I gave this patch set a spin this week and I agree that a less granular lock would be a good thing. In addition (and this is a nit), when I looked at the logs, I think it would help if the main thread were identified by (main_thread) for consistency with how other threads report. Here's an example of what shows up now:
2016-04-01T16:11:28.489Z|00008|ovsdb_jsonrpc_server(sessions_thread1)|ERR| thread receive command msg 2016-04-01T16:11:28.489Z|00015|memory|INFO|cells:266325 json-caches:3 monitors:3 sessions:5 2016-04-01T16:21:27.781Z|00016|ovsdb_file| INFO|/opt/stack/data/ovs/ovnsb.db: compacting database online (1458329582.712 seconds old, 1573 transactions, 12266630 bytes) 2016-04-01T16:31:28.950Z|00009|ovsdb_file(sessions_thread1)| INFO|/opt/stack/data/ovs/ovnsb.db: compacting database online (1458329582.971 seconds old, 1623 transactions, 15439453 bytes) 2016-04-01T16:31:30.078Z|00010|timeval(sessions_thread1)|WARN|Unreasonably long 1129ms poll interval (589ms user, 46ms system) Lastly, I'm not sure what the first line above is conveying, but there were a *LOT* of them, so I'd suggest either removing or lowering the log level to DEBUG. Ryan _______________________________________________ dev mailing list dev@openvswitch.org http://openvswitch.org/mailman/listinfo/dev
