Brendan Cully wrote:
On Friday, 07 December 2007 at 12:16, Jason Joines wrote:< a0006 OK STATUS completed. mutt_num_postponed: 4 postponed IMAP messages found. > a0007 NOOP Error talking to mail.okstate.edu (Connection reset by peer) imap_cmd_step: Error reading server response. Mailbox closed imap_exec: command failed:If I'm interpreting this correctly, the first NOOP worked. An attempt was made to send a second NOOP but the mail server couldn't be contacted. Looks to me like a connection issue instead of a server issue. What do you think?Jason ===========I restored my 2s settings and ran the debug version. It does show a successful NOOP every 2s until the "Connection reset by peer" error appears. Even with default settings, sometimes there are several successful NOOPs before the error.It doesn't sound like there's much you can do about this. The server seems to be disconnecting you for some reason. Maybe there's another connection opening up to the same mailbox and your IMAP server doesn't like it?
The earlier post by Kyle in this thread suggested running the debug version to see if exchange was failing to recognize the NOOP option as something to keep the mailbox open. The connection stays open a lot longer when I send one every 2 seconds. At the moment the connection has been open for 33 minutes and I have had it stay open for a few hours this way. The logs look to me like the NOOPs are being successful.
Your earlier post in this thread pointed me to the FAQ at http://wiki.mutt.org/?MuttFaq/RemoteFolder. Part of the next to last entry says, "There is probably a firewall or NAT router between you and your server which drops connections that don't get traffic frequently enough.". Since my sending more traffic by decreasing the time between imap_keepalive messages since to make the problem better and keeps connections open way beyond the 30 minute time that the server could disconnect me via RFC if it hadn't been told to stay open, I'm leaning toward the FAQ suggestion being correct. I don't think the "connection reset by peer" is an actual reset from the server but a router dropping the connection. I have servers here on the same network but different subnet from the exchange servers. From other locations in the same network I can stay connected via SSH for weeks. However, from outside this network my SSH connections get killed with a "connection reset by peer" message unless I'm actively using the connection. At the same time, those connection from within the same network stay up and running. That along with the IMAP behavior I'm seeing makes me think the network admins have implemented some sort of firewall rule that's affecting both applications.
Just wanted to know what some others think before I start pushing on the network and exchange admins.
Jason ===========