-
Type:
Bug
-
Resolution: Done
-
Priority:
Critical - P2
-
Affects Version/s: None
-
Component/s: None
-
Labels:None
There is a (new) deadlock between drop waiting for all operations to drain and an LSM worker thread getting the checkpoint lock in order to start a bulk load:
1 sched_yield,__wt_yield,__lsm_tree_close,__wt_lsm_tree_drop,__wt_schema_drop,__wt_session_drop,__session_drop,_wrap_Session_drop
1 __lll_lock_wait,pthread_mutex_lock,__wt_spin_lock,__wt_curfile_open,__session_open_cursor_int,__session_open_cursor,__wt_bloom_finalize,__lsm_bloom_create,__wt_lsm_work_bloom,__lsm_worker_general_op,__lsm_worker,start_thread,clone
First seen in http://build.wiredtiger.com:8080/job/wiredtiger-test-unit-long/1120/console
- is depended on by
-
SERVER-22388 WiredTiger changes for MongoDB 3.3.2
-
- Closed
-
-
-
- Closed
-