Pipeline was working on kubernetes plugin 3580.v78271e5631dc, upgraded to 3636.v84b_a_1dea_6240 and it no longer works.

Brief synopsis - when using a podTemplate(containerTemplate()){node{

with a container(){} and an SH step, the pipeline completely fails (to the extent that even try-catch-finally is skipped). Errors seem to suggest a null pointer.

Other steps tested, and not using container selection, don't seem to cause a problem.

JNLP reports (I think):

17:29:23 Caused by: java.net.ProtocolException: Expected HTTP 101 response but was '500 Internal Server Error'
17:29:23 at okhttp3.internal.ws.RealWebSocket.checkResponse(RealWebSocket.java:229)
17:29:23 at okhttp3.internal.ws.RealWebSocket$2.onResponse(RealWebSocket.java:196) 

jenkins logs report (I think, there is an adjustment for BST, but also, it is 1 second after...)

Computer.threadPoolForRemoting [#572] for test-node-4kjng-lb021 terminated: java.nio.channels.ClosedChannelException
Jun 13, 2022 4:29:24 PM WARNING jenkins.util.Listeners lambda$notify$0
null
java.lang.NullPointerException 

Simplest pipeline (take my word perhaps that try/catch isn't working), and logs from pipeline/jenkins attached. Note that the openshift installation is slightly wonky regarding minimum resource specifications (out of our control), so a merge of container specification with resources is a necessary evil