Description of problem: need 'gluster peer update' command/utility for updating IPs of peers in cluster information If servers are migrated from one network to another or because of some other reasons IP address of all or few peers get changed, - User is not be able to use existing cluster as IP has changed and adding same peer will give below error gluster peer status, will show old IP which is always discoonected ####################################### [root@Rhs3 ~]# gluster p s Number of Peers: 1 Hostname: 10.70.1.111 Uuid: bb035a88-8a41-4fea-9e93-caca9a096d0a State: Peer in Cluster (Disconnected) [root@Rhs3 ~]# gluster peer probe 10.70.45.11 Probe on host 10.70.45.11 port 0 already in peer list ####################################### solution can be --> force detach and attach again but in that case it need to be done (no of peers -1) times and that too from each peer or --> delete /var/lib/glusterd and restart glusterd (which will delete everything) above solutions will fail if user want to use existing volumes and data stored in that ####################################### [root@Rhs3 ~]# gluster volume info Volume Name: d1 Type: Distribute Volume ID: b13d261b-652d-41ba-98f4-743a88e4b1ee Status: Started Number of Bricks: 4 Transport-type: tcp Bricks: Brick1: 10.70.1.3:/home/d11 Brick2: 10.70.1.3:/home/d22 Brick3: 10.70.1.111:/home/d11 Brick4: 10.70.1.111:/home/d22 [root@Rhs3 ~]# gluster volume status Status of volume: d1 Gluster process Port Online Pid ------------------------------------------------------------------------------ NFS Server on localhost 38467 Y 4628 ####################################### Can we have gluster peer update utility which will update peers IP addresses and also updates volume info, so we can use same volume by mounting it again .
Reducing the priority to medium for the following reasons, - The issue is not seen in a normal storage administration workflow. ie, migrating of the cluster to a different LAN is not likely to happen often. - The issue/limitation has been present even before RHS-2.0.
Since our recommendation is to use fqdns over IP, we don't have any plan to introduce this feature at this time and hence closing this bug. Feel free to reopen if you have any strong use case where it is absolutely essential.