Debian Codegraph @

mysql-test/suite/group_replication/include/gr_crash_donor_server.inc

raw
1 ################################################################################
2 # This test verifies whether the group_replication works fine if the
3 # donor server is killed when the joiner is already on recovery phase-2.
4 #
5 # Test:
6 # 0. The test requires three servers: M1,M2 and M3.
7 # 1. Start GR on servers M1 and M2.
8 # 2. Execute some operations on the group through server M1
9 # using procedure dml_operations.
10 # 3. Join M3 to the group and wait until the server is in Recovery.
11 # 4. Identify the donor server for M3.
12 # 5. perform the transactions on the group.
13 # 6. Wait until recovery phase-1 on M3 ends.
14 # 7. Kill and restart the donor server during phase-2 of M3 recovery.
15 # 8.Wait until all the members are online.
16 # 9.Verify that all the members have same data.
17 # 10.Cleanup
18 ################################################################################
19
20 # This test does crashes servers, thence we skip it on valgrind.
21 --source include/not_valgrind.inc
22 --source include/big_test.inc
23 --source include/force_restart.inc
24
25 --source include/set_privilege_checks_user_as_system_user.inc
26 --source include/have_group_replication_plugin.inc
27 --let $rpl_server_count= 3
28 --let $rpl_skip_group_replication_start= 1
29 --source include/group_replication.inc
30
31 # START GR on Two servers M1 and M2
32 --let $rpl_connection_name= server1
33 --source include/rpl_connection.inc
34
35 SET sql_log_bin=0;
36 CREATE TABLE t1(a int primary key);
37 SET sql_log_bin=1;
38
39 # Extract the server_uuid
40 --let $server1_uuid= query_get_value(SELECT @@SERVER_UUID, @@SERVER_UUID, 1)
41
42 --source include/start_and_bootstrap_group_replication.inc
43
44 --let $rpl_connection_name= server2
45 --source include/rpl_connection.inc
46
47 SET sql_log_bin=0;
48 CREATE TABLE t1(a int primary key);
49 SET sql_log_bin=1;
50
51 # Extract the server_uuid
52 --let $server2_uuid= query_get_value(SELECT @@SERVER_UUID, @@SERVER_UUID, 1)
53
54 --source include/start_group_replication.inc
55
56 # Perform the transactions on group through server1 so
57 # that when M3 joins it will be in recovery for some time.
58
59 --let $rpl_connection_name= server1
60 --source include/rpl_connection.inc
61
62 # Create the procedure to perform dml operations
63 delimiter $$;
64 CREATE PROCEDURE dml_operations()
65 BEGIN
66 declare x INT;
67 set x=1;
68 while x<100 do
69 insert into t1 values (x);
70 update t1 set a=x+200 where a=x;
71 delete from t1 where a<250;
72 set x=x+1;
73 end while;
74 end$$
75 delimiter ;$$
76 --echo
77
78 --echo ----call procedure----
79 call dml_operations();
80 --echo
81
82 # Wait for the data to get synced on server2
83 --source include/rpl_sync.inc
84
85 # JOIN M3 to the group
86 --let $rpl_connection_name= server3
87 --source include/rpl_connection.inc
88
89 SET sql_log_bin=0;
90 CREATE TABLE t1(a int primary key);
91 SET sql_log_bin=1;
92 LOCK TABLES t1 READ;
93
94 --let $rpl_connection_name= server_3
95 --source include/rpl_connection.inc
96
97 SET SESSION sql_log_bin= 0;
98 call mtr.add_suppression("There was an error when connecting to the donor server.*");
99 call mtr.add_suppression("For details please check performance_schema.replication_connection_status table and error log messages of Replica I/O for channel group_replication_recovery.");
100 call mtr.add_suppression("Replica I/O for channel 'group_replication_recovery': Source command COM_REGISTER_REPLICA failed: failed registering on source, reconnecting to try again.*");
101 SET SESSION sql_log_bin= 1;
102
103 --let $group_replication_start_member_state= RECOVERING
104 --source include/start_group_replication.inc
105
106 # Identify the donor_server for M3
107
108 --let $wait_timeout= 100
109 --let $wait_condition= SELECT COUNT(*)=1 FROM performance_schema.replication_connection_status WHERE channel_name='group_replication_recovery' AND service_state='ON'
110 --source include/wait_condition.inc
111
112 let $donor_id= 1;
113 let $other_id= 2;
114
115 if(`SELECT COUNT(*)=1 FROM performance_schema.replication_connection_status WHERE channel_name='group_replication_recovery' AND service_state='ON' AND source_uuid='$server2_uuid'`)
116 {
117 let $donor_id= 2;
118 let $other_id= 1;
119 }
120
121 # Connect to server which is neither M3 nor the donor server and
122 # Create procedure to perform dml operations and call it using send.
123 # So that these operations gets executed while server M3 is in recovery.
124
125 --let $rpl_connection_silent= 1
126 --let $rpl_connection_name= server$other_id
127 --source include/rpl_connection.inc
128 # Create the procedure to perform dml operations
129 delimiter $$;
130 CREATE PROCEDURE dml_operations_2()
131 BEGIN
132 declare x INT;
133 set x=100;
134 while x<200 do
135 insert into t1 values (x);
136 update t1 set a=x+200 where a=x;
137 delete from t1 where a<320;
138 set x=x+1;
139 end while;
140 end$$
141 delimiter ;$$
142 --echo
143
144 --echo ----call procedure----
145 send call dml_operations_2();
146 --echo
147
148 # Wait until Recovery phase1 ends.(until the service_state of group_replication_recovery goes to OFF state.)
149 --let $rpl_connection_name= server3
150 --source include/rpl_connection.inc
151 UNLOCK TABLES;
152 --let $wait_timeout=200
153 --let $wait_condition= SELECT COUNT(*)=1 FROM performance_schema.replication_connection_status WHERE channel_name='group_replication_recovery' AND service_state='OFF';
154 --source include/wait_condition.inc
155
156 # Kill the donor server during phase 2 of M3's recovery
157 --let $rpl_connection_name= server$donor_id
158 --source include/rpl_connection.inc
159
160 SET sql_log_bin=0;
161 call mtr.add_suppression(".*Replica SQL for channel 'group_replication_applier': ... The replica coordinator and worker threads are stopped, possibly leaving data in inconsistent state*");
162 SET sql_log_bin=1;
163
164 # kill donor server
165 --echo # killing
166 --let $group_replication_local_address= `SELECT @@GLOBAL.group_replication_local_address`
167 --let $group_replication_group_seeds= `SELECT @@GLOBAL.group_replication_group_seeds`
168 --let $restart_parameters=restart:--group_replication_local_address=$group_replication_local_address --group_replication_group_seeds=$group_replication_group_seeds --group_replication_group_name=$group_replication_group_name
169 --replace_result $group_replication_local_address GROUP_REPLICATION_LOCAL_ADDRESS $group_replication_group_seeds GROUP_REPLICATION_GROUP_SEEDS $group_replication_group_name GROUP_REPLICATION_GROUP_NAME
170 --source include/kill_and_restart_mysqld.inc
171 --echo # restarting
172
173 # Needed as we are not using rpl_restart_server.inc
174 --let $rpl_server_number= $donor_id
175 --source include/rpl_reconnect.inc
176
177 --let $rpl_connection_name= server$other_id
178 --source include/rpl_connection.inc
179 reap;
180 # Wait until group has only 2 members, that is, donor server's death is detected by
181 # the group.
182 --echo # check that there are 2 mebers in the group
183 --let $group_replication_number_of_members= 2
184 --source include/gr_wait_for_number_of_members.inc
185
186 --sleep 5
187
188 # Start GR on killed and restarted donor server and wait until all the servers are ONLINE.
189 # Start GR on donor server
190 --let $rpl_connection_name= server$donor_id
191 --source include/rpl_connection.inc
192
193 --let $wait_timeout= 200
194 --source include/start_group_replication.inc
195
196 --let $rpl_connection_name= server$other_id
197 --source include/rpl_connection.inc
198
199 --let $wait_timeout= 200
200 --let $wait_condition= SELECT COUNT(*) = 3 FROM performance_schema.replication_group_members where MEMBER_STATE='ONLINE';
201 --source include/wait_condition.inc
202
203 --source include/rpl_sync.inc
204
205 # Verify that table t1 has same data on all the servers.
206 --let $diff_tables=server1:t1, server2:t1, server3:t1
207 --source include/diff_tables.inc
208
209 # Clean up.
210 drop table t1;
211 drop procedure dml_operations;
212 drop procedure dml_operations_2;
213 --let $skip_restore_connection= 0
214 --source include/group_replication_end.inc