I inherited an issue where a MariaDB Galera Cluster has only one out of 3 nodes working.
Node1: up and running in production and is accepting reads and writes
Node2: fresh install it just need to sync with node 1
Node3: fresh install it just need to sync with node 1
If I run service mysql start
on Node2/Node3 it changes the state of wsrep_local_state_comment
from Synced
to Donor/Desynced
and a SST is triggered but I will have a lot of downtime.
I have a 2TB of databases and it's going to take about 5-7 hours for a newly added node to sync, and I can't afford the downtime (as an idea, during the day I have 1GB/minute of binlogs and at night about 4GB/hour of binlogs).
I take a db backup every night with Percona XtraBackup and I know how to restore it on a fresh server.
I have galera.cache
set at 32GB.
Is there a way of restoring the other 2 nodes without triggering an SST?
I was thinking of restoring the backup of Node1 to Node2 maybe it will trigger an IST.