This is why the mysql PXC cluster is down
Our architecture is 3 main strong consistent.
Due to executing an SQL to update a table with 10W records, but this table does not have a primary key, after the successful update of master 1, when the strong consistent synchronization is performed, each update is a full table scan, so that when the master 2 is synchronized, the synchronization spends The time is 10W*10W, the main 2 collapses instantly, and then refuses to serve, which will also hang the main 1.
At this time, the external service is no longer provided, so the mysql read initial packet error is reported.