hcq/MaxScale
hcq/MaxScale
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Fix prepared_statement test

Browse Source 
The test did not synchronize the slaves before doing a read that depended
on replication being in sync. It also didn't clear the previous value from
the buffer used to store the test form results which caused a misleading
error message.
This commit is contained in:
Markus Mäkelä 2017-08-13 10:38:31 +03:00
parent 2430f1b6b1
commit 24bec67d90
1 changed files with 4 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
maxscale-system-test/prepared_statement.cpp
View File

@ -55,6 +55,7 @@ void test_routing(TestConnections& test)
// Test reads inside transactions are routed to master
strcpy(buf, "-1");
test.try_query(test.conn_rwsplit, "BEGIN");
test.add_result(find_field(test.conn_rwsplit, "EXECUTE ps1", "@@server_id", buf),
"Execute should succeed");
@ -63,6 +64,7 @@ void test_routing(TestConnections& test)
test.try_query(test.conn_rwsplit, "COMMIT");
// Test reads inside read-only transactions are routed slaves
strcpy(buf, "-1");
test.try_query(test.conn_rwsplit, "START TRANSACTION READ ONLY");
test.add_result(find_field(test.conn_rwsplit, "EXECUTE ps1", "@@server_id", buf),
"Execute should succeed");
@ -71,10 +73,12 @@ void test_routing(TestConnections& test)
test.try_query(test.conn_rwsplit, "COMMIT");
// Test prepared statements that modify data
strcpy(buf, "-1");
test.try_query(test.conn_rwsplit, "CREATE OR REPLACE TABLE test.t1 (id INT)");
test.try_query(test.conn_rwsplit, "PREPARE ps2 FROM 'INSERT INTO test.t1 VALUES (?)'");
test.try_query(test.conn_rwsplit, "SET @a = 1");
test.try_query(test.conn_rwsplit, "EXECUTE ps2 USING @a");
test.repl->sync_slaves();
test.add_result(find_field(test.conn_rwsplit, "SELECT id FROM test.t1", "id", buf),
"Read should succeed");
res = atoi(buf);