64 1 28.6G 0 disk ├─sde1 8:65 1 1M 0 part └─sde2 8:66 1 28.6G 0 part / zram0 252:0 0 8G 0 disk [SWAP]
The sde drive is a usb that I am using to boot with until I get the hdd fixed.
The Harddisks are <2 TB and it looks like a DOS/MBR partitioning. Fedora nowadays used GPT for all disks. As said, you can keep DOS/MBR, but may cause organizational and administrative problems later due to the different default partitioning.
Another issue:
(a) sdb has just sdb2, no sdb1. So we have to know, how it is partitioned in detail to determine, how we can use the disk as part of a raid.
(b) Regarding the mentioned UUID issue you have to check the disk ID.
Therefore, use as root / with rood permission (sudo -i)
fdisk -l | less
You can scroll up and down to check all partition entries.
Would be good if you could post the listing here
Another question:
how do you use sdc/sdd ? They are raid type 1 as well. but obviously not mounted anywhere.
To install Fedora Server you have several options:
(a) To create a Raid of 14 GB on sda and sdb, format at as xfs (no LVM as Fedora uses by default) and use it as the root file system for system data. Mount the raid on sda2 und sdb2 as /home in the root file system. That’s a bit tricky, but nevertheless just a series of several step-by-step actions.
(b) Install the system (root file system) just on sda1 without raid (and therefore without redundancy in case of a disk failure for the system files) and mount the current raid on /home. That’s the easiest way.
(c) If you don’t use sdc/sdd, then install Fedora on those disk, and - again - mount sda2/sdb2 as /home. That’s quite easy as well.
(d) Backup /home, completely reformat the disks and restore /home.
Comparing the alternatives:
(c) seems the best solution if you can completely overwrite disks sdc/sdd.
(b) is best, if you mainly want to play around and explore Fedora Server
(a) Is best if you want to seriously use Fedora Server, but without touching you precious data (provided you can add a partition to sdb as required).
(d) is best if you want a long term solution in line with Fedora's development and administrative principles and (upcomming) administrative tools.
That’s your choice.
-- Peter Boy https://fedoraproject.org/wiki/User:Pboy pboy@fedoraproject.org
Timezone: CET (UTC+1) / CEST (UTC+2)
Fedora Server Edition Working Group member Fedora docs team contributor Java developer and enthusiast
On 5/1/23 12:01, Peter Boy wrote:
Am 01.05.2023 um 19:39 schrieb David Woodyard dwoody5654@gmail.com: NAME MAJ:MIN RM SIZE RO TYPE MOUNTPOINTS sda 8:0 1 931.5G 0 disk ├─sda1 8:1 1 14.7G 0 part └─sda2 8:2 1 916.9G 0 part └─md127 9:127 0 916.7G 0 raid1 /home sdb 8:16 1 931.5G 0 disk └─sdb2 8:18 1 916.9G 0 part └─md127 9:127 0 916.7G 0 raid1 /home sdc 8:32 0 111.8G 0 disk └─sdc1 8:33 0 111.8G 0 part └─md0 9:0 0 111.7G 0 raid1 sdd 8:48 0 111.8G 0 disk └─sdd1 8:49 0 111.8G 0 part └─md0 9:0 0 111.7G 0 raid1 sde 8:64 1 28.6G 0 disk ├─sde1 8:65 1 1M 0 part └─sde2 8:66 1 28.6G 0 part / zram0 252:0 0 8G 0 disk [SWAP]
The sde drive is a usb that I am using to boot with until I get the hdd fixed.
Another question:
how do you use sdc/sdd ? They are raid type 1 as well. but obviously not mounted anywhere.
Since he's booting from sde, I assume that's the original / raid.
On 2023-05-01 1:05 p.m., Peter Boy wrote:
Am 01.05.2023 um 18:33 schrieb David Woodyard dwoody5654@gmail.com:
the server is what I need to install.
OK, o I guess Server will be the only Operating System on the device. And because you ‚see‘ sda and sdb, you don’t have a hardware raid but use a software raid.
Second bunch of questions:
(a) Do you need to keep data? If yes, how they are stored? On its own partition?
If not, is there anything against completely restructuring the hard disks? Then you wouldn't have to worry about the UUID at all.
(b) how are the disk currently formatted, what is the output of
lsblk
(c) Do you know, if it is a BIOSboot machine or an UEFI machine?
Because you can see 2 drives, I assume that any fake-RAID BIOS settings are disabled. You appear to be looking at a 2-drive JBOD setup with identical UUIDs for some unknown reason. Since this is going to be the only operating system on the machine and we have no idea what was done to those drives, just format them as part of the install and get them into a known state.
Instead of messing around with a fairly complicated and error-prone mechanism using EXT4 or XFS and MD-RAID layers, just install using BTRFS in its RAID-1 configuration. Its faster (as fast as XFS on reads), more reliable due to everything being checksummed and duplexed, simpler, handles failing drives in a more obvious way, and is the default filesystem in Fedora Workstation anyways.
On Mon, May 1, 2023 at 11:33 AM David Woodyard dwoody5654@gmail.com wrote:
the server is what I need to install.
David
On Mon, May 1, 2023 at 11:28 AM Peter Boy pboy@uni-bremen.de wrote:
Am 01.05.2023 um 18:12 schrieb David Woodyard dwoody5654@gmail.com:
I have searched the web for a solution and have found nothing on this
topic.
The error I get is sda1 and sdb1 have the same UUID. I would rather not
remove a drive from raid and unplug it to do the install.
I must be missing something.
It’s a well proven procedure. What do you want to install? Workstation, Server, one of the Spins?
Gmail has deleted your second reply, so I am replying to this one.
I, perhaps, was not as clear as I should have been on my first email. I have a computer that has a raid1 and it has been working for several years. I want to install Fedora-Server 37. When I run the install (anaconda) it gives an error about duplicate UUID's and will not do the install.
That tells me that I can not install on a computer that ALREADY has raid1. I was not expecting that error.
Is it correct that anaconda can NOT install to a hard drive that ALREADY has a RAID1? Is there a workaround other than removing one drive from the raid and unplugging it from the computer?
Thanks,
David
-- Peter Boy https://fedoraproject.org/wiki/User:Pboy pboy@fedoraproject.org
Timezone: CET (UTC+1) / CEST (UTC+2)
Fedora Server Edition Working Group member Fedora docs team contributor Java developer and enthusiast
users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-leave@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
Am 01.05.2023 um 22:22 schrieb David Woodyard dwoody5654@gmail.com:
On Mon, May 1, 2023 at 11:33 AM David Woodyard dwoody5654@gmail.com wrote: the server is what I need to install.
David
On Mon, May 1, 2023 at 11:28 AM Peter Boy pboy@uni-bremen.de wrote:
Am 01.05.2023 um 18:12 schrieb David Woodyard dwoody5654@gmail.com:
I have searched the web for a solution and have found nothing on this topic. The error I get is sda1 and sdb1 have the same UUID. I would rather not remove a drive from raid and unplug it to do the install.
I must be missing something.
It’s a well proven procedure. What do you want to install? Workstation, Server, one of the Spins?
Gmail has deleted your second reply, so I am replying to this one.
I suppose you mean my last reply. I add it at the bottom.
I, perhaps, was not as clear as I should have been on my first email. I have a computer that has a raid1 and it has been working for several years. I want to install Fedora-Server 37. When I run the install (anaconda) it gives an error about duplicate UUID's and will not do the install.
That tells me that I can not install on a computer that ALREADY has raid1. I was not expecting that error.
Is it correct that anaconda can NOT install to a hard drive that ALREADY has a RAID1? Is there a workaround other than removing one drive from the raid and unplugging it from the computer?
No, that’s not correct. I think, no system can use 2 disk which have the same UUID at the same time (besides maybe one of the Windows BIOS fake controller). Duplicate UUID is a contradictio in adiecto and should be fixed. Anaconda can install on existing raid anyway. And there are several ways to fix the issue with the UUID. But I have to know more details. First question is, which of your disks is affected and what is the output of 'fdisk -l'.
- repeating my previous post - - - - - - <
Am 01.05.2023 um 19:39 schrieb David Woodyard dwoody5654@gmail.com:
... Fedora is the only OS.
That’s good. So it is quite easy.
Second bunch of questions:
(a) Do you need to keep data? If yes, how they are stored? On its own partition? Both / and /home has raid1. / of course will be over written. /home needs to be preserved.
That’s not so good, but perfectly doable. Keeping sda2/sdb2 and mounting as /home may cause some headaches in the long run, in organizational and administrative terms. But does not affect the reliability.
Maybe, you study https://docs.fedoraproject.org/en-US/fedora-server/installation/#_storage_or... and https://docs.fedoraproject.org/en-US/fedora-server/installation/interactive-...
If you haven’t already done so.
If not, is there anything against completely restructuring the hard disks? Then you wouldn't have to worry about the UUID at all.
(b) how are the disk currently formatted, what is the output of
Formatted as ext4 for everything.
lsblk
NAME MAJ:MIN RM SIZE RO TYPE MOUNTPOINTS sda 8:0 1 931.5G 0 disk ├─sda1 8:1 1 14.7G 0 part └─sda2 8:2 1 916.9G 0 part └─md127 9:127 0 916.7G 0 raid1 /home sdb 8:16 1 931.5G 0 disk └─sdb2 8:18 1 916.9G 0 part └─md127 9:127 0 916.7G 0 raid1 /home sdc 8:32 0 111.8G 0 disk └─sdc1 8:33 0 111.8G 0 part └─md0 9:0 0 111.7G 0 raid1 sdd 8:48 0 111.8G 0 disk └─sdd1 8:49 0 111.8G 0 part └─md0 9:0 0 111.7G 0 raid1 sde 8:64 1 28.6G 0 disk ├─sde1 8:65 1 1M 0 part └─sde2 8:66 1 28.6G 0 part / zram0 252:0 0 8G 0 disk [SWAP]
The sde drive is a usb that I am using to boot with until I get the hdd fixed.
The Harddisks are <2 TB and it looks like a DOS/MBR partitioning. Fedora nowadays used GPT for all disks. As said, you can keep DOS/MBR, but may cause organizational and administrative problems later due to the different default partitioning.
Another issue:
(a) sdb has just sdb2, no sdb1. So we have to know, how it is partitioned in detail to determine, how we can use the disk as part of a raid.
(b) Regarding the mentioned UUID issue you have to check the disk ID.
Therefore, use as root / with rood permission (sudo -i)
fdisk -l | less
You can scroll up and down to check all partition entries.
Would be good if you could post the listing here
Another question:
how do you use sdc/sdd ? They are raid type 1 as well. but obviously not mounted anywhere.
To install Fedora Server you have several options:
(a) To create a Raid of 14 GB on sda and sdb, format at as xfs (no LVM as Fedora uses by default) and use it as the root file system for system data. Mount the raid on sda2 und sdb2 as /home in the root file system. That’s a bit tricky, but nevertheless just a series of several step-by-step actions.
(b) Install the system (root file system) just on sda1 without raid (and therefore without redundancy in case of a disk failure for the system files) and mount the current raid on /home. That’s the easiest way.
(c) If you don’t use sdc/sdd, then install Fedora on those disk, and - again - mount sda2/sdb2 as /home. That’s quite easy as well.
(d) Backup /home, completely reformat the disks and restore /home.
Comparing the alternatives:
(c) seems the best solution if you can completely overwrite disks sdc/sdd.
(b) is best, if you mainly want to play around and explore Fedora Server
(a) Is best if you want to seriously use Fedora Server, but without touching you precious data (provided you can add a partition to sdb as required).
(d) is best if you want a long term solution in line with Fedora's development and administrative principles and (upcomming) administrative tools.
That’s your choice.
-- Peter Boy https://fedoraproject.org/wiki/User:Pboy pboy@fedoraproject.org
Timezone: CET (UTC+1) / CEST (UTC+2)
Fedora Server Edition Working Group member Fedora docs team contributor Java developer and enthusiast
On Mon, 2023-05-01 at 23:12 +0200, Peter Boy wrote:
I think, no system can use 2 disk which have the same UUID at the same time (besides maybe one of the Windows BIOS fake controller). Duplicate UUID is a contradictio in adiecto and should be fixed.
Just a guess, but it probably could, if you didn't use the UUIDs to mount the drives by. Such as you mounted them by device names, or volume names. Though, I think it'd be best not to have cloned IDs.
Briefly looking at RAID information, there are things that should be unique, and there are some things that can be duplicated (not so sure that they should be, though). Drive IDs would need to be unique for anything that uses IDs to differentiate one drive from other. There's partition and volume IDs that are used for different purposes.
Am 02.05.2023 um 09:05 schrieb Tim via users users@lists.fedoraproject.org:
On Mon, 2023-05-01 at 23:12 +0200, Peter Boy wrote:
I think, no system can use 2 disk which have the same UUID at the same time (besides maybe one of the Windows BIOS fake controller). Duplicate UUID is a contradictio in adiecto and should be fixed.
Just a guess, but it probably could, if you didn't use the UUIDs to mount the drives by. Such as you mounted them by device names, or volume names. Though, I think it'd be best not to have cloned IDs.
Briefly looking at RAID information, there are things that should be unique, and there are some things that can be duplicated (not so sure that they should be, though). Drive IDs would need to be unique for anything that uses IDs to differentiate one drive from other. There's partition and volume IDs that are used for different purposes.
I agree. But it’s hard not to use UIEDs and to ignore misconfigured UUIDs. Many Fedora tools use UUID by default, e.g. Cockpit and - if I remember correctly - dbus. Therefore, cloning a disc often ends up in more work than cloning saves.
-- Peter Boy https://fedoraproject.org/wiki/User:Pboy pboy@fedoraproject.org
Timezone: CET (UTC+1) / CEST (UTC+2)
Fedora Server Edition Working Group member Fedora docs team contributor Java developer and enthusiast
Tim:
Briefly looking at RAID information, there are things that should be unique, and there are some things that can be duplicated (not so sure that they should be, though). Drive IDs would need to be unique for anything that uses IDs to differentiate one drive from other. There's partition and volume IDs that are used for different purposes.
Peter Boy:
I agree. But it’s hard not to use UIEDs and to ignore misconfigured UUIDs. Many Fedora tools use UUID by default, e.g. Cockpit and - if I remember correctly - dbus. Therefore, cloning a disc often ends up in more work than cloning saves.
Since my experience with RAID is minimal (using a motherboard with built-in hardware RAID that could not be shut-off on the drive ports I had to use), I assume that if a mirror drive dies, you swap it, and let the RAID do its own magic to incorporate the replacement drive into the system, it handles filling up the new drive with partitions and data from the other drive without cloning IDs.
Am 02.05.2023 um 16:45 schrieb Tim via users users@lists.fedoraproject.org:
Tim:
Briefly looking at RAID information, there are things that should be unique, and there are some things that can be duplicated (not so sure that they should be, though). Drive IDs would need to be unique for anything that uses IDs to differentiate one drive from other. There's partition and volume IDs that are used for different purposes.
Peter Boy:
I agree. But it’s hard not to use UIEDs and to ignore misconfigured UUIDs. Many Fedora tools use UUID by default, e.g. Cockpit and - if I remember correctly - dbus. Therefore, cloning a disc often ends up in more work than cloning saves.
Since my experience with RAID is minimal (using a motherboard with built-in hardware RAID that could not be shut-off on the drive ports I had to use), I assume that if a mirror drive dies, you swap it, and let the RAID do its own magic to incorporate the replacement drive into the system, it handles filling up the new drive with partitions and data from the other drive without cloning IDs.
Software Raid write various meta data on the drive. Therefor - as I understand the docs - you MUST NOT clone the drive. I use SW Raid on my private servers only, and luckily I had not so much drive failures so far. But SW Raid restored the drive on its own when I tested it a while ago. However, according to my memory, I had created a (empty) partition table. I don't remember if that was necessary or if it just happened.
-- Peter Boy https://fedoraproject.org/wiki/User:Pboy pboy@fedoraproject.org
Timezone: CET (UTC+1) / CEST (UTC+2)
Fedora Server Edition Working Group member Fedora docs team contributor Java developer and enthusiast
Hardware or software raid typically makes and exact copy of what they are mirroring.
Hardware raid typically mirrors the entire disk (all UUIDS on the disk will be duplicated on the 2nd device, but in real hw raid the raid controller hides the underlying devices), software raid can do the entire disk or simply do one or more partitions, but exactly clones what is in the mirrored partition, there will be design be duplicate UUIDs for any FSes and/or PV/VG/LVM's.
It sounds like MD-RAID is creating the correct raid1 MD device(with duplicate UUID for the fs as would be expected), but anaconda is not correctly ignoring the underlying components in use and being managed by MD-RAID.
On Tue, May 2, 2023 at 9:45 AM Tim via users users@lists.fedoraproject.org wrote:
Tim:
Briefly looking at RAID information, there are things that should be unique, and there are some things that can be duplicated (not so sure that they should be, though). Drive IDs would need to be unique for anything that uses IDs to differentiate one drive from other. There's partition and volume IDs that are used for different purposes.
Peter Boy:
I agree. But it’s hard not to use UIEDs and to ignore misconfigured UUIDs. Many Fedora tools use UUID by default, e.g. Cockpit and - if I remember correctly - dbus. Therefore, cloning a disc often ends up in more work than cloning saves.
Since my experience with RAID is minimal (using a motherboard with built-in hardware RAID that could not be shut-off on the drive ports I had to use), I assume that if a mirror drive dies, you swap it, and let the RAID do its own magic to incorporate the replacement drive into the system, it handles filling up the new drive with partitions and data from the other drive without cloning IDs.
--
uname -rsvp Linux 3.10.0-1160.88.1.el7.x86_64 #1 SMP Tue Mar 7 15:41:52 UTC 2023 x86_64
Boilerplate: All unexpected mail to my mailbox is automatically deleted. I will only get to see the messages that are posted to the mailing list.
users mailing list -- users@lists.fedoraproject.org To unsubscribe send an email to users-leave@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
On Tue, May 2, 2023, at 8:14 AM, Roger Heflin wrote:
Hardware or software raid typically makes and exact copy of what they are mirroring.
I was just preparing a reply to this. I had been confused since I do run Linux MD raid 1 and I was pretty sure I had matching UUID. Here are two examples that show older and newer raid version/type:
(note that they use matching UUID for the members, but use a unique UUID for the raid, and that there is also a unique PARTUUID for each)
8:39-doug@wombat-~>blkid | grep 7: /dev/sdb7: UUID="a1efda72-102e-139f-6435-186fb839f57c" TYPE="linux_raid_member" PARTUUID="28f4447c-07" /dev/md7: UUID="8688b29b-c244-4dc6-88bc-02f282396dc3" BLOCK_SIZE="4096" TYPE="ext4" /dev/sda7: UUID="a1efda72-102e-139f-6435-186fb839f57c" TYPE="linux_raid_member" PARTUUID="87e58b90-07"
8:39-doug@wombat-~>blkid | grep 1: /dev/sdb1: UUID="deb9b913-0a73-90d8-6b37-12e5156fce03" UUID_SUB="50b1ecf9-a7bd-6ba5-131e-4bbde91735c8" LABEL="wombat.wombatz.com:boot" TYPE="linux_raid_member" PARTUUID="28f4447c-01" /dev/sda1: UUID="deb9b913-0a73-90d8-6b37-12e5156fce03" UUID_SUB="8efb2a67-f08a-4ba9-186f-11f7afec25c7" LABEL="wombat.wombatz.com:boot" TYPE="linux_raid_member" PARTUUID="87e58b90-01" /dev/md1: UUID="72f36fca-b84e-45f4-8621-d85b14a32ff3" BLOCK_SIZE="4096" TYPE="ext4"
On Tue, May 2, 2023 at 10:46 AM Doug Herr fedoraproject.org@wombatz.com wrote:
On Tue, May 2, 2023, at 8:14 AM, Roger Heflin wrote:
Hardware or software raid typically makes and exact copy of what they are mirroring.
I was just preparing a reply to this. I had been confused since I do run Linux MD raid 1 and I was pretty sure I had matching UUID. Here are two examples that show older and newer raid version/type:
(note that they use matching UUID for the members, but use a unique UUID for the raid, and that there is also a unique PARTUUID for each)
8:39-doug@wombat-~>blkid | grep 7: /dev/sdb7: UUID="a1efda72-102e-139f-6435-186fb839f57c" TYPE="linux_raid_member" PARTUUID="28f4447c-07" /dev/md7: UUID="8688b29b-c244-4dc6-88bc-02f282396dc3" BLOCK_SIZE="4096" TYPE="ext4" /dev/sda7: UUID="a1efda72-102e-139f-6435-186fb839f57c" TYPE="linux_raid_member" PARTUUID="87e58b90-07"
8:39-doug@wombat-~>blkid | grep 1: /dev/sdb1: UUID="deb9b913-0a73-90d8-6b37-12e5156fce03" UUID_SUB="50b1ecf9-a7bd-6ba5-131e-4bbde91735c8" LABEL="wombat.wombatz.com:boot" TYPE="linux_raid_member" PARTUUID="28f4447c-01" /dev/sda1: UUID="deb9b913-0a73-90d8-6b37-12e5156fce03" UUID_SUB="8efb2a67-f08a-4ba9-186f-11f7afec25c7" LABEL="wombat.wombatz.com:boot" TYPE="linux_raid_member" PARTUUID="87e58b90-01" /dev/md1: UUID="72f36fca-b84e-45f4-8621-d85b14a32ff3" BLOCK_SIZE="4096" TYPE="ext4" _______________________________________________
If your only duplicates are the above member UUID's then anaconda must have a bug in dealing with pre-existing md-raid configs and not ignoring the raid_members that will be duplicate.
Given 38 is out you may simply want to try to see if 38 works or not.
On 05/01/2023 10:12 AM, David Woodyard wrote:
I have searched the web for a solution and have found nothing on this topic. The error I get is sda1 and sdb1 have the same UUID. I would rather not remove a drive from raid and unplug it to do the install.
Are you able to have both drives mounted at the same time, or is there an error when you try? If you're unable to get both of them mounted together, use gparted (I think it's part of the standard install.) to edit the UUID on sdb1 (changing one digit should be enough) as well as fstab, along with any other config files that need it.
On Mon, May 1, 2023 at 11:13 AM David Woodyard dwoody5654@gmail.com wrote:
I have searched the web for a solution and have found nothing on this topic. The error I get is sda1 and sdb1 have the same UUID. I would rather not remove a drive from raid and unplug it to do the install.
I must be missing something.
Any help appreciated,
David
Linux MDRAID raid 1 or Vendor/Bios Fake Raid 1? Most of the MB built-in's are fake raid 1.
Depending on the fake raid 1 vendor/version it may not be supported in the kernel by default and may need extra (vendor supplied) modules to even work (or have no option at all to work).
And if you have to use an extra vendor supplied module, upgrades may or may not always work right.
On Mon, 2023-05-01 at 11:12 -0500, David Woodyard wrote:
I have searched the web for a solution and have found nothing on this topic. The error I get is sda1 and sdb1 have the same UUID. I would rather not remove a drive from raid and unplug it to do the install.
I assume you cloned a drive. If the drives are simply prepared blanks, you could modify one of the drives (change its UUID) with a simple reformat. And, of course, change the /etc/fstab file to use the new uuid.
If the drives are blanks/empty, then I'd have thought the install routine would give you some method to prepare them to suit ignoring the current status of the drives. n.b. It's usually a waste of time to prepare blank drives before starting an install.
I'm not sure about modifying a drive with data on it. It should be doable without destroying content (research the tune2fs or btrfstune and uuidgen commands for ext or btrfs file systems on some OSs). Possibly gparted might help.
You could use something other than the UUID to mount them (device name, volume name).
I know I'm being a bit vague, but if either of those steps sound understandable, we can elaborate.