Not great. Really, it should be up to the client to identify and adapt
to a NAT, not the server. After all, the client can 'done' an idle
connection any time it wants.
Is there a way to tune or configure this behaviour at compile time in
uw-imap? My server is port-mapped rather than nat'd so shouldn't
suffer the problem of nat timeouts. Not sure about my phone but I can
at least experiment :)
Robert
p.s. I did look though the archives but couldn't find anything. Maybe
I just didn't look back far enough - thanks for finding the post for
me :)
On 14 Jul 2008, at 15:05, Paul Roub wrote:
> Any ideas why there is this 2min packet exchange?
>
From the archives:
http://mailman1.u.washington.edu/pipermail/imap-uw/2008-March/001965.html
"With UW imapd, every 2 minutes the client should receive a
* OK Timeout in 28 minutes
message (this will count down). This is an attempt to work around
the NAT
box problem (not to mention effectively defeating the entire point of
IDLE) by keeping the session active enough to avoid the loss of NAT
mapping. "
...
"What this all means is that IDLE is completely pointless. UW imapd
implements IDLE only as a checklist item, but because of the bugs in
NAT
and in Outlook is compelled to implement IDLE in a way that no
better (and
arguably worse) than client based polling. "
---
GnuPG public key:
http://www.Far-Blue.co.uk
_______________________________________________
Imap-uw mailing list
[email protected]
https://mailman1.u.washington.edu/mailman/listinfo/imap-uw
