On 1/3/23 09:15, Stephen John Smoogen wrote:


On Tue, 3 Jan 2023 at 08:12, Robert Moskowitz <rgm@htt-consult.com> wrote:
This is software RAID1, not hardware RAID1?

Would hardware hit the same problem?  the Terramaster looks like it is hardware RAID?


Hardware RAID for ESP partitions requires that the EUFI firmware to understand how to talk to the hardware raid device. If the EUFI doesn't have the right driver it will try to talk to the hardware raid as a raw device in a different way and it goes to pot. [You would think that server hardware would be built to do this out of the box, but I have had a couple where hardware raid was only supported to boot non-EUFI or required the EUFI to be on a separate drive.]

Argh.

I want to buy a small RAID platform for my mail server which really needs updating.  After what just happened on my NAS, I am totally sold on RAID.  And for my small size, RAID1 is ok.

So I can't buy a decent small business turnkey mail server, need to build it.  CentOS is gone, that means AlmaLinux and probably iRedMail...

But if I get the wrong box that won't give me AL supporting RAID, I have shot a few hundred.

Sigh.


 
On 1/3/23 08:04, Stephen John Smoogen wrote:


On Tue, 3 Jan 2023 at 06:19, Alessandro Baggi <alessandro.baggi@gmail.com> wrote:
Hi Jack,
I'm sorry to bother you during holidays.

I encountered a strange problem installing AlmaLinux 9.1 on a RAID1
(MDADM) with the current configuration:

- /boot/efi on md125
- swap on md126
- / on md127


My limited understanding is that RAID on EFI has been something of a hack as the backing store that EFI uses is a slightly modified VFAT. What happens is that there is some code to 'clone' the data across but it isn't really RAID1. My guess is that something in the 9.1 kernel broke that hack. Could you try CentOS Stream 9 kernel (you can install that with your existing Alma or Rocky system) and see if the problem still occurs? If it does then it is a bug that needs to be tracked upstream at bugzilla.redhat.com and if it doesn't then it should have been fixed in an upcoming kernel. You could continue to then use the CS9 kernel until whatever works in Alma/Rocky 9

 
disks are 2 SSD MLC Type.

After the installation, if I reboot the system I get:

"md: md125 stopped" (it is printed many times like in a loop)

alternated with:

"systemd-shutdown[1]: Not all MD devices stopped, 1 left
Stopping MD Devices
Stopping MD /dev/md125 (9:125)"

and the system hangs on this loop until I cut the power.

I encountered this issue on my Workstation with Asus Prime Z490-A /
i9-10850k.

I tried with another workstation that runs on Asus Prime Z370-A / i7
8700K, to exclude bad SATA controller and bad cables.

Also on the second workstation the problem is found.

I tried to replicate this using 9.0 ISO. The problem does not occur
until I update to 9.1

I tried also with 8.7. No problem here.

I tried also RockyLinux 9.1 and got the same problem but with different
messages:

"block device autoconfig is deprecated and will removed"

alternated with:

"blkdev_get_no_open: 270 callbacks suppressed."

To stop the machine I need to cut the power.

I tried also Debian 11.5 without problems.

So seems that the problem is 9.1 related. Actually I can't test the same
with RHEL 9.1 but probably the problem will occour also on RHEL 9.1

There is a way to fix this or I should wait an upgrade?

Thank you in advance.
_______________________________________________
AlmaLinux Users mailing list -- users@lists.almalinux.org
To unsubscribe send an email to users-leave@lists.almalinux.org


--
Stephen J Smoogen.
Let us be kind to one another, for most of us are fighting a hard battle. -- Ian MacClaren

_______________________________________________
AlmaLinux Users mailing list -- users@lists.almalinux.org
To unsubscribe send an email to users-leave@lists.almalinux.org



--
Stephen J Smoogen.
Let us be kind to one another, for most of us are fighting a hard battle. -- Ian MacClaren