Tables Cannot Sync Among Galera Clustering Nodes
I built a 3-node cluster using MariaDB 10.3.10. It is for production.
When inserting records into MariaDB by 'source script', records can be sync to node 2 and 3. But when records were inserted by java program (connecting to node 1 and 2 using jdbc url connection string), when one record was inserted into node 1, it couldn't be sync to node 2 and 3.
After I stopped MariaDB in node 2 and 3, started them and rejoined the cluster, the records could be sync.
There is already a 2-node cluster in SIT, MariaDB version 10.3.10. Galera clustering was adopted.
After installation and cluster up and running, we don't stop it. Yesterday, I have checked all the tables' row count. Some tables are sync with same no. of records between the two nodes, but some are not!
For tables not sync issue in SIT, it was caused by the same reason. Some tables were defined as MyISAM at the very beginning. Then ppl performed transactions. Later we discovered that the engine we used was wrong. Then we updated the engine to InnoDB. So row counts were not matched among nodes.
Now all the tables are InnoDB. And we sync them again. Keep monitoring and see whether there are discrepanies.
Please sign in to leave a comment.