Set thd's state to 'committed' only after it has done closing tables
(called indirectly by relay_group_info::cleanup_context()).

During server start, as wsrep initialization happens before
plugin_init(), segfault may occur if wsrep THDs try to access
the uninitialized maria_hton.

In wsrep, CTAS should be handled like a regular transaction.
Added a test case.

..print a false positive stack trace into the log

Suppress the stacktrace that innobackupex dumps on receiving SIGPIPE.

Restoring the line affected by 5.5=>10.0 merge

Do not use format function attribute for sql_print_xxx() family of
functions as they use a MariaDB-specific extension of printf instead
of one provided by the system.

on REPAIR don't do table-specific stuff for views
(because even if the view has a temp table opened for it,
it's not opened all the way down the engine. In particular,
Aria crashes in maria_status() because MARIA_HA* info - that is
table->table->file->file - is NULL)

Refs codership/mysql-wsrep#33 - change const initialization to assignment to follow the genral style of the file.

Make sure that thd uses the same method to begin and end
OSU operation.

Refs codership/mysql-features#33 - this fixes a bug where wsrep_start_position_update() would pass local_uuid and local_seqno to wsrep_sst_received() instead of real, submitted position.

Sidno initialization was skipped in wsrep_sst_received() if the
received position matched to current position. However, sidno
initialization must be done every time server is started to
add inverted group uuid into sidno map.

  1. factored XID-related functions to a separate wsrep_xid.cc unit.
  2. refactored them to take refrences instead of pointers where appropriate
  3. implemented wsrep_get/set_SE_position to take wsrep_uuid_t and wsrep_seqno_t instead of XID
  4. call wsrep_set_SE_position() in wsrep_sst_received() to reinitialize SE checkpoint after SST was received, avoid assert() in setting code by first checking current position.

The following FLUSH commands are now executed under total
order isolation:
* FLUSH DES_KEY_FILE
* FLUSH HOSTS
* FLUSH PRIVILEGES
* FLUSH QUERY CACHE
* FLUSH STATUS
* FLUSH USER_RESOURCES

refs codership/mysql-wsrep#53 - fencing LOAD DATA splitting related commit to not happen with ALTER processing

Refs #25 - made sure signals that may be set to ignored in mysqld were set to default in the child process.

Going more MTR-friendly - for SST prefer to use mysql client, mysqldump and my_print_defaults from the local build tree.

refs #7 - merged fix from 5.5 branch. However, 5.6 branch has not (yet) surfaced similar applier hang as 5.5