Prev: Failover control of Postfix content-filter
Next: GSSAPI Authentication
From: Mark Goodge on 15 Dec 2009 11:48
Wietse Venema wrote:
> Mark Goodge:
>>
>> The cache has two main functions:
>>
>> 1. To reduce traffic between the gateway and destination server, and
>> reduce load on both.
>>
>> 2. To allow the gateway server to correctly handle mail
>> acceptance/rejection when the destination server is unreachable.
>
> If the cache is good enough for 2) then it makes no sense to
> skip the cache for 1).

It would if even the short lag for cached failures (or, for that matter,
cached successes) to expire is unacceptable. A typical example might be
a webmail system which allows users to change or create addresses
whenever they want to, and have those changes instantly reflected in
what mail is accepted and rejected.

Mark

From: Wietse Venema on 15 Dec 2009 11:58
Mark Goodge:
> Wietse Venema wrote:
> > Mark Goodge:
> >>
> >> The cache has two main functions:
> >>
> >> 1. To reduce traffic between the gateway and destination server, and
> >> reduce load on both.
> >>
> >> 2. To allow the gateway server to correctly handle mail
> >> acceptance/rejection when the destination server is unreachable.
> >
> > If the cache is good enough for 2) then it makes no sense to
> > skip the cache for 1).
>
> It would if even the short lag for cached failures (or, for that matter,
> cached successes) to expire is unacceptable. A typical example might be

If the cache is NOT good enough for 1) then it is NOT good enough
for 2), either. Therefore do NOT use the cache.

Wietse

From: Mark Goodge on 15 Dec 2009 12:08
Wietse Venema wrote:
> Mark Goodge:
>> Wietse Venema wrote:
>>> Mark Goodge:
>>>> The cache has two main functions:
>>>>
>>>> 1. To reduce traffic between the gateway and destination server, and
>>>> reduce load on both.
>>>>
>>>> 2. To allow the gateway server to correctly handle mail
>>>> acceptance/rejection when the destination server is unreachable.
>>> If the cache is good enough for 2) then it makes no sense to
>>> skip the cache for 1).
>> It would if even the short lag for cached failures (or, for that matter,
>> cached successes) to expire is unacceptable. A typical example might be
>
> If the cache is NOT good enough for 1) then it is NOT good enough
> for 2), either.

It would be under the scenario that I described.

> Therefore do NOT use the cache.

That's the workaround, yes. But it's not the ideal!

Mark

First  |  Prev  | 
Pages: 1 2 3 4
Prev: Failover control of Postfix content-filter
Next: GSSAPI Authentication