Storage replication - Avoid data loss when migrating between nodes

[rafafell]

Hi,

In the following scenario:

A VM XXX from node A is replicated on node B. Then VM XXX is migrated to Node B and there are files and folders (for example "folder z") from VM XXX located on node A that have not yet been replicated to Node B VM.


If the replication of VM XXX now located on node B is made, will the extra files that are in the replica of VM XXX on node A ("folder z") be lost?

I believe that it would be interesting for replication to have the option of not deleting data located in a given node when we migrate a vm from one node to another. Having an option to increment the data more in the replicated VM that has less data would be good.

I think of the following scenario:

a node A needs unexpected maintenance. With that, I use the replication (I migrate) of the VM on node B (with less data than the VM on node A, because data replication had not yet been done), new data is inserted in the VM located on node B. When node A comes back from unexpected maintenance, the extra data it owns is synchronized to the Node B VM. And the new data from the Node B VM is synchronized to the Node A VM.

Is this not possible with Storage Replication (pvesr)? Do you have a native proxmox solution to avoid this "data loss"?

best,

 

[velocity08]

Hi, In the following scenario: A VM XXX from node A is replicated on node B. Then VM XXX is migrated to Node B and there are files and folders (for example "folder z") from VM XXX located on node A that have not yet been replicated to Node B VM. If the replication of VM XXX now located on node B is made, will the extra files that are in the replica of VM XXX on node A ("folder z") be lost? I believe that it would be interesting for replication to have the option of not deleting data located in a given node when we migrate a vm from one node to another. Having an option to increment the data more in the replicated VM that has less data would be good. I think of the following scenario: a node A needs unexpected maintenance. With that, I use the replication (I migrate) of the VM on node B (with less data than the VM on node A, because data replication had not yet been done), new data is inserted in the VM located on node B. When node A comes back from unexpected maintenance, the extra data it owns is synchronized to the Node B VM. And the new data from the Node B VM is synchronized to the Node A VM. Is this not possible with Storage Replication (pvesr)? Do you have a native proxmox solution to avoid this "data loss"? best,

hey @rafafell

from my understanding proxmox replication is based on snapshots which are taken in increments at given time intervals.

its not streaming replication in real time.

if you are looking for failover in close to real-time with close to zero data loss then some form of shared storage is required.

you can use a San, nas, if you have 3 or more hosts can setup a ceph cluster.

last 2 option for 2 nodes;

- application aware replication where there is a streaming DB, load balancer etc and the application replicated from within the VM to a second VM on another node.

- drbd on the 2 VM’s on different nodes can give you active/ passive replication for DR.

the last 2 options are not ProxMox aware and don’t need to be as it’s all handled in the vm OS.

DRBD is built into the Linux kernel, Windows has its own replication options you’ll need to investigate.

all have their own positives and negatives you just need to work out which ones meet your requirements.

””Cheers
G