[BUG] fix transfer during rollback to
This commit is contained in:
@ -131,7 +131,6 @@ public:
|
||||
WRITE_SQL_BY_CONN(connection, "set GLOBAL ob_query_timeout = 10000000000");
|
||||
WRITE_SQL_BY_CONN(connection, "alter system set enable_early_lock_release = False;");
|
||||
WRITE_SQL_BY_CONN(connection, "alter system set undo_retention = 1800;");
|
||||
WRITE_SQL_BY_CONN(connection, "alter system set partition_balance_schedule_interval = '10s';");
|
||||
sleep(5);
|
||||
}
|
||||
|
||||
@ -213,6 +212,7 @@ public:
|
||||
MTL_SWITCH(tenant_id) {
|
||||
TRANS_LOG(INFO, "worker to do partition_balance");
|
||||
auto b_svr = MTL(rootserver::ObTenantBalanceService*);
|
||||
b_svr->stop();
|
||||
b_svr->reset();
|
||||
int64_t job_cnt = 0;
|
||||
int64_t start_time = OB_INVALID_TIMESTAMP, finish_time = OB_INVALID_TIMESTAMP;
|
||||
@ -410,6 +410,22 @@ TEST_F(ObTransferBetweenRollbackTo, transfer_between_rollback_to)
|
||||
}
|
||||
ASSERT_EQ(loc1, loc2);
|
||||
|
||||
while (true) {
|
||||
int64_t transfer_task_count = 0;
|
||||
if (OB_FAIL(SSH::g_select_int64(tenant_id, "select count(*) as val from __all_transfer_task", transfer_task_count))) {
|
||||
TRANS_LOG(WARN, "fail to wait for transfer task");
|
||||
} else if (transfer_task_count == 0) {
|
||||
fprintf(stdout, "succeed wait for balancer v2\n");
|
||||
break;
|
||||
} else if (ObTimeUtility::current_time() - begin_time > 300 * 1000 * 1000) {
|
||||
fprintf(stdout, "ERROR: fail to wait for balancer\n");
|
||||
break;
|
||||
} else {
|
||||
fprintf(stdout, "wait for balancer v2\n");
|
||||
ob_usleep(200 * 1000);
|
||||
}
|
||||
}
|
||||
|
||||
usleep(1000 * 1000);
|
||||
|
||||
global_ls_id.reset();
|
||||
|
||||
Reference in New Issue
Block a user