Nodes view fails when a node is temporarily down

The Cloud->Nodes view lists all nodes, grouped by host. However, if a node is temporarily down (not in live_nodes), the whole screen is broken, and instead error messages like

Requested node example.com:8983_solr is not part of cluster

This happens on the Ajax request to fetch admin/metrics and admin/info/system. A better approach would be to skip requesting metrics for downed nodes but still display them in the list, perhaps with a DOWN label or other background colour, to clearly tell that the node is configured but not live.