Migrate UCS Standalone server to new hardware

UCS - Univention Corporate Server
1

Hello,

we need to migrate a standalone UCS 5.0-2 (will be updated to 5.0-3) to a new server hardware.
The server is mainly used for file shares. Installed apps are:

  • AD compat. DC
  • DHCP-Server
  • CUPS
  • Self Service
  • Self Service Backend

I found some guides for that, but I’m not sure if they’re still valid and what will be the most reliable approach?

https://wiki.univention.de/index.php/Cool_Solution_-_Single_Server_Backup_and_Restore

https://wiki.univention.de/index.php/Backup_und_Wiederherstellung_eines_UCS-Systems#Wiederherstellung_eines_UCS-Systems

Or should we add the new server as backup server and change backup to primary after migration of all resources (mainly file shares)?
http://docs.software-univention.de/manual/5.0/en/domain-ldap/backup2master.html#domain-backup2master

I would prefer a fail-safe approach with the possibility to switch (temporarily) back to the old hardware in case of any severe problems.

2

Hi thanq
I did it with veeam for linux. A backup to my nas and a restore to the new server.

Ben

3

Hi Ben,

thanks for that hint! Somehow I forgot that there’s a veeam agent for linux for some time now.
I’ll give it a try!

4

Yes - that was my idea too. I have migrate my servers last year to new hardware - from physical Server to virtual Servers (with promox) and i use the veeam agent.

5

or you can use “REAR”, which allows rempapping, resizing of drives.

Another solution is to just pull the drive out and make a disk image, then attach that “software image” to a standard VMware & it will boot up directly. (just before you boot, set a snapshot on your soft image)
you might have to fiddle with the network cards a bit (names) & once it is up shut it down & snapshot it again.

6

Like to give some update here.
My first post was missing some important details: New server has less (the half size!) storage, both servers are equipped with hardware raid controllers. LVM is used on the old system.
I tried with veeam for Linux in my lab with some VMs - apart from some dependencies during the installation, backup and restore was seamless - as long as the target disk was not smaller than the original. I remember also some problem with LVM - not sure if it was generally not supported or only in combination with smaller target disks.
Next try was with Rear - had to watch some tutorial for the beginning, first backup and recovery of a VM was successfull even with a smaller target disk.
Unfortunatly, on the real hardware there popped up some more problems with incompatible (raid / dm) driver and controller firmware. Was fixed after firmware upgrade of the controller (which took hours to get the correct firmware ad broadcoms site). Next try ended with reboot of the server as soon as the recovery image booted. Fixed this by using a more recent version of Rear. Finally I could boot the rescue iso and select the backup and target disk, but to migrate to a substantially smaller disk I would have to manually adapt the disklayout.conf file before running “rear recover” (s. release notes of rear 2.7). Not sure if I’ll give it a try or migrate to a VM on the server by setting up proxmox or ESXi before.