I've gotten in the bad habit of ignoring this specific Alert that occurs about ten times a day on my Datomic Cloud platform:
{"Msg":"Uncaught Exception: Connection reset","Ex":{"Via":[{"Type":"java.net.SocketException","Message":"Connection reset","At":["sun.nio.ch.SocketChannelImpl","throwConnectionReset","SocketChannelImpl.java",394]}],"Trace":[["sun.nio.ch.SocketChannelImpl","throwConnectionReset","SocketChannelImpl.java",394],["sun.nio.ch.SocketChannelImpl","read","SocketChannelImpl.java",426],["jdk.internal.net.http.SocketTube","readAvailable","SocketTube.java",1170],["jdk.internal.net.http.SocketTube$InternalReadPublisher$InternalReadSubscription","read","SocketTube.java",833],["jdk.internal.net.http.SocketTube$SocketFlowTask","run","SocketTube.java",181],["jdk.internal.net.http.common.SequentialScheduler$SchedulableTask","run","SequentialScheduler.java",230],["jdk.internal.net.http.common.SequentialScheduler","runOrSchedule","SequentialScheduler.java",303],["jdk.internal.net.http.common.SequentialScheduler","runOrSchedule","SequentialScheduler.java",256],["jdk.internal.net.http.SocketTube$InternalReadPublisher$InternalReadSubscription","signalReadable","SocketTube.java",774],["jdk.internal.net.http.SocketTube$InternalReadPublisher$ReadEvent","signalEvent","SocketTube.java",957],["jdk.internal.net.http.SocketTube$SocketFlowEvent","handle","SocketTube.java",253],["jdk.internal.net.http.HttpClientImpl$SelectorManager","handleEvent","HttpClientImpl.java",979],["jdk.internal.net.http.HttpClientImpl$SelectorManager","lambda$run$3","HttpClientImpl.java",934],["java.util.ArrayList","forEach","ArrayList.java",1511],["jdk.internal.net.http.HttpClientImpl$SelectorManager","run","HttpClientImpl.java",934]],"Cause":"Connection reset"},"Type":"Alert","Tid":48900,"Timestamp":1762258371599}
Is this something to worry about? Can it be mitigated? Should I continue to ignore it?I suspect these are transient HTTP noise. TCP connection resets. LBs, NAT gateways can kill an idle connection and then the first read on that dead connection would throw this exception. I'd worry about it if you saw a sudden spike or any user visible impact and if all attempts fail not just a handful.
Seat-of-the-pants, they seem to happen mostly after-hours, which lends credence to the theory they represent idle connections killed by the infrastructure. I have not noticed any user impacts. Thanks for the interpretation @jaret, I'll sleep better ignoring them.