How to remove peer info in Donor node

– status info –
wsrep_cluster_name=‘g_test1’

node1 - Donor (192.168.0.1)
node2 - New node (192.168.0.2)

Node2 was safely stopped after node1 and node2 joined successfully

However, node1 is constantly attempting to reconnect node2.

[Note] WSREP: (d45ecf8d, 'tcp://0.0.0.0:4567') reconnecting to d9b63250 (tcp://192.168.0.2:4567), attempt 48720
[Note] WSREP: (d45ecf8d, 'tcp://0.0.0.0:4567') reconnecting to d9b63250 (tcp://192.168.0.2:4567), attempt 48750

How to remove peer info (192.168.0.2) in Donor node?
I want stop trying reconnecting

Because, Node2 handshake fail when joining other cluster group (g_test_new)

Node2 error log

[Note] WSREP: handshake failed, my group: 'g_test_new', peer group: 'g_test1'

This should be the same for all nodes. If the name is not the same, then a node cannot join. node1 will continue to attempt to node2 because node2 had a failure condition.
You can force node1 back to bootstrap mode, which should stop connection attempts:

SET GLOBAL wsrep_provider_options='pc.bootstrap=YES';
1 Like

Thank you for your kind reply

1 Like

Sorry I tried to execute the command, but it didn’t work (ignore command)

I stopped node2 and ran the command on node1.

SET GLOBAL wsrep_provider_options=‘pc.bootstrap=YES’;

node1 error log
[Note] WSREP: Ignore ‘pc.bootstrap’ in PRIM
[Note] WSREP: (d45ecf8d, ‘tcp://0.0.0.0:4567’) reconnecting to d9b63250 (tcp://192.168.0.2:4567), attempt 1020
[Note] WSREP: (d45ecf8d, ‘tcp://0.0.0.0:4567’) reconnecting to d9b63250 (tcp://192.168.0.2:4567), attempt 1050
[Note] WSREP: (d45ecf8d, ‘tcp://0.0.0.0:4567’) reconnecting to d9b63250 (tcp://192.168.0.2:4567), attempt 1080

I hope Node1 forgets Node2

1 Like

Can you SET GLOBAL wsrep_cluster_address=gcomm:// or is that read-only? If that is read-only, then you must restart node1 to make it forget about node2. It will never forget on its own.

2 Likes