Hi,
Background; we have a 22 node cluster, KVMs stored on a SAN via iSCSI / LVM.
One of the nodes went down (KSMTUNED was spewing errors onto the console, unable to do anything without segfaults - rebooted and it never came back). I was hoping that as I had a backup of the config file it would be a simple case of copying it onto another node and starting the KVM. However, when started up on a new node the KVM was unable to boot from the disk image.
Fdisk/gdisk/parted etc ran from other nodes were all unable to list the partitions on the disk image; but I can list partitions of other images on the SAN so we assumed it has somehow also become corrupted.
I reinstalled proxmox on the screwed HN this morning, and thought I'd try booting up the previously mentioned KVM....and it works. I'm also able to fdisk -l the image and see all the partitions etc.
The only difference I can think of between this VM and all the others, is the image is 3.2TB and has a GPT partition table.
Has anyone experienced anything like this? Or able to shed some light on how a disk image could be tied to a single machine like this, even through a restart?!
Thanks in advance,
Lee
Background; we have a 22 node cluster, KVMs stored on a SAN via iSCSI / LVM.
One of the nodes went down (KSMTUNED was spewing errors onto the console, unable to do anything without segfaults - rebooted and it never came back). I was hoping that as I had a backup of the config file it would be a simple case of copying it onto another node and starting the KVM. However, when started up on a new node the KVM was unable to boot from the disk image.
Fdisk/gdisk/parted etc ran from other nodes were all unable to list the partitions on the disk image; but I can list partitions of other images on the SAN so we assumed it has somehow also become corrupted.
I reinstalled proxmox on the screwed HN this morning, and thought I'd try booting up the previously mentioned KVM....and it works. I'm also able to fdisk -l the image and see all the partitions etc.
The only difference I can think of between this VM and all the others, is the image is 3.2TB and has a GPT partition table.
Has anyone experienced anything like this? Or able to shed some light on how a disk image could be tied to a single machine like this, even through a restart?!
Thanks in advance,
Lee