It has been pointed out that the callout timeout, e.g. as set by
verify = sender/callout=10s
applies to individual connections and commands within the callout. If a
domain has a very large number of hosts (and some do have dozens: look
at newsabuse.net) and all of them are timing out, the overall time taken
for the callout to fail can be riduculously long. This can cause trouble
further back.
I think it is clear that we need an overall timeout on the callout
process. The question is whether
(a) To invent a totally new parameter ("overall callout timeout");
OR
(b) The redefine the existing callout time to be an overall timeout as
well as a timeout on individual connections and SMTP commands.
OR
(c) To take some multiple of the existing callout time (e.g. 4 times).
The default timeout, incidentally, is 30s.
I feel that inventing a totally new parameter is making this whole thing
too complicated, but you may think differently.
Opinions?
Regards,
Philip
--
Philip Hazel University of Cambridge Computing Service,
ph10@??? Cambridge, England. Phone: +44 1223 334714.
