Debian Codegraph @

mysql-test/suite/group_replication/t/gr_recovery_donor_failover.test

raw
1 ################################################################################
2 # Recovery donor failover test
3 # This test focus on the failover mechanism where a joiner changes it's donor
4 # whenever the one in use leaves the group while recovery is happening.
5 #
6 # To test this, a new member is started but recovery blocked. While at this state
7 # the donor is killed, forcing the joiner to change to a new one. As there is
8 # no donor identification as of now, we just unblock and check that recovery
9 # was successful
10 #
11 # Test:
12 # 0. The test requires three servers: M1, M2 and M3.
13 # 1. Bootstrap group on M1. Insert some data. Start GR on M2.
14 # 2. Stop the applier on both servers to block recovery.
15 # 3. Start GR on M3. It should go into RECOVERING state. Identify the DONOR.
16 # 4. Stop GR on the donor making it leave the group.
17 # 5. Unblock recovery i.e. Start the applier on failover member and watch M3 go
18 # ONLINE despite the donor exit. Vaidate data on M3.
19 # 6. Clean up.
20 ################################################################################
21
22 --source ../include/gr_recovery_donor_failover.inc