mtr tests for cluster node

[Seppo Jaakola]

This bug will be used as a tracker for all the development that is needed for enabling the testing of single wsrep node cluster with mysql test run (mtr) framework.

There is a sibling bug: lp:856162, for which the goal is to make sure that mtr testing for wsrep node, with no wsrep_provider, will be successful. Whereas, In this bug, we are testing node with wsrep_provider functionality.

Following sub tasks should be worked on:
* figure out and document a suitable way for mtr to launch wsrep nodes
* find out which mtr tests pass, and which cannot be supported
   For non supported tests:
   * document the reason why they must be skipped
   * maintain a list of non-supported tests, so that they can be skipped by mtr
* some tests, that should pass, will show errors, work on these case by case

[Seppo Jaakola (seppo-jaakola)]

innodb/innodb test for a debug build wsrep node failed for a debug assert in handler.cc:

in wsrep_innobase_mysql_sort()
...
  DBUG_ASSERT(tmp_length <= str_length);

This assert is too strong, fix was pushed in wsrep-5.6, in revision:
http://bazaar.launchpad.net/~codership/codership-mysql/5.6/revision/4034

[Seppo Jaakola (seppo-jaakola)]

innodb/innodb caused problems for wsrep-5.5 debug build in part:

create table t1 (a int not null , b int, primary key (a)) engine = innodb;
create table t2 (a int not null , b int, primary key (a)) engine = myisam;
insert into t1 VALUES (1,3) , (2,3), (3,3);
select * from t1;
insert into t2 select * from t1;

...where the insert..select from innodb table to myisam table caused debug level assert during binlog commit. Problem was that wsrep_cleanup_transaction had wiped out trx cache earlier (but not stmt cache)

Fix was pushed in wsrep-5.5, in revision: http://bazaar.launchpad.net/~codership/codership-mysql/wsrep-5.5/revision/3942
and, as said, wsrep-5.6 is not affected by this

[Philip Stoev (philip-stoev-f)]

All of the items in this ticket have been completed.

[Shahriyar Rzayev (rzayev-sehriyar)]

Percona now uses JIRA for bug reports so this bug report is migrated to: https://jira.percona.com/browse/PXC-1172