I've seen this very recently. In that case, it was a partially-failed master, but it's possible that any partially-failed node could cause _cat/nodes to error in this way. The failure can't be complete, otherwise the node would simply be skipped. It has to be a partial failure which leaves the node running but unable to properly return a result to the methods triggered by _cat/nodes.
I suggest taking a support diagnostic via https://github.com/elastic/support-diagnostics and then looking at the "nodes.json" file in the output. That's how I discovered which node was at fault.
No problem. Of course now that you found that one node had a disk problem, you probably don't need to run a diagnostic. If that node is fixed, _cat/nodes should work again.
Apache, Apache Lucene, Apache Hadoop, Hadoop, HDFS and the yellow elephant
logo are trademarks of the
Apache Software Foundation
in the United States and/or other countries.