diff options
author | Monty <monty@mariadb.org> | 2017-01-29 22:10:56 +0200 |
---|---|---|
committer | Sergei Golubchik <serg@mariadb.org> | 2017-02-28 16:10:46 +0100 |
commit | e65f667bb60244610512efd7491fc77eccceb9db (patch) | |
tree | ff549849324d917615ab896afb7e25ee0bfc7396 /mysys/mf_iocache.c | |
parent | d5c54f3990b49d7c7d6a410016e85e8e58803895 (diff) | |
download | mariadb-git-e65f667bb60244610512efd7491fc77eccceb9db.tar.gz |
MDEV-9573 'Stop slave' hangs on replication slave
The reason for this is that stop slave takes LOCK_active_mi over the
whole operation while some slave operations will also need LOCK_active_mi
which causes deadlocks.
Fixed by introducing object counting for Master_info and not taking
LOCK_active_mi over stop slave or even stop_all_slaves()
Another benefit of this approach is that it allows:
- Multiple threads can run SHOW SLAVE STATUS at the same time
- START/STOP/RESET/SLAVE STATUS on a slave will not block other slaves
- Simpler interface for handling get_master_info()
- Added some missing unlock of 'log_lock' in error condtions
- Moved rpl_parallel_inactivate_pool(&global_rpl_thread_pool) to end
of stop_slave() to not have to use LOCK_active_mi inside
terminate_slave_threads()
- Changed argument for remove_master_info() to Master_info, as we always
have this available
- Fixed core dump when doing FLUSH TABLES WITH READ LOCK and parallel
replication. Problem was that waiting for pause_for_ftwrl was not done
when deleting rpt->current_owner after a force_abort.
Diffstat (limited to 'mysys/mf_iocache.c')
-rw-r--r-- | mysys/mf_iocache.c | 5 |
1 files changed, 4 insertions, 1 deletions
diff --git a/mysys/mf_iocache.c b/mysys/mf_iocache.c index f891a22b17d..8687c2e0c48 100644 --- a/mysys/mf_iocache.c +++ b/mysys/mf_iocache.c @@ -1815,6 +1815,7 @@ int my_b_flush_io_cache(IO_CACHE *info, It's currently safe to call this if one has called init_io_cache() on the 'info' object, even if init_io_cache() failed. This function is also safe to call twice with the same handle. + Note that info->file is not reset as the caller may still use ut for my_close() RETURN 0 ok @@ -1850,10 +1851,12 @@ int end_io_cache(IO_CACHE *info) if (info->type == SEQ_READ_APPEND) { /* Destroy allocated mutex */ - info->type= TYPE_NOT_SET; mysql_mutex_destroy(&info->append_buffer_lock); } info->share= 0; + info->type= TYPE_NOT_SET; /* Ensure that flush_io_cache() does nothing */ + info->write_end= 0; /* Ensure that my_b_write() fails */ + info->write_function= 0; /* my_b_write will crash if used */ DBUG_RETURN(error); } /* end_io_cache */ |