-
Type:
Improvement
-
Resolution: Fixed
-
Priority:
Major - P3
-
Affects Version/s: 5.0.0-rc1
-
Component/s: None
-
Fully Compatible
-
v5.0
-
Service Arch 2021-06-28
-
135
Running any given set of repl/sharding jsTests on 4.4 and 5.0 has a difference in the amount of information-level logs emitted when a connection is closed.
4.4:
[js_test:hash_crud] d20021| 2021-06-10T15:36:57.357+00:00 I NETWORK 22944 [conn30] "Connection ended","attr":{"remote":"10.122.10.199:59354","uuid":"ca7a0c27-d846-4fc1-942b-b33289bd2498","connectionId":30,"connectionCount":12}
5.0:
[js_test:hash_crud] d20021| 2021-06-10T15:36:57.357+00:00 W EXECUTOR 4910400 [conn30] "Terminating session due to error","attr":{"error":{"code":6,"codeName":"HostUnreachable","errmsg":"Connection closed by peer"}} [js_test:hash_crud] d20021| 2021-06-10T15:36:57.357+00:00 I NETWORK 5127900 [conn30] "Ending session","attr":{"error":{"code":6,"codeName":"HostUnreachable","errmsg":"Connection closed by peer"}} [js_test:hash_crud] d20021| 2021-06-10T15:36:57.357+00:00 I NETWORK 22944 [conn30] "Connection ended","attr":{"remote":"10.122.10.199:59354","uuid":"ca7a0c27-d846-4fc1-942b-b33289bd2498","connectionId":30,"connectionCount":12}
This behavior can quickly swamp the log – in a given jsTest, I see an average of 200 connection ended logs (mostly due to the hello command*). On top of this, the server triage team has concerns that the "terminating session due to error" message will scare users who are trying to upgrade from 4.4 to 5.0 (HELP-25086).
This ticket is aimed to suppress these extra logs if possible, and change them to debug level.
*This was verified by adding printStackTrace to the lowest-level of these messages, and the vast majority of these messages were coming from a connection being closed while the recipient side of a hello command was waiting in ReplicationCoordinator::awaitHelloResponse
- related to
-
-
- Backlog
-