0x80004005 recurs quite often, but there are no other errors. Only 4 of 10 users are affected. There have been no changes to the cluster, and running handheldcleanup -u on the server mitigated the issue for about a day, then went right back to failing.