After a long slog, I am now able to answer my own question, just in case anybody is listening.
We determined that when we deploy our application to Tomcat using the Tomcat deploy service, which happens when we deploy with Jenkins and Ansible, these errors start. Conversely, if we re-start Tomcat from scratch, the errors go away. Nothing else we tried (and we tried a lot) worked. Our guess is that the Zookeeper libraries we build into our application do something that do not go away, even when the application is re-deployed. This isn't a great answer from us, as we use Ansible to deploy our application to production, and we use Jenkins to continuously deploy in development. But, it is what it is, and at least our logs are readable now. Joe -- Sent from: http://lucene.472066.n3.nabble.com/Solr-User-f472068.html