| 1 |
####################################################################################### |
| 2 |
# WL11570 - GR: options to defer member eviction after a suspicion |
| 3 |
# |
| 4 |
# In a group of 3 servers, we crash one of them, restart it and make it join the group |
| 5 |
# again, which fails. When it finally is expelled from the group, as its suspicion |
| 6 |
# on other members times out, it is finally able to rejoin the group. |
| 7 |
# Test : |
| 8 |
# 0. This test requires a group with 3 servers. |
| 9 |
# 1. Set the group_replication_member_expel_timeout parameter to 3600 seconds. |
| 10 |
# 2. Create table t1 and insert some data, checking that it is replicated to all |
| 11 |
# servers. |
| 12 |
# 3. Crash server 3 and it will restart. |
| 13 |
# This will make server 3 not answer to "I am alive" GCS messages and it will |
| 14 |
# eventually be considered faulty. |
| 15 |
# 4. Check that all members are still in the group on servers 1 and 2, which should |
| 16 |
# both be ONLINE. |
| 17 |
# Server 3 should still be in the group but UNREACHABLE. |
| 18 |
# 5. Insert data into table t1 on server 2 and check that it is replicated to server 1. |
| 19 |
# 6. Restart GR on server 3 and it should fail since its |
| 20 |
# suspicion still hasn't timed out on other members. |
| 21 |
# 7. Wait until server 3 is expelled from the group. |
| 22 |
# 8. Server 3 was expelled! Start GR on server 3. |
| 23 |
# 9. Check that server 3 retrieves the data inserted while |
| 24 |
# it was suspended. Then, use it to insert new data into |
| 25 |
# table t1 which should be replicated to servers 1 and 2. |
| 26 |
# 10. Clean up. |
| 27 |
####################################################################################### |
| 28 |
|
| 29 |
--source ../include/gr_suspect_member_resumes_after_crash.inc |