-
Type:
Bug
-
Resolution: Duplicate
-
Priority:
Major - P3
-
None
-
Affects Version/s: None
-
Component/s: None
-
None
In the repro script for HELP-23347 the MongoClient is not configured with minPoolSize and yet the background thread still attempts to create a connection for the pool. This is a bug.
2021-04-08 11:20:45,234 INFO Connecting to: mongodb://XXX:XXX@redactedhost:27017,redactedhost2:27017,redactedhost3:27017/test?retryWrites=true&authSource=admin ... ... redactedhost:27017 becomes unreachable and minPoolSize is not configured ... 2021-04-08 11:36:48,033 INFO [pool ('redactedhost', 27017)] pool cleared 2021-04-08 11:36:48,529 INFO Topology description updated for topology id 606ee70d3e8b5aff0ebb6fcb 2021-04-08 11:36:49,441 INFO Command insert with request id 1469543483 on server ('redactedhost', 27017) failed in 933415140 microseconds, with error: {'errmsg': '[Errno 113] No route to host', 'errtype': 'OSError'} 2021-04-08 11:36:49,441 INFO [pool ('redactedhost', 27017)][conn #1] connection closed, reason: error 2021-04-08 11:36:49,441 INFO [pool ('redactedhost', 27017)][conn #1] connection checked into pool ... 2021-04-08 11:36:49,530 INFO [pool ('redactedhost', 27017)][conn #2] connection created 2021-04-08 11:36:51,105 INFO [pool ('redactedhost', 27017)][conn #2] connection closed, reason: error Traceback (most recent call last): File "/usr/local/lib64/python3.7/site-packages/pymongo/pool.py", line 1180, in connect sock = _configured_socket(self.address, self.opts) File "/usr/local/lib64/python3.7/site-packages/pymongo/pool.py", line 988, in _configured_socket sock = _create_connection(address, options) File "/usr/local/lib64/python3.7/site-packages/pymongo/pool.py", line 972, in _create_connection raise err File "/usr/local/lib64/python3.7/site-packages/pymongo/pool.py", line 965, in _create_connection sock.connect(sa) OSError: [Errno 113] No route to host During handling of the above exception, another exception occurred: Traceback (most recent call last): File "/usr/local/lib64/python3.7/site-packages/pymongo/mongo_client.py", line 1752, in _process_periodic_tasks self._topology.update_pool(self.__all_credentials) File "/usr/local/lib64/python3.7/site-packages/pymongo/topology.py", line 441, in update_pool server._pool.remove_stale_sockets(generation, all_credentials) File "/usr/local/lib64/python3.7/site-packages/pymongo/pool.py", line 1152, in remove_stale_sockets sock_info = self.connect(all_credentials) File "/usr/local/lib64/python3.7/site-packages/pymongo/pool.py", line 1187, in connect _raise_connection_failure(self.address, error) File "/usr/local/lib64/python3.7/site-packages/pymongo/pool.py", line 286, in _raise_connection_failure raise AutoReconnect(msg) pymongo.errors.AutoReconnect: redactedhost:27017: [Errno 113] No route to host
What's odd is that the background thread only starts attempting to create connections after the closed connection (which was stalled for 16 minutes) is checked back into the pool. So the order of events are:
- Create pool with minPoolSize 0
- Pool creates a connection for insert operations. This connection is checked in and out many times.
- The primary becomes unreachable and connection remains checked out for 16 minutes.
- SDAM heartbeats fail due to timeout errors and clear the pool many times.
- After 16 minutes, the closed connection is checked back into the pool
- Finally, the background thread immediately attempts to create a connection for the same pool.
I suspect something about the pool's remove_stale_sockets is incorrect or clearing the pool potentially leaves the internal state inconsistent.
- duplicates
-
-
- Closed
-