Prevent infinite retry in hudson.remoting.Launcher.parseJnlpArguments()

[jan-z]

Sometimes when restarting the jenkins master the swarm-client gets stuck in an infinite retry loop in hudson.remoting.Launcher.parseJnlpArguments().

The error looks like this:

Failing to obtain http://jenkins:8080/computer/slave1/slave-agent.jnlp
java.io.IOException: Failed to load http://jenkins:8080/computer/slave1/slave-agent.jnlp: 404 Not Found
	at hudson.remoting.Launcher.parseJnlpArguments(Launcher.java:399)
	at hudson.plugins.swarm.SwarmClient.connect(SwarmClient.java:284)
	at hudson.plugins.swarm.Client.run(Client.java:135)
	at hudson.plugins.swarm.Client.main(Client.java:87)
Waiting 10 seconds before retry
Failing to obtain http://jenkins:8080/computer/slave1/slave-agent.jnlp
java.io.IOException: Failed to load http://jenkins:8080/computer/slave1/slave-agent.jnlp: 404 Not Found
	at hudson.remoting.Launcher.parseJnlpArguments(Launcher.java:399)
	at hudson.plugins.swarm.SwarmClient.connect(SwarmClient.java:284)
	at hudson.plugins.swarm.Client.run(Client.java:135)
	at hudson.plugins.swarm.Client.main(Client.java:87)
Waiting 10 seconds before retry
...
(Log from swarm-client-3.5)

Setting launcher.noReconnect makes Launcher throw an IOException that gets handled by the swarm-plugin retry mechanism.

[oleg-nenashev]

I need to review potential issues with the external failover logic, but LGTM in general

[oleg-nenashev]

Tested it yesterday. It works as explained in the PR description. 👍

[oleg-nenashev]

Thanks @jan-z! It has been released in 3.7