-
Bug
-
Resolution: Fixed
-
Minor
This is especially critical when JNLP agents are operating behind a NAT router with a short timeout on the routing table (i.e. running in a cloud who's name may or may not rhyme with awe-sure)
- is related to
-
JENKINS-27509 Operations on slave fails after long idle period
-
- Resolved
-
- links to
Code changed in jenkins
User: Stephen Connolly
Path:
src/main/java/hudson/remoting/Engine.java
src/main/java/hudson/remoting/Launcher.java
src/main/java/hudson/remoting/jnlp/Main.java
http://jenkins-ci.org/commit/remoting/ca8702173424f343ddd0bfec0b817f39c30d4549
Log:
[FIXED JENKINS-38539] Turn on SO_KEEPALIVE and provide CLI option to turn it off again
The master side of the connection already has SO_KEEPALIVE enabled, this just allows both OSes to keep their own
guidance and therefore assist when tuning the agent side is more appropriate than changing the kernel parameters on
the master side (as the master is handling the HTTP requests of users)
But there may be users that have the requirement to disable, so safer to provide the option