So I've been trying out delivering mail to maildirs in coda, and it
worked fine for about two weeks. Then something happened where any
command trying to access the coda volume hangs. Of course this happened
when I was out of town, so I have a huge queue of mail.

First, I thought this was because I had too many files in my venus cache
(250,000), so I reduced it to 10000.

So I got coda tokens and told the mailserver to start processing the
queue backlog. But maildrop is apparently hanging on accessing any files
in Maildir/tmp, and I have the following messages in venus.log:


[ W(92) : 0000 : 18:04:31 ] WAITING(63d36648.7f000005.127.7cb): level =
RD, readers = 0, writers = 1

[ W(94) : 0000 : 18:05:03 ] WAITING(63d36648.7f000005.127.7cb): level =
RD, readers = 0, writers = 1

[ T(01) : 0506 : 18:06:00 ] BeginRvmFlush (1, 40056, T)
[ T(01) : 0506 : 18:06:00 ] EndRvmFlush

[ X(00) : 0000 : 18:18:41 ] DispatchWorker: out of workers (max 20),
queueing message
18:29:58 DispatchWorker: signal received (seq = 3836)
[ X(00) : 0000 : 18:30:00 ] DispatchWorker: out of workers (max 20),
queueing message



>From an earlier codalist message, I gather that something is going wrong
with the sftp transfers .. 
http://www.coda.cs.cmu.edu/maillists/codalist/codalist-2003/5103.html

I have also had reproduceable problems transfering large files (15mb)
over a 256k link, where it would time out on the backfetch.

How do I go about debugging this?