dist-upgrade broke GRUB... I think?

Hey Wizard,

Thanks for the link, and apologies for the delayed reply. Life gets busy sometimes, eh? Anyway, I attempted to install grub onto the RAID, as that last link seems to suggest, but no dice, unfortunately. I've attached my commands run and their output, for reference, and the output of the boot-repair tool's "Boot info" option.

Hopefully something there gives you a clue. I'm close to whacking a standalone drive in to install a new OS onto, and boot off, and then just mounting the RAID off my existing drives in that new OS.

Thanks for all your help so far, mate.

Cheers,
Benjamin.
 

Attachments

  • Boot-Info_20190108_1020.txt
    80.9 KB · Views: 884
  • chroot boot fix.txt
    9.2 KB · Views: 977


Life gets busy sometimes, eh?

I'll say :) haven't forgotten you mate, found a couple of links and I'll be back with more tomorrow.

Chris
 
Benjamin

Search parameters I used were

grub-install: warning: Couldn't find physical volume `(null)'. Some modules may be missing from core image..

To my untrained eye, this one looked promising

https://www.claudiokuenzler.com/blog/688/grub-grub2-install-warning-couldnt-find-physical-volume

You bootinfo xxx.txt is good but beyond my ability to track through. I'll ping a couple of people again.

@JasKinasis , @Lazydog , @mal-2 - folks, OP has RAID on GPT disks with an MBR / - can you assist?

I'm close to whacking a standalone drive in to install a new OS onto, and boot off, and then just mounting the RAID off my existing drives in that new OS.

I can only guess :(

If you go offsite and get an answer, can you bring it back so we can add it to our Wiki?

Sorry can't help more.

Chris
 
What does your grub.cfg file look like? What type of RAID are you running? I would not mix GTP and MBR in fact I don't think you can.
 
GPT.

Thanks for coming on board, Robert, I have been struggling with this :)

Wiz
 
I try to help where I can when I can. This is how we learn. ;)
 
Hi all :)

I've just done an apt-get dist-upgrade on my Debian 9.6 server, and during the process, it attempted to update GRUB. This failed as it could not find the disk when it attempted to install. I told it not to install GRUB, thinking a simple grub-install /dev/sda1 would work after the dist-upgrade, but this returns a similar error.

As I understand it, I need to install GRUB over the top of itself to make sure it loads the new kernel on boot. I'm fuzzy on this part, as my previous GRUB already pointed to the right drive/partition/etc., so I thought it wouldn't need changing, but as the installer wanted to update it, I'm not confident I'm understanding this part correctly at all.

My setup is 4 3TB drives each with a BIOS partition, and the rest of the drive as a second partition. These second partitions are joined in a RAID5, which is presented to LVM as the PV.

My highlighting.

Benjamin, can you clarify for Robert and others (hopefully) - had you had the system running successfully for a period of time prior to the attempt to update/upgrade?

Cheers

Wiz
 
Hey guys, thanks for coming in.

The system was live and working, yes. It's been a media server and other things for about a year.

I realised that I had a degraded RAID array - this hasn't happened to me before, and this is my first RAID, so I didn't see the "activated with 3 of 4 devices" as a sign that there was a problem. My RAID is rebuilding now, and I'll come back with the state I'm in once that's done :)
 
For anyone that finds this thread in future, all I did was this:

1. Work out which partitions are supposed to be part of your RAID. I did this with # cat /proc/partitions, and looking at which partitions were my 2.9TB ones. For me, this was sda2, sdb2, sdd2, and sde2. sdc was my boot USB.
2. Work out which partitions are currently in the RAID, and which aren't, by running # cat /proc/mdstat. On the line starting with "md0", I saw that sdd2 was missing.
3. Re-add the missing drive to the array: # mdadm /dev/md0 -a /dev/sdd2. The array will automatically start rebuilding itself.
4. Watch the re-build progress: # watch -n2 cat /proc/mdstat

The rebuild shows you its ETA under "finish=". For me, this started around 170,000 minutes, but it completed in about 20 minutes. I think it's a bit like the Windows file copy dialog:
https://xkcd.com/612/

Once the rebuild was complete, I did a clean shutdown, removed the USB, and booted back into my real OS.

Cheers,
Benjamin.
 
WOW :), I follow a bit of that. I will bookmark this Thread for when I need it.

Regrets it took so long, thanks for your patience, and for sharing.

Cheers

Chris
 

Staff online


Top