cluster storage gone

[elmacus]

Hi.
I test with new server as Cluster Master. No machines created.
Add old server as node.
Then all storage is gone from old, only local storage and no ISO mapping.
I try to make old server as Master, storage still gone, all old KVM running, but no links to disk.
After restart of old server as Master, no machines start.
How do i find and attach the storage ?
What did i do wrong ? Must the old server be master, or do you always loose storage links when building cluster ? Shall i fix this now as standalone servers, and wait for 2.0 to build cluster ?

pveca -l shows the same on both servers, ok.
Both is 1.8, Kernel 2.6.35 ,identical pveversion except this line:
proxmox-ve-2.6.32: 1.8-32 ,New, installed from new 1.8 iso
proxmox-ve-2.6.35: 1.8-10 ,Old, installed from 1.7 and updated to 1.8, tried aptitude but nothing new.

 

[udo]

Hi.
I test with new server as Cluster Master. No machines created.
Add old server as node.
Then all storage is gone from old, only local storage and no ISO mapping.
I try to make old server as Master, storage still gone, all old KVM running, but no links to disk.
After restart of old server as Master, no machines start.
How do i find and attach the storage ?
What did i do wrong ? Must the old server be master, or do you always loose storage links when building cluster ? Shall i fix this now as standalone servers, and wait for 2.0 to build cluster ?

pveca -l shows the same on both servers, ok.
Both is 1.8, Kernel 2.6.35 ,identical pveversion except this line:
proxmox-ve-2.6.32: 1.8-32 ,New, installed from new 1.8 iso
proxmox-ve-2.6.35: 1.8-10 ,Old, installed from 1.7 and updated to 1.8, tried aptitude but nothing new.

Hi,
in a cluster, some files from the master will be synced to the other nodes.
This ist also with /etc/pve/storage.cfg and all iso-files and so on.
So it was an bad idea to use an "empty" node as master...

Udo

 

[udo]

Hi.
I test with new server as Cluster Master. No machines created.
Add old server as node.
Then all storage is gone from old, only local storage and no ISO mapping.
I try to make old server as Master, storage still gone, all old KVM running, but no links to disk.
After restart of old server as Master, no machines start.
How do i find and attach the storage ?
What did i do wrong ? Must the old server be master, or do you always loose storage links when building cluster ? Shall i fix this now as standalone servers, and wait for 2.0 to build cluster ?

pveca -l shows the same on both servers, ok.
Both is 1.8, Kernel 2.6.35 ,identical pveversion except this line:
proxmox-ve-2.6.32: 1.8-32 ,New, installed from new 1.8 iso
proxmox-ve-2.6.35: 1.8-10 ,Old, installed from 1.7 and updated to 1.8, tried aptitude but nothing new.

BTW - you must only define the storage again on the master - then all should work (perhaps uploads some cdrom-isos).

Udo

 

[elmacus]

Thanks for quick reply.

>So it was an bad idea to use an "empty" node as master...
But i read the wiki, and searched before doing this, and i thought: what could go wrong?
I didnt expect it to remove my storage.
You mean define in Gui or in config files somewhere ?

Is this a problem ? why are not they the same :
proxmox-ve-2.6.32: 1.8-32
proxmox-ve-2.6.35: 1.8-10

http://pve.proxmox.com/wiki/Proxmox_VE_Cluster

 

[udo]

Thanks for quick reply.

>So it was an bad idea to use an "empty" node as master...
But i read the wiki, and searched before doing this, and i thought: what could go wrong?
I didnt expect it to remove my storage.
You mean define in Gui or in config files somewhere ?

It's the same - you can use the gui - this will write /etc/pve/storage.cfg, which is synced to all nodes

Is this a problem ? why are not they the same :
proxmox-ve-2.6.32: 1.8-32
proxmox-ve-2.6.35: 1.8-10

http://pve.proxmox.com/wiki/Proxmox_VE_Cluster

You use different kernels. Should not be an problem, but you can't use online-migration if you have shared storage. If you want online-migration the kernels must be the same.
(apt-get install proxmox-ve-2.6.35) or 2.6.32 - after that, select the right kernel in /boot/grub/menu.lst if you don't want the newest kernel and reboot.

Udo

 

[elmacus]

I can backtrack the storage in gui, and they get synced to new server, thanks.
But not all, maybe there is some logfile that lists old storage ? I had some complex iSCSI and LVM mixed with NFS.

 

[udo]

I can backtrack the storage in gui, and they get synced to new server, thanks.
But not all, maybe there is some logfile that lists old storage ? I had some complex iSCSI and LVM mixed with NFS.

Hi,
it's allways a good idea to have an backup of /etc/pve, /etc/qemu-server, /etc/vz and /etc/network/interfaces
The required names of the storages are in the config-files of the VMs. See in /etc/qemu-server/VMID.conf on the nodes (and master if there VMs defined).

AFAIK there a no logfiles which described the storage...

Udo

 

[elmacus]

Everything is working, THANKS Udo,
Case closed.

Someone with account on wiki should update to reflect this for future clusternewbies.

 

[tom]

Everything is working, THANKS Udo,
Case closed.

Someone with account on wiki should update to reflect this for future clusternewbies.

feel free to register yourself and add your comments.