You will find step-by-step instructions to prepare your HPC Cloud project for the datacenter move below the announcement.
The HPC Cloud will not be available from 2 to 11 September.
During this time the HPC Cloud hardware will move to the new data center. We expect to resume normal operation on 12 September.
Make sure that all your VMs are properly shut down: your list of VMs should be empty by the end of Thursday 1 September. If you feel you really need to keep some VM instances, please contact our advisors to minimize the risk of data loss.
We are doing everything possible to minimize the risk of data loss. If your VM list is empty, this risk will not be greater than during normal operation. Nevertheless, we always recommend users with valuable datasets to keep a copy elsewhere. See also our backup policy.
Please call or email your advisor, or contact our helpdesk: firstname.lastname@example.org if you have questions about the move.
The safe way to prepare for the move is to shut down your VMs in time. “Shut down” means using the button from the Cloud’s user interface and check that the VM is gone. As part of the shut down procedure, the persistent disks of the VM will be copied back to our Ceph cluster.
If your VM changed valuable data on non-persistent disks, you should copy that to another location before shutting down: changes to data on non-persistent disks is lost when the VM shuts down.
For more information on (non-) persistency, see Image Persistence.
How can you check your disks for persistency?
Important: do not shut down using a command or button inside your VM. This will leave your VM alive in “power down” mode.
Check that the VM proceeds with shutdown: use the refresh button (upper left ). The VM should take no more than a few minutes to shut itself down, after which the Cloud will copy your disks back to the disk storage.
If, after a few minutes, the VM is running again (“State” is “ACTIVE” and “LCM State” is “RUNNING”), then your VM did not react correctly on the BIOS signal to shut down. See VM not reacting to Shutdown for a solution.
Repeat this for the other running VMs. VMs that are suspended, stopped or otherwise alive but not running (or shutting down) need to be made running and then shut down as described above.