|
When I run the test locally on a rotating disk, it takes very long and/or frequently fails:
|
10.4 6adb0d2f
|
rpl.rpl_shutdown_wait_slaves 'innodb,mix' [ fail ]
|
Test ended at 2020-05-09 21:52:54
|
|
CURRENT_TEST: rpl.rpl_shutdown_wait_slaves
|
analyze: sync_with_master
|
mysqltest: In included file "/data/bld/10.4-debug-nightly/mysql-test/suite/rpl/include/rpl_shutdown_wait_slaves.inc":
|
included from /data/bld/10.4-debug-nightly/mysql-test/suite/rpl/t/rpl_shutdown_wait_slaves.test at line 11:
|
At line 52: sync_with_master failed: 'select master_pos_wait('master-bin.000001', 162475, 300, '')' returned -1 indicating timeout after 300 seconds
|
|
The result from queries just before the failure was:
|
include/rpl_init.inc [topology=1->2, 1->3, 1->4]
|
connection server_1;
|
CREATE TABLE t1 (a INT) ENGINE=innodb;
|
connection server_2;
|
connection server_3;
|
connection server_4;
|
include/stop_slave.inc
|
connection server_1;
|
connection server_1;
|
SET @@GLOBAL.debug_dbug="+d,simulate_delay_at_shutdown";
|
connection server_4;
|
include/start_slave.inc
|
connection server_1;
|
SHUTDOWN WAIT FOR ALL SLAVES;
|
connection server_4;
|
It might be not a particularly fast disk, but it's not exceptionally slow either, works all right for general purposes.
The master only manages to do a few INSERTs per second, so with 1000 INSERTs in the test it takes ~300 seconds. I'm not sure what happens afterwards and why sync fails after another 300 seconds, it needs investigation. Vardir is attached.
|
var.tar.gz