[SOLVED] After host update: interrupted script address not found. scsi bus reset detected.

elterminatore

Active Member
Jun 18, 2018
47
3
28
49
Hello,
I am not sure if I am correct in my assumptions or if something has broken.

these are the latest updates i have installed

Code:
Start-Date: 2023-02-28  05:19:42
Commandline: apt-get -y dist-upgrade
Install: libslirp0:amd64 (4.4.0-1+deb11u2, automatic)
Upgrade: libcurl4:amd64 (7.74.0-1.3+deb11u5, 7.74.0-1.3+deb11u7), libcurl3-gnutls:amd64 (7.74.0-1.3+deb11u5, 7.74.0-1.3+deb11u7), pve-qemu-kvm:amd64 (7.1.0-4, 7.2.0-5), curl:amd64 (7.74.0-1.3+deb11u5, 7.74.0-1.3+deb11u7)
End-Date: 2023-02-28  05:19:48

Start-Date: 2023-03-07  07:14:52
Commandline: apt-get -y dist-upgrade
Upgrade: libproxmox-acme-perl:amd64 (1.4.3, 1.4.4), swtpm-libs:amd64 (0.8.0~bpo11+2, 0.8.0~bpo11+3), swtpm-tools:amd64 (0.8.0~bpo11+2, 0.8.0~bpo11+3), swtpm:amd64 (0.8.0~bpo11+2, 0.8.0~bpo11+3), lxc-pve:amd64 (5.0.2-1, 5.0.2-2), novnc-pve:amd64 (1.3.0-3, 1.4.0-1), qemu-server:amd64 (7.3-3, 7.3-4), libproxmox-acme-plugins:amd64 (1.4.3, 1.4.4), pve-i18n:amd64 (2.8-2, 2.8-3), pve-kernel-helper:amd64 (7.3-4, 7.3-5)
End-Date: 2023-03-07  07:15:52

this morning i booted all pve hosts in the cluster (HA and live migrations). after that one vm stopped responding. the screen showed the following:

1678267599460.png

a reboot or restore from backup showed the same result.

now i have to say that the vm is a very old one ;-)
a debian 4 with LSI53C895A controller and intel e1000 network card. virtio support was not available in the kernel at that time?

i don't need a solution for this!
i took the opportunity to replace the vm with one with a new operating system (and some pita things with the applications).
but i would like to know if this can be related to the last update(s).

regards
stefan
 
Last edited:
FYI we were able to reproduce (same issue hit us running proxmox on Debian 11 on an Azure VM using LSI 53C895A SCSI controller), and work around using the below downgrade:

Bash:
apt-get install pve-qemu-kvm=7.1.0-4

Note that it looks like the issue was introduced with pve-qemu-kvm (7.2.0-5).

We aren't able to use the VirtIO controller as it is fairly unstable on the hyper-v nested virtualisation that Azure uses (gets through bios but crashes after any sizable IO volume)

Some notes here - https://github.com/MicrosoftDocs/Virtualization-Documentation/pull/1833/files , for our use case (Lab environments in public cloud) running on nested virtualisation has been very effective except for the storage gotchas.
 
Last edited:
  • Like
Reactions: Muzarmo
Yes, my very old Debian 4 organizing webapplication has stoppt working too. Not sure, if it is a good idea to downgrade pve-qemu...
 
Yes, my very old Debian 4 organizing webapplication has stoppt working too. Not sure, if it is a good idea to downgrade pve-qemu...
Why not? If the previous version of pve-qemu-kvm worked fine, there should be no issue downgrading. The only thing you can't do is live-migrate a VM running with QEMU 7.2 with machine type latest or 7.2 back to a node with QEMU 7.1.

EDIT: If you don't want to downgrade, you can also just use a different SCSI controller.

That said, we'll likely need to revert the problematic patch for now, as it requires to change the LSI controller emulation code and that needs to be evaluated more carefully: https://lists.nongnu.org/archive/html/qemu-devel/2023-03/msg03739.html
 
Why not? If the previous version of pve-qemu-kvm worked fine, there should be no issue downgrading. The only thing you can't do is live-migrate a VM running with QEMU 7.2 with machine type latest or 7.2 back to a node with QEMU 7.1.

EDIT: If you don't want to downgrade, you can also just use a different SCSI controller.

That said, we'll likely need to revert the problematic patch for now, as it requires to change the LSI controller emulation code and that needs to be evaluated more carefully: https://lists.nongnu.org/archive/html/qemu-devel/2023-03/msg03739.html
OK, thanks. I've done the downgrade and it is running again.

Then I observe when a new Qemu version comes out and then test again.
 
Yes, my very old Debian 4 organizing webapplication has stoppt working too. Not sure, if it is a good idea to downgrade pve-qemu...
Hi, debian4 with kernel 3.2.23 is working fine with virtio-blk. (you should have better performance than not accelerated lsi controller).
the only difference is that disk are /dev/vdx instead /dev/sdx.
 
Hi, debian4 with kernel 3.2.23 is working fine with virtio-blk. (you should have better performance than not accelerated lsi controller).
the only difference is that disk are /dev/vdx instead /dev/sdx.
ah, ok... my debian 4 had a 2.6 kernel. (i dont know why... maybe because of an update from debian 3 or older. don't know... it's too lang ago)
virtio doens't work with this kernel.
but as i wrote: i used the better (more time intensive) solution and replaced the system with a current operating system and new versions of the application
 
Hi, debian4 with kernel 3.2.23 is working fine with virtio-blk. (you should have better performance than not accelerated lsi controller).
the only difference is that disk are /dev/vdx instead /dev/sdx.
Have tried it. But... With the downgrad it works and it is no problem to wait for a fix...

1678698405947.png
 
ah, ok... my debian 4 had a 2.6 kernel. (i dont know why... maybe because of an update from debian 3 or older. don't know... it's too lang ago)
virtio doens't work with this kernel.
but as i wrote: i used the better (more time intensive) solution and replaced the system with a current operating system and new versions of the application
I'm runing this backport kernel in my (last) debian4 vm:

64bit:
http://archive.debian.org/debian-ba...st/linux-image-amd64_3.2+46~bpo60+1_amd64.deb

32bit:
http://archive.debian.org/debian-ba...est/linux-image-amd64_3.2+46~bpo60+1_i386.deb
 
To be honest I don't want to touch the VM anymore. This is an ancient web application that is not scheduled to be used for very long. I'd rather wait for a qemu update in the next few days or weeks. But thanks for this possible solution!
 
We had this issue on a new cluster importing an old VM which worked previously. Had that SCSI error message and installing the older version over the top got them booting correctly.

Now my worry is that updates have probably been installed at another 5 sites, but the nodes haven't' been restarted. My concern now is that they probably will not boot unless I do the same package update.

I tried to boot using VirtIO and didn't have any success, the LSI was the only compatible combination we found to run legacy warehousing on.
These machines use a kernel from hundreds of years ago vmlinuz-2.4.21-144-smp4G.

I don't know enough about that kernel nor the scsi controller to know how to fix it so was lucky this post exists.
When I migrated dozens of VMs, the usual thing was to get them to boot and then figure out if the disks / partition names had shifted around and reattach everything - for this one only one combination worked.
 

About

The Proxmox community has been around for many years and offers help and support for Proxmox VE, Proxmox Backup Server, and Proxmox Mail Gateway.
We think our community is one of the best thanks to people like you!

Get your subscription!

The Proxmox team works very hard to make sure you are running the best software and getting stable updates and security enhancements, as well as quick enterprise support. Tens of thousands of happy customers have a Proxmox subscription. Get yours easily in our online shop.

Buy now!