5.0.0-rc0 clusters (standalone mongod, mongos) running on macOS, I see many errors like this "ctx":"listener","msg":"Can't set socket option","attr":{"optname":"TCP_KEEPALIVE","error":"Invalid argument"}}
The log is flooded with these errors:
{"t":{"$date":"2021-06-04T13:16:52.682-07:00"},"s":"E", "c":"NETWORK", "id":23206, "ctx":"listener","msg":"Can't set socket option","attr":{"optname":"TCP_KEEPALIVE","error":"Invalid argument"}} {"t":{"$date":"2021-06-04T13:16:52.682-07:00"},"s":"W", "c":"NETWORK", "id":23023, "ctx":"listener","msg":"Error accepting new connection","attr":{"error":{"code":9001,"codeName":"SocketException","errmsg":"remote_endpoint: Invalid argument"}}} {"t":{"$date":"2021-06-04T13:16:52.682-07:00"},"s":"I", "c":"NETWORK", "id":23018, "ctx":"listener","msg":"Error accepting new connection on local endpoint","attr":{"localEndpoint":"127.0.0.1:27017","error":"Invalid argument"}}
The connections that error are coming from an otherwise idle load balancer (haproxy). When a driver connects directly to the mongod/s the error does not occur. What does this error mean and why do these errors occur for haproxy connections?
The load balancer is started and stopped with this run-load-balancer.sh script, like this:
$ sudo MONGODB_URI='mongodb://localhost:27017,localhost:27018/' .evergreen/run-load-balancer.sh start Starting HAProxy... Single Mongos LB: mongodb://127.0.0.1:8000/?loadBalanced=true Multiple Mongos LB: mongodb://127.0.0.1:8001/?loadBalanced=true $ sudo MONGODB_URI='mongodb://localhost:27017,localhost:27018/' .evergreen/run-load-balancer.sh stop Stopping HAProxy...
- causes
-
SERVER-69063 Fix TCP keepalive option setting
- Closed
- is depended on by
-
SERVER-60603 Allow connection reset errors without assertion in ASIOSession::ensureSync()
- Closed
- is related to
-
SERVER-61016 Swallow connection reset-related errors received during ASIOSession creation on outbound connection.
- Closed