The configuration of Kuberenetes starts with
Connection Timeout 5
Read Timeout 15

After several runs of jobs (happens after a day and sometimes a month) the configuration is reseted to
Connection Timeout 0
Read Timeout 0

Any change via browser to change it with Apply/Save doesn't pass and the values are kept to 0

This result is creating slaves that starts with error, probably due to lack of contact with master.

Workaround
----------------

kubectl exec -it jenkins_pod bash
edit the /var/jenkins_home/config.xml

edit the variables.

Delete all slaves, afterward kill the master pod. The running pod starts with 5 and 15. and activity is back to normal

I put it on Major but it should be Critical, only because I have a workaround for this.