| 1 |
####################################################################################### |
| 2 |
# WL11570 - GR: options to defer member eviction after a suspicion |
| 3 |
# |
| 4 |
# In a group of 4 servers, we suspend one of them for 300 seconds and test if it is |
| 5 |
# possible to remove a node. The current behavior does not allow the removal of a |
| 6 |
# node from the group and this is verified. |
| 7 |
# |
| 8 |
# 1. Create a group with 4 servers and a table on it. |
| 9 |
# 2. Set the group_replication_member_expel_timeout parameter to 300 seconds |
| 10 |
# 3. Suspend server 3 by sending a signal SIGSTOP to it. |
| 11 |
# This will make server 3 not answer to "I am alive" GCS messages and it will |
| 12 |
# eventually be considered faulty. |
| 13 |
# 4. Check that all members are still in the group on servers 1, 2 and 4, which should |
| 14 |
# be ONLINE. |
| 15 |
# Server 3 should still be in the group but UNREACHABLE. |
| 16 |
# 5. Make server 4 leave the group and notice that it is reported as ONLINE and server |
| 17 |
# 3 as UNREACHABLE. |
| 18 |
# 6. Reset the group_replication_member_expel_timeout parameter to 0 seconds thus |
| 19 |
# forcing server 3 to be expelled. |
| 20 |
# 7. Make server 4 successfully join the group after reseting the above option. |
| 21 |
# 8. Resume server 3 and make it rejoin the group. |
| 22 |
# 9. Clean up. |
| 23 |
####################################################################################### |
| 24 |
|
| 25 |
# Don't test this under valgrind, memory leaks will occur |
| 26 |
--source include/not_valgrind.inc |
| 27 |
# Test involves sending SIGSTOP and SIGCONT signals using kill Linux command. |
| 28 |
--source include/linux.inc |
| 29 |
--source include/big_test.inc |
| 30 |
--source include/force_restart.inc |
| 31 |
--source include/have_group_replication_plugin.inc |
| 32 |
|
| 33 |
--echo |
| 34 |
--echo ############################################################ |
| 35 |
--echo # 1. Create a group with 4 members and a table on it. |
| 36 |
--let $rpl_server_count= 4 |
| 37 |
--source include/group_replication.inc |
| 38 |
|
| 39 |
--let $rpl_connection_name= server1 |
| 40 |
--source include/rpl_connection.inc |
| 41 |
--let $server1_local_address= `SELECT @@GLOBAL.group_replication_local_address` |
| 42 |
|
| 43 |
--echo # |
| 44 |
--echo # Create a table |
| 45 |
--let $rpl_connection_name= server1 |
| 46 |
--source include/rpl_connection.inc |
| 47 |
CREATE TABLE t1 (c1 INT NOT NULL PRIMARY KEY) ENGINE=InnoDB; |
| 48 |
INSERT INTO t1 VALUES (1); |
| 49 |
--source include/rpl_sync.inc |
| 50 |
|
| 51 |
--let $rpl_connection_name= server4 |
| 52 |
--source include/rpl_connection.inc |
| 53 |
set session sql_log_bin=0; |
| 54 |
call mtr.add_suppression("Timeout on wait for view after joining group"); |
| 55 |
call mtr.add_suppression("The member is leaving a group without being on one"); |
| 56 |
call mtr.add_suppression("The member has left the group but the new view will not be installed"); |
| 57 |
set session sql_log_bin=1; |
| 58 |
|
| 59 |
--echo |
| 60 |
--echo ############################################################ |
| 61 |
--echo # 2. Set group_replication_member_expel_timeout to |
| 62 |
--echo # 300 seconds. |
| 63 |
--let $rpl_connection_name= server1 |
| 64 |
--source include/rpl_connection.inc |
| 65 |
SET GLOBAL group_replication_member_expel_timeout = 300; |
| 66 |
SELECT @@GLOBAL.group_replication_member_expel_timeout; |
| 67 |
|
| 68 |
--let $rpl_connection_name= server2 |
| 69 |
--source include/rpl_connection.inc |
| 70 |
SET GLOBAL group_replication_member_expel_timeout = 300; |
| 71 |
SELECT @@GLOBAL.group_replication_member_expel_timeout; |
| 72 |
|
| 73 |
--let $rpl_connection_name= server3 |
| 74 |
--source include/rpl_connection.inc |
| 75 |
SET GLOBAL group_replication_member_expel_timeout = 300; |
| 76 |
SELECT @@GLOBAL.group_replication_member_expel_timeout; |
| 77 |
|
| 78 |
--let $rpl_connection_name= server4 |
| 79 |
--source include/rpl_connection.inc |
| 80 |
SET GLOBAL group_replication_member_expel_timeout = 300; |
| 81 |
SELECT @@GLOBAL.group_replication_member_expel_timeout; |
| 82 |
|
| 83 |
--echo |
| 84 |
--echo ############################################################ |
| 85 |
--echo # 3. Suspend server 3 by sending signal SIGSTOP to it. |
| 86 |
--echo # This will make server 3 not answer to "I am alive" |
| 87 |
--echo # GCS messages and it will eventually be considered |
| 88 |
--echo # faulty. |
| 89 |
--let $rpl_connection_name= server3 |
| 90 |
--source include/rpl_connection.inc |
| 91 |
|
| 92 |
--echo # |
| 93 |
--echo # Get server 3 pid. |
| 94 |
SET SESSION sql_log_bin= 0; |
| 95 |
CREATE TABLE pid_table(pid_no INT); |
| 96 |
--let $pid_file= `SELECT @@GLOBAL.pid_file` |
| 97 |
--replace_result $pid_file pid_file |
| 98 |
--eval LOAD DATA LOCAL INFILE '$pid_file' INTO TABLE pid_table |
| 99 |
--let $server_pid=`SELECT pid_no FROM pid_table` |
| 100 |
DROP TABLE pid_table; |
| 101 |
SET SESSION sql_log_bin= 1; |
| 102 |
|
| 103 |
--echo # |
| 104 |
--echo # Suspending server 3... |
| 105 |
--exec kill -19 $server_pid |
| 106 |
|
| 107 |
--echo |
| 108 |
--echo ############################################################ |
| 109 |
--echo # 4. Check that all members are still in the group on |
| 110 |
--echo # servers 1, 2 and 4 which should be ONLINE. |
| 111 |
--echo # Server 3 should still be in the group but UNREACHABLE. |
| 112 |
|
| 113 |
--let $rpl_connection_name= server1 |
| 114 |
--source include/rpl_connection.inc |
| 115 |
let $wait_condition=SELECT COUNT(*)=3 FROM performance_schema.replication_group_members where MEMBER_STATE="ONLINE"; |
| 116 |
--source include/wait_condition.inc |
| 117 |
let $wait_condition=SELECT COUNT(*)=1 FROM performance_schema.replication_group_members where MEMBER_STATE="UNREACHABLE"; |
| 118 |
--source include/wait_condition.inc |
| 119 |
|
| 120 |
--echo |
| 121 |
--echo ############################################################ |
| 122 |
--echo # 5. Make server 4 leave the group and notice that it is |
| 123 |
--echo # reported as ONLINE and server 3 as UNREACHABLE. |
| 124 |
--echo # |
| 125 |
--echo # Stop GR on server4. |
| 126 |
--let $rpl_connection_name= server4 |
| 127 |
--source include/rpl_connection.inc |
| 128 |
--source include/stop_group_replication.inc |
| 129 |
|
| 130 |
--let $rpl_connection_name= server1 |
| 131 |
--source include/rpl_connection.inc |
| 132 |
let $wait_condition=SELECT COUNT(*)=3 FROM performance_schema.replication_group_members where MEMBER_STATE="ONLINE"; |
| 133 |
--source include/wait_condition.inc |
| 134 |
let $wait_condition=SELECT COUNT(*)=1 FROM performance_schema.replication_group_members where MEMBER_STATE="UNREACHABLE"; |
| 135 |
--source include/wait_condition.inc |
| 136 |
|
| 137 |
--echo |
| 138 |
--echo ############################################################ |
| 139 |
--echo # 6. Reset the group_replication_member_expel_timeout |
| 140 |
--echo # parameter to 0 seconds thus forcing server 3 to be |
| 141 |
--echo # expelled. |
| 142 |
--echo # |
| 143 |
--echo # Reset the group_replication_member_expel_timeout to 0. |
| 144 |
|
| 145 |
--let $rpl_connection_name= server1 |
| 146 |
--source include/rpl_connection.inc |
| 147 |
SET GLOBAL group_replication_member_expel_timeout = 0; |
| 148 |
|
| 149 |
--let $rpl_connection_name= server2 |
| 150 |
--source include/rpl_connection.inc |
| 151 |
SET GLOBAL group_replication_member_expel_timeout = 0; |
| 152 |
|
| 153 |
--let $rpl_connection_name= server4 |
| 154 |
--source include/rpl_connection.inc |
| 155 |
SET GLOBAL group_replication_member_expel_timeout = 0; |
| 156 |
|
| 157 |
--echo # |
| 158 |
--echo # Wait until until server 3 is expelled and 2 servers are online |
| 159 |
--let $rpl_connection_name= server1 |
| 160 |
--source include/rpl_connection.inc |
| 161 |
let $wait_condition=SELECT COUNT(*)=2 FROM performance_schema.replication_group_members where MEMBER_STATE="ONLINE"; |
| 162 |
--source include/wait_condition.inc |
| 163 |
|
| 164 |
--echo |
| 165 |
--echo ############################################################# |
| 166 |
--echo # 7. Make server 4 successfully join the group after reseting |
| 167 |
--echo # the above option. |
| 168 |
--echo # |
| 169 |
--echo # Start GR on server4 |
| 170 |
--let $rpl_connection_name= server4 |
| 171 |
--source include/rpl_connection.inc |
| 172 |
--replace_result $server1_local_address SERVER1_LOCAL_ADDRESS |
| 173 |
--eval SET GLOBAL group_replication_group_seeds= "$server1_local_address" |
| 174 |
--source include/start_group_replication.inc |
| 175 |
|
| 176 |
--echo ############################################################# |
| 177 |
--echo # 8. Resume server 3 and make it rejoin the group |
| 178 |
--echo # |
| 179 |
--exec kill -18 $server_pid |
| 180 |
|
| 181 |
|
| 182 |
--let $rpl_connection_name= server3 |
| 183 |
--source include/rpl_connection.inc |
| 184 |
# Due to BUG#28068548 and BUG#28224165 we cannot check the server status as it |
| 185 |
# may report that the server never left the group. |
| 186 |
#let $wait_condition=SELECT COUNT(*)=1 FROM performance_schema.replication_group_members where MEMBER_STATE="ONLINE" or MEMBER_STATE="ERROR"; |
| 187 |
#--source include/wait_condition.inc |
| 188 |
--source include/stop_group_replication.inc |
| 189 |
--source include/start_group_replication.inc |
| 190 |
|
| 191 |
--echo # |
| 192 |
--echo # Wait until until all servers are online again |
| 193 |
--let $rpl_connection_name= server1 |
| 194 |
--source include/rpl_connection.inc |
| 195 |
let $wait_condition=SELECT COUNT(*)=4 FROM performance_schema.replication_group_members where MEMBER_STATE="ONLINE"; |
| 196 |
--source include/wait_condition.inc |
| 197 |
|
| 198 |
--echo |
| 199 |
--echo ############################################################ |
| 200 |
--echo # 9. Clean up. |
| 201 |
--let $rpl_connection_name= server1 |
| 202 |
--source include/rpl_connection.inc |
| 203 |
DROP TABLE t1; |
| 204 |
--source include/rpl_sync.inc |
| 205 |
|
| 206 |
--source include/group_replication_end.inc |