We were on 2.98 previously, and in the recent week have upgraded from 2.97 to 2.98. When we did the 2.97 to 2.98 conversion we faced an the issue where the nodes were not connecting. In addition to updating the Swarm plugin from 3.6 to 3.7 we rebuilt the docker images containing our nodes to update Jenkins remoting client from 3.12 to 3.15
I noticed the N/A a few times with jobs, after that, but it was when we went to 2.100 that they seemed to happen all the time.
I did just watch a job which does most of the work on the slave, and was N/A for the entire time on the slave, jump to perform a task on the master. While it was on the master, it did have an estimate. So it seems to be something in the remoting which may not be reporting the time.