Koozali.org: home of the SME Server

SME Server 9.2 Raid status

Offline AdamHolland

  • 2
  • +0/-0
SME Server 9.2 Raid status
« on: June 25, 2020, 03:55:05 PM »
I have a strange one regarding the raid on a SME server 9.2 install and its RAID that looks like it has got confused. I will try and describe as best I can.
This has been up and running for a good couple of years now and every quarter I have been imaging the disk using clonezilla. We decided to test each of the individual drives by booting from each one by itself and both booted up fine. Since this was done the RAID does not sync with the other disk. the second disk boots fine but missing the latest data.
I think it has got confused as If i boot from a single disk it thinks there is a second disk present as when go to raid status it prompts to rebuild array. See pic 1
When disk 2 is put back in it thinks there is 3 drives, see pic 2.
I am by no means a specialist when it comes to SME server so any suggestions would be most welcome as this is obviously a live system.

Thanks

Adam

Offline ReetP

  • *
  • 3,722
  • +5/-0
Re: SME Server 9.2 Raid status
« Reply #1 on: June 25, 2020, 06:17:22 PM »
First question is how good are your backups??????

Never play with raid drives without being sure. Right now I woudln't start any more swapping around until we have an idea of what is happening.

Note that things can go wrong if you have a 2 drive mirror and you boot with a missing drive. Grub expects things in a  certain way and removing a drive entirely can have unexpected consequences.

Any drives should be properly failed, or replaced with another drive entirely. You should not just pull one drive and hope it works as it can lead to unintended consequences, as you have discovered.

You should also wipe a drive completely before adding it back - that includes properly clearing the MBR etc other wise it can also have odd consequences.

If you are going to to this you should....

Fail the drive. Shut down. Remove drive. Replace or completely wipe and replace. Reboot. Resync.

Do not be tempted to swap drives about etc. Leave the working one in situ and just replace the 'failed' one. (If you do this on says Raid 5/6/10 etc you will completely trash your array....)

OK. The trouble is now knowing which is the most "up to date" data source?

Curious why it see 3 drives.

What does dmesg tell you?

Code: [Select]
dmesg |grep sd
And fdisk?

Code: [Select]
fdisk -l
Or per drive:

Code: [Select]
fdisk -l /dev/sda
This should show you the same as the raid panel:

Code: [Select]
cat /proc/mdstat
And then this:

Code: [Select]
cat /etc/mdadm.conf
Have you got a USB drive or a.n. other drive plugged in my mistake? My son actually had this only a few days ago on a new little server - someone had left a micro USB drive key in a USB socket on the motherboard...

...
1. Read the Manual
2. Read the Wiki
3. Don't ask for support on Unsupported versions of software
4. I have a job, wife, and kids and do this in my spare time. If you want something fixed, please help.

Bugs are easier than you think: http://wiki.contribs.org/Bugzilla_Help

If you love SME and don't want to lose it, join in: http://wiki.contribs.org/Koozali_Foundation

Offline AdamHolland

  • 2
  • +0/-0
Re: SME Server 9.2 Raid status
« Reply #2 on: June 25, 2020, 07:10:05 PM »
Thank you for your reply. Guess this is a scenario where I'm repeating tests with what I'm used to with hardware raid.
Backups are good. They runs nightly to a NAS box. I would however do a clone again if was to perform any actions that could mess the system up to be on safe side.
I know which drive has the correct data so that's not a problem.
I will run those commands next time I am at that site which will hopefully be next week. Its a pub chain and with them being shut I have to arrange access.
There's is no USB sticks in the server and there never has been at any time to my knowledge.
I think that once this mystery disc is resolved that getting a resync would be a success.
Like I say I will run those commands and post when I have the results.
Thanks
Adam