That is weird. It looks like the hikari pool is losing the connection to the database, which then means that the node can't check in, so it then gets killed.
I'm not sure why the pool would be losing the connection; hard to know what is happening.
If you have detailed repo steps (which it sounds like you do), please file a bug in the github issues.
I will say that kubernetes is being run by plenty of our customers and I haven't heard of this issue before, but would love to get to the bottom of this.