hdfs dfsadmin refreshSuperUserGroupsConfiguration does not work when config changes are made via CM

hdfs dfsadmin has the ability to refresh the superuser proxy mappings without having to restart all of HDFS via the -refreshSuperUserGroupsConfiguration flag.

However, when I try to use this feature, it doesn't seem to work on CDH clusters.

The procedure I expected to work was:
1. Update proxy config in CDH UI
2. Push client configuration out
3. Run refresh command on namenode host: sudo -u hdfs hdfs dfsadmin -refreshSuperUserGroupsConfiguration

However, the updated configs are not recognized until a full HDFS restart. It looks like the new configs are staged under /var/run/cloudera-scm-agent/process/ccdeploy_<service> but the refresh command doesn't recognize them.

What seems to work is for me to modify the running configs under the current version of the process in /var/run/cloudera-scm-agent/process/<service> and then run refresh.

It seems like a bug that a feature put in place to allow for config refreshes requires a full service restart when running under CDH.

Some other notes about our environment:
-we currently do not use kerberos auth, only simple auth
-we're making use of the impersonation feature with httpfs
-we're putting a fixed list of groups for impersonation for some security controls