-
Type:
Task
-
Resolution: Done
-
Priority:
Major - P3
-
Affects Version/s: None
-
Component/s: None
-
None
In cases where there is a lot of dirty content in cache for a table, calling drop can be slow.
For example, I created the following wtperf configuration file:
conn_config="cache_size=5G" table_config="type=file" icount=5000 random_range=100000000 report_interval=5 run_time=30 populate_threads=1 threads=((count=8,insert=1))
Then made a modification to wtperf so that it explicitly drops tables before shutting down. When I do that the drop takes between 345 seconds and 510 seconds.
The most common stack from pmp is:
__split_parent,__wt_row_ikey,__wt_split_multi,__evict_page_dirty_update,__wt_evict,__wt_evict_file,__wt_cache_op,__checkpoint_worker,__wt_checkpoint_close,__wt_conn_btree_sync_and_close,__wt_conn_dhandle_close_all,__drop_file,__wt_schema_drop,__drop_table,__wt_schema_drop,__session_drop,drop_all_tables,start_run,start_all_runs,main
- is related to
-
- Closed