Re: closing all inactive server connections

[Ted Zlatanov <tzz@lifelogs.com>]

On Tue, 31 Jul 2007 17:52:42 +0900 Katsumi Yamaoka <yamaoka@jpl.org> wrote: 

>>>>>> Ted Zlatanov wrote:
KY> --- nntp.el~	2007-07-18 12:07:39 +0000
KY> +++ nntp.el	2007-07-31 03:04:52 +0000

>> I tried it and the IMAP server still hangs.  Maybe the fix is in
>> nnimap.el?  I haven't looked deeper into this issue but I'll be glad to
>> debug things on my side.

KY> Is the patch effective for the hanged nntp connections?

My NNTP connections are OK.

KY> Well, I tried looking into the IMAP code although I've never
KY> used nnimap.  And I found the functions that communicate with
KY> the server all have the following form:
KY>   (while (and (memq (process-status PROCESS) '(open run))
KY> 	      (wait-for-a-string-that-the-PROCESS-returns))
KY>     (accept-process-output PROCESS 1)
KY>     (sit-for 1))

KY> This can be an infinite loop if the process-status keeps `open'
KY> or `run' even though the connection has actually died.  In
KY> particular, the `imap-wait-for-tag' function can be the cause of
KY> the hanging since it sends the LOGOUT command to the dead server
KY> and waits for the response.  So, I tried modifying it so as to
KY> quit in a certain time.  The patch to imap.el is below.  Anyway,
KY> it needs to be reviewed by someone who is skilled in IMAP.  The
KY> default value of `imap-timeout-seconds' might be too small for
KY> slow connections.

I tried this and it works fine.  If Simon is around, he can take a look,
but I tested it over 6 hours of IMAP/NNTP use and had no problems.  If
anyone else can verify that it works, this would be a great addition to
Gnus.

Maybe it should be generalized, however, to an IMAP timeout per (server,
group name, command) combination (as a regex or a function), so you can
assign "LOGOUT" a longer timeout with a slower server.  This could also
be a server tuning parameter for nnimap instead of an imap.el parameter.

Thank you for the help!
Ted