How to migrate a ceph RBD pool to a new cluster

[Mimis]

Greetings!!

We are having 2 PvE clusters with 1 ceph RBD pool each, hosted into the same rack. The 2 clusters have not a shared private network.

PvE version:

cluster1: pve-manager/4.4-20/2650b7b5 (running kernel: 4.4.95-1-pve)​

cluster2: pve-manager/5.4-13/aee6f0ec (running kernel: 4.15.18-18-pve)​

Ceph version:

cluster1: ceph version 12.2.12​

cluster2: ceph version 0.94.10​

What we want to achieve, if it is possible, to move the data from cluster1 to cluster2, without or minimum downtime. Is there a way that we can move the data from cluster1 rbd pool to cluster2 rbd pool over public network?
Would it be possible to use a new pool shared between the 2 clusters (maybe with cephfs?) and move disk image with this way, and then move them to the rbd pool of cluster2?
Both clusters are in production and we want to smoothly decommission cluster1.

This is my 1st post, so in case I did not include any valuable info, let me know.

Thank you in advance!

 

[t.lamprecht]

Hi,

cluster1: pve-manager/4.4-20/2650b7b5 (running kernel: 4.4.95-1-pve)

That version is EOL since June 2018...

Would it be possible to use a new pool shared between the 2 clusters (maybe with cephfs?) and move disk image with this way, and then move them to the rbd pool of cluster2?
Both clusters are in production and we want to smoothly decommission cluster1.

Due to the age of your installation cephfs isn't even really available on cluster 1, it got declared stable with ceph Jewel (v10.2.0).

What we want to achieve, if it is possible, to move the data from cluster1 to cluster2, without or minimum downtime. Is there a way that we can move the data from cluster1 rbd pool to cluster2 rbd pool over public network?

There's RBD Mirroring:
https://pve.proxmox.com/wiki/Ceph_RBD_Mirroring

I'd suggest to create a small test RBD pool with a small linux test VM and try all out with that.

If you get issues due to features not available on older cluster1, you could also upgrade the cluster1 to Proxmox VE 5 ( https://pve.proxmox.com/wiki/Upgrade_from_4.x_to_5.0 ) - that upgrade is relatively straight forward.

 

[Alwin]

cluster2: ceph version 0.94.10

To add, Ceph hammer didn't have the rbd-mirror capability yet. This requires Ceph Jewel at least.

 

[sg90]

Greetings!!

We are having 2 PvE clusters with 1 ceph RBD pool each, hosted into the same rack. The 2 clusters have not a shared private network.

PvE version:

cluster1: pve-manager/4.4-20/2650b7b5 (running kernel: 4.4.95-1-pve)​

cluster2: pve-manager/5.4-13/aee6f0ec (running kernel: 4.15.18-18-pve)​

Ceph version:

cluster1: ceph version 12.2.12​

cluster2: ceph version 0.94.10​

What we want to achieve, if it is possible, to move the data from cluster1 to cluster2, without or minimum downtime. Is there a way that we can move the data from cluster1 rbd pool to cluster2 rbd pool over public network?
Would it be possible to use a new pool shared between the 2 clusters (maybe with cephfs?) and move disk image with this way, and then move them to the rbd pool of cluster2?
Both clusters are in production and we want to smoothly decommission cluster1.

This is my 1st post, so in case I did not include any valuable info, let me know.

Thank you in advance!

How big is each VM? What kinda size of CEPH cluster/performance on each do you have?
With such a big difference in versions, I really think you would be limited to powering off and migrating the disk manually and then powering a VM back on, or doing an upgrade on the cluster1 on both PvE and CEPH to be able to use some of the CEPH tools.

Also if cluster2 is a new cluster your making live why still an old version of PvE?

 

[Mimis]

Thank you for sharing all this information.

It has been decided to backup and restore the VMs at the new cluster as the safest solution at the current moment.

Thank you again!!