Forums

Home / Forums

You need to log in to create posts and topics. Login · Register

Proper way to move data and nodes between 2 different petasan clusters

I have two petasan clusters.

Cluster 1:
6 nodes - 2 nodes are nearing end of life and need to be decommissioned
provides iscsi service

Cluster 2:
4 nodes - 1 node is nearing end of life and needs to be decommissioned.
Provides iSCSI and CIFS service

We have made extensive changes to our network and the network settings differ between cluster1 and cluster2. iSCSI disks are in different 10.x.x.x networks, etc.

Goal:
Keep Cluster 2 with 6 nodes, 3 of which are currently in Cluster 1 and 3 of which are in Cluster 2. We also want these 6 nodes split between 3 locations, 1 Manager and 1 data node in each location, so that any one location could go down and the cluster remains up. so we want to make sure that the 3/2 copies of the data are set per data center.

The reason we want to keep cluster 2 is it was built after all the network changes. We are not wanting to use the networks that are being used by cluster 1 anymore.

On cluster 2, the old node is failing and needs to be removed. It has crashed / been fenced 2 of the last 3 nights. I have set it so that no iscsi disk is being served from this node. Should I go through and set all of its OSD's to have a crush weight of 0? I did this with 1 OSD this morning and it rebalanced and the cluster health came back to OK, so I think it 'drained' the disk, but the OSD usage still shows 3%. How do I know for sure all data has been moved off that disk before I set the next disk to crush weight of 0?

 

 

What is the best way to accomplish the movement of nodes from Cluster1 to Cluster 2?

Should I start by removing 1 of the nodes from cluster 1, and then adding it to cluster 2 (cluster 1 has more data on it than will fit on cluster 2 with just 3 nodes)
Can I migrate the iSCSI disks from cluster 1 to cluster 2 using petasan replication software? The iscsi networks on  cluster 2 are different than on cluster 1. Can I move the disk and then change the disk IP address on cluster 2 and then reattach to the client virtual machines?

Do you have any suggestions on the best way to handle this?

Thanks,
Neil