Details
-
New Feature
-
Status: Stalled (View Workflow)
-
Major
-
Resolution: Unresolved
-
None
-
None
Description
There's replication specific assertion in mdl.cc:
/*
|
Ensure that if we are trying to get an exclusive lock for a slave
|
running parallel replication, then we are not blocked by another
|
parallel slave thread that is not committed. This should never happen as
|
the parallel replication scheduler should never schedule a DDL while
|
DML's are still running.
|
*/
|
DBUG_SLOW_ASSERT((mdl_request->type != MDL_INTENTION_EXCLUSIVE &&
|
mdl_request->type != MDL_EXCLUSIVE) ||
|
!(get_thd()->rgi_slave &&
|
get_thd()->rgi_slave->is_parallel_exec &&
|
lock->check_if_conflicting_replication_locks(this)));
|
Please trying moving it out to replication code. Needed for another InnoDB cleanup, where we want to remove THD from InnoDB background threads and use MDL_context instead.