Closed timja closed 5 years ago
Moving this to JENKINS rather than INFRA, the latter of which is only for Jenkins project infrastructure tasks.
I'm not sure if this is a real bug or configuration error however :-/
No problem... When I had tried to open the request I wanted to choose "Jenkins", but there was only Jenkins(website) option.
And I'm also sure that the source of the problem is in correct configuration, but anyway I did not found the correct working configuration.
One more thing (not related to JNLP, but adding information on connectivity) - automatically scaled agents (AWS ec2 windows slaves) in such situation continue to work (with other problems not relevant to the context of this issue).
Sorry, I pushed the wrong button.
ilyag, you say that there was a high CPU load. Was that on the master? An overloaded system can certainly experience a number of different problems, including slow response causing a loss of connection. Have you been able to determine anything about the cause of that high load? Is it the Jenkins master process itself? Or something else on that system? If it is the Jenkins server, then you should see if you can determine what it is doing and what is causing the load. Perhaps specific jobs, maintenance, or plugins.
The first thing to do is to try and isolate something of the cause. Only after that is it possible to try to figure out what needs to be corrected in which component.
frequently Windows JNLP agents can't connect to Master. When it happens we see high CPU load. Only restart of master helps.
I did not found that specific problem between 6 existing issues that looks the same.
In logs of client:
and
in logs of master:
Reconnect of manual JNLP does not helps.
Windows service continues running launched state without crash.
I've attached logs of master and slave before restart.
Here the list of plugins:
Thank you in advance,
Ilya
Originally reported by ilyag, imported from: frequently Windows JNLP agents can't connect to Master