mariadb, galera cluster, haproxy, keepalived

hello guys

nice being part of you.

recently I’ve setup a galera cluster with haproxy/keepalived as follow

*3 nodes setup
Node1 act as writing node
node2,3 are for reading. ( in case of failure of node1 , node2 will be the writing node )

  • i have haproxy-keepalived on all 3 machines.
  • i have a VIP pointing at node1 but the reading sequence configured on haproxy is node2,node3,node1.

my problem is when i was testing the failover scenario.
I shutdown node1 and everything moved to node2/3 ( node2 took the place of writing node )

after getting back node1 the mysql was not able to start as was not able to sync from the other 2 nodes.

my configuration on 3 nodes is as following

#node1
binlog_format = ROW
wsrep_provider = /usr/lib64/galera/libgalera_smm.so
wsrep_cluster_name = “cluster”
wsrep_cluster_address = gcomm://
wsrep_node_address = 192.168.116.68
wsrep_node_name = ‘MariaDBNode1’
wsrep_node_incoming_address = 192.168.116.68
wsrep_sst_method = rsync
wsrep_sst_donor = 192.168.116.69,192.168.116.70

#node2
binlog_format = ROW
wsrep_provider = /usr/lib64/galera/libgalera_smm.so
wsrep_cluster_name = “cluster”
wsrep_cluster_address = gcomm://192.168.116.68,192.168.116.70
wsrep_node_address = 192.168.116.69
wsrep_node_name = ‘MariaDBNode2’
wsrep_node_incoming_address = 192.168.116.69
wsrep_sst_method = rsync

#node3

binlog_format = ROW
wsrep_provider = /usr/lib64/galera/libgalera_smm.so
wsrep_cluster_name = “cluster”
wsrep_cluster_address = gcomm://192.168.116.68,192.168.116.69
wsrep_node_address = 192.168.116.70
wsrep_node_name = ‘MariaDBNode3’
wsrep_node_incoming_address = 192.168.116.70
wsrep_sst_method = rsync

please note that before my final testing i had
wsrep_cluster_address = gcomm://192.168.116.68,192.168.116.69,192.168.116.70
on all nodes.

i changed it now as shown on the above configuration

i would appreciate if you could help me.

any help guys?