rht_error returns any permanent error that exists on a replica through err_ptr. A permanent error is an error from which the replica cannot recover. Most RHT calls (except for rht_fini) on the handle of such a replica will fail with the same permanent error code until rht_fini is called.

Use of the default RHT_DIRTY_READ flag will cause rht_error to return the error condition immediately without waiting for any locally generated updates to be applied. Whereas, specifying the RHT_CLEAN_READ will first apply all locally generated updates, and then return the error code.