Open timja opened 12 years ago
Another point of information. The socket cleanups do seem to occur, but NOT while a job is running. I have a single executor and a job that takes a couple of hours. The sockets from the CLI invocations appeared to accumulate and remained open until the that job finished and then were finally released.
Could this be a threading issue?
after using CLI in a loop to update dozens or hundreds of jobs, frequently see (lsof) many many open socks to the jenkins.war process and see (netstat) many tcp sockets in TIME_WAIT.
not sure if I'm interpreting it correctly, but if I am, perhaps
jenkins / cli / src / main / java / hudson / cli / CLI.java
might consider using
http://docs.oracle.com/javase/1.4.2/docs/api/java/net/Socket.html#setSoTimeout%28int%29
with a reasonable short value (10s?) so that when the client finishes, the jenkins.war process can release resources from its end faster?
Originally reported by moncreaff, imported from: Dangling socket connections to jenkins after using cli jar?