rbd-mirror support

[Kmgish]

Alwin,

Can you provide specific instructions for a one-way mirror between two ceph clusters installed via pveceph?

 

[Alwin]

We now have a one-way RBD mirror how-to on our wiki. It is a standalone service on the remote ceph cluster that pulls the changes.
https://pve.proxmox.com/wiki/Ceph_RBD_Mirroring

Please feel free to provide feedback on the how-to article and the resulting setup.

 

[Kmgish]

Thank You Alwin!!
The how-to worked perfectly as documented.

K

 

[velocity08]

Also very interested in this - we're looking at migrating a large Xen estate that resides in two datacentres to Proxmox, and would love to be able to replace our DRBD storage with Ceph as part of this.

Hey @hacman happy NY.

just curious you mentioned DRBD are you using this per vm or per host to replicate all VM’s?
are you still using Xen or have you made the switch to Proxmox?

””Cheers
G

 

[velocity08]

We now have a one-way RBD mirror how-to on our wiki. It is a standalone service on the remote ceph cluster that pulls the changes.
https://pve.proxmox.com/wiki/Ceph_RBD_Mirroring

Please feel free to provide feedback on the how-to article and the resulting setup.

Hey @Alwin happy NY.

im just checking in on if there has been any more detailed testing on the rbd-mirror feature in ProxMox?
it’s something that has caught my eye vs ZFS replication.

wondering if the replication is any better or worse for DR.

would it be considered HA or DR ?

””Cheers
G

 

[Alwin]

would it be considered HA or DR ?

It can't be HA, since the two cluster don't know anything about each other.

 

[velocity08]

It can't be HA, since the two cluster don't know anything about each other.

OK so its DR.

Is there an automatic way to trigger a fail-over or is this all manually done at this stage?

So would you say its equivalent to ZFS replication between sites?

Any additional positives/ negatives of using RMB-Mirroring?

thanks always appreciate your input.

speak soon.

""Cheers
G

 

[Alwin]

Is there an automatic way to trigger a fail-over or is this all manually done at this stage?

That's up to your automation and procedures.

So would you say its equivalent to ZFS replication between sites?
Any additional positives/ negatives of using RMB-Mirroring?

Not quite. And see Ceph's docs.
https://docs.ceph.com/en/latest/rbd/rbd-mirroring/

 

[velocity08]

That's up to your automation and procedures.


Not quite. And see Ceph's docs.
https://docs.ceph.com/en/latest/rbd/rbd-mirroring/

Thanks for this reference

so the point of difference if selected is Journal-based replication as an option compared to snapshot replication which is what ZFS uses.

both are crash consistent except that "Journal-based" replication can be more accurate up to the second compared to snapshot replication.

did i miss anything?

Are there automation process built into Ceph or Proxmox for the failover between rempte clusters or is this something that would need to be created from scratch?

""Cheers
G

 

[Alwin]

did i miss anything?

That you need Ceph Octopus to have snapshot based mirroring.

Are there automation process built into Ceph or Proxmox for the failover between rempte clusters or is this something that would need to be created from scratch?

As said, this is up to your automation. The rbd-mirror is a daemon that pulls from the remote cluster.