Two servers one storage

[triangle]

Good day!

Our organization is considering various virtualization for further implementation. Proxmox quite satisfied and we are willing to pay for commercial support. But is not possible to evaluate all the possibilities Proxmox, since it is impossible to correctly connect the existing data store.

We have storage IBM DS 3500 and two servers connected to it by FiberChannel multipath. It seems to be connected correctly, but the synchronization takes place only in the case of a remount or reboot servers (which of course is not acceptable). And without a single repository, of course most of the necessary functions Proxmox not possible.

With respect and hope for help, triangel.

 

[udo]

Good day!

Our organization is considering various virtualization for further implementation. Proxmox quite satisfied and we are willing to pay for commercial support. But is not possible to evaluate all the possibilities Proxmox, since it is impossible to correctly connect the existing data store.

We have storage IBM DS 3500 and two servers connected to it by FiberChannel multipath. It seems to be connected correctly, but the synchronization takes place only in the case of a remount or reboot servers (which of course is not acceptable). And without a single repository, of course most of the necessary functions Proxmox not possible.

With respect and hope for help, triangel.

Hi,
how do you use the FC-Storages?
I use some FC-Storage as blockdevice for volumegroups (LVM-storage) without trouble (on multible nodes).

Udo

 

[triangle]

Storage should be used for all. Backups, disk images, virtual machine images, snapshots.

I got cut off small pieces of LVM storage for virtual machines and all works, but it is very inconvenient.

Also turned out to connect it as nfs storage resource, and everything works - and backups, and snapshots, and live migration. However, fault tolerance is lost storage appears as a single point of failure for the nfs resource.

With thanks for the response, triangle.

 

[udo]

Storage should be used for all. Backups, disk images, virtual machine images, snapshots.

Hi,
lvm-storage don't have an filesystem - so backups, CTs need an different storage (like you wrote eg. nfs).

I got cut off small pieces of LVM storage for virtual machines and all works, but it is very inconvenient.

Why?
work very well for me:

root@proxmox1:~# pvs
  PV         VG                    Fmt  Attr PSize   PFree  
...
  /dev/sdg1  es2_sata_lun06        lvm2 a--  952.74g 694.73g
  /dev/sdh1  es2_sata_lun16        lvm2 a--    3.64t   2.38t
  /dev/sdk1  es3_sata_lun1         lvm2 a--    2.82t 141.10g
...
root@proxmox2:~# pvs
  PV         VG                    Fmt  Attr PSize   PFree  
...
  /dev/sdg1  es2_sata_lun06        lvm2 a--  952.74g 694.73g
  /dev/sdh1  es2_sata_lun16        lvm2 a--    3.64t   2.38t
  /dev/sdk1  es3_sata_lun1         lvm2 a--    2.82t 141.10g
 ---

root@proxmox1:~# lvs 
...
  LV            VG                    Attr      LSize   Pool Origin Data%  Move Log Copy%  Convert
  vm-302-disk-2 es2_sata_lun06        -wi-a----  12.00g                                           
  vm-312-disk-1 es2_sata_lun06        -wi-a----  32.00g                                           
  vm-118-disk-1 es2_sata_lun16        -wi-ao---  10.00g                                           
  vm-118-disk-2 es2_sata_lun16        -wi-ao---  48.00g                                           
  vm-215-disk-2 es2_sata_lun16        -wi-a---- 160.00g                                           
  vm-304-disk-1 es2_sata_lun16        -wi-a----  12.00g
...

You see the VG is active on several nodes and pve activate only the LVs which are needed

Also turned out to connect it as nfs storage resource, and everything works - and backups, and snapshots, and live migration. However, fault tolerance is lost storage appears as a single point of failure for the nfs resource.

With thanks for the response, triangle.

Udo