>>>The above is caught by:
>>>
>>>/[0-9]{1,3}(\.|\-)[0-9]{1,3}(\.|\-)[0-9]{1,3}(\.|\-)[0-9]{1,3}.*\[/
>>>client_filter.class
>>remove the \[ part.
>>and BTW, here is a shorter version:
>>/(\d{1,3}[-\.]){3}\d{1,3}/ class_client_filter
>>
>>
>>>... but is not going to the restriction class as actual SMTP client.
>
>After looking at my and mouss' advice, I realize the expressions we suggested
>will match _all_ clients since it matchs an IP address.
I've wanted to match patterns in the PTR domain name, that's why I anchored the
expr with \[, to make sure I didn't match the domain.name[ip.ad.re.ss]
>Probably not what you intend.
of course not
>Here's what I use to match IP-containing hostnames, should work for you.
>/(\d{1,3}[-\.]){3}\d{1,3}[^.]+\.[^.]/ class_client_filter
>ie. there must be a dot somewhere after the IP-containing section.
OK, same idea as mine, but I say "there must be a [ after the PTR domain name",
because without the post-PTR-domain-name anchor, it was matching/shunting all
connections into the (suspect) client_filter.class instead of staying in the
non_suspect (main) branch.
I'm really struggling with this because without your domain.name\. or my \[,
not all domain.name[a.b.c.d] get matched into the client_filter.class. iow,
with the trailing anchor,
So what string does the restriction class matching look at?
1. label.domain.tld[ip.ad.re.ss]
2. label.domain.tld ip.ad.re.ss
3. label.domain.tld
4. ip.ad.re.ss
btw, the policy service passes the client_name and client_address as two
separate values.
Len
