Thanks, Jan.  I did 'cfs strong' and 'cfs wr' and started looking around
again, and now I see things like this:

13:11:38 Building callback conn.
13:11:38 RevokeWBPermit on conn 1229d80 returned 0
13:11:38 ValidateVolumes: 0x7f000001 failed!
13:11:38 RevokeWBPermit on conn 1229d80 returned 0
13:12:36 RevokeWBPermit on conn 1229d80 returned 0
13:12:36 RevokeWBPermit on conn 1229d80 returned 0
13:12:39 Caution, replicated volume 0x3000002, resolution is turned off.
[...]
13:21:21 Callback failed RPC2_DEAD (F) for ws 199.4.104.153:1301
13:21:21 Unbinding RPC2 connection 169520436
13:21:21 Unbinding RPC2 connection 552446497
13:21:21 Unbinding RPC2 connection 25564687
13:21:21 Unbinding RPC2 connection 703255599

>
> You could fix this in two ways. Either try to avoid disconnections, 'cfs
> wr && cfs strong'. Or set up the client with a larger local cache,
> something like 100MB, so that even when it goes disconnected it will not
> run out of space during the check-in.