Marc Sherman wrote:
> Ted Cooper wrote:
>> If you keep the "spam = user" verb/condition/thingy the same each time
>> it is called (just keeping the user the same, you can add/remove the
>> true bit), the result is cached and only calls SA once so it's no less
>> efficient than any of other condition. I use it for a yes/no condition
>> in my spam ACLs.
>
> Please reread my message that Johann was replying to:
> http://lists.exim.org/lurker/message/20080310.143243.4dbec324.en.html
>
> Caching the repeated lookups only works when Spamassassin is working.
> When Spamassassin fails (such as due to a timeout), there's no result to
> cache, so each invocation in the ACL re-runs it. And a message that
> fails with a spamassassin timeout is exactly the worst possible time to
> be rerunning it 5 times for your CPU and disk.
So I guess the answer to Johann is "yes", you can rewrite it as per the
template your provided, but isn't it nice that included some extra fun
knowledge - The more you know! I hadn't thought of SA failing inside
Exim before.
Urh. I used "There's" instead of "There are". I'm sorry I murdered the
English language.
So how does one detect if SA isn't working correctly and failing from
inside Exim? I'm already keeping a watch out for it - If it's not
running, a restart of the service is attempted and if that fails, Exim
is killed and I am notified. But if SA is still running and failing
inside Exim I have no way of knowing.
--
The Exim Manual
http://www.exim.org/docs.html
http://www.exim.org/exim-html-current/doc/html/spec_html/index.html
