Another issue we can run into is the following set of messages:

2019-06-08T04:56:24.518538Z 0 [ERROR] WSREP: failed to open gcomm backend connection: 110: failed to reach primary view: 110 (Connection timed out)
         at gcomm/src/pc.cpp:connect():158
2019-06-08T04:56:24.518591Z 0 [ERROR] WSREP: gcs/src/gcs_core.cpp:gcs_core_open():209: Failed to open backend connection: -110 (Connection timed out)
2019-06-08T04:56:24.518764Z 0 [ERROR] WSREP: gcs/src/gcs.cpp:gcs_open():1458: Failed to open channel 'galera_cluster1' at 'gcomm://192.168.0.126,192.168.0.107,192.168.0.114': -110 (Connection timed out)
2019-06-08T04:56:24.518793Z 0 [ERROR] WSREP: gcs connect failed: Connection timed out
2019-06-08T04:56:24.518812Z 0 [ERROR] WSREP: wsrep::connect(gcomm://192.168.0.126,192.168.0.107,192.168.0.114) failed: 7
2019-06-08T04:56:24.518835Z 0 [ERROR] Aborting

To solve this, take a look at your /etc/my.cnf file. The following three fields have to match the server that you are on:

wsrep_node_address="192.168.0.126"
wsrep_node_name="mysql01"
server_id=1
bind-address=192.168.0.126

If they don't, the above error is thrown.  server_id must be unique for each node.  This can happen when you're restoring your database from one node to another or trying recovery steps where you copy the /etc/my.cnf over to another host. 

 

Another solution involves copying the data directory from the master or most current host to the target host we want as a new master.  We do this because we wanted to test when the master is offline while working only with 2/3 nodes.  Here is our situation:

mysql01 GOOD
mysql02 BAD
mysql03 BAD

Copy the data dir from mysql01 to mysql02:

mysql02 # cd /var/lib/mysql; scp -rp mysql01:/var/lib/mysql/  .

Set the safe_to_bootstrap flag to 1:

cat grastate.dat
# GALERA saved state
version: 2.1
uuid:    f25fc12b-8a0b-11e9-b58d-bfb801e3b36d
seqno:   -1
safe_to_bootstrap: 1

Bootstrap this node:

mysql02 # /usr/bin/mysqld_bootstrap

On the third node, mysql03, remove all files from the /var/lib/mysql folder because we'll let it sync up from mysql02:

mysql03 # cd /var/lib/mysql; rm -rf *

Start mysql on mysql03 so it sync's from mysql02:

mysql03 # systemctl start mysqld

Let it sync.  You should have an accessible 2/3 node cluster at this point.

Thx,
TK