reverse dns configuration for IPV4, IPV6+ dns+ mail ?

[Reindl Harald]

Am 19.06.2017 um 15:00 schrieb Matus UHLAR - fantomas:
>>> On 19.06.17 01:05, Reindl Harald wrote:
>>>> it's nearly always misleading and results in randomness on the 
>>>> receiving server which name get logged and if A/PTR matches
>>>>
>>>> normally you should always have:
>>>>
>>>> * IP with *one* PTR
>>>> * the A-Record for the PTR matches
> 
> these two are correct.
> 
>>>> * smtp_helo_name of your MTA matches the same name
> 
> this one is incorrect and my next comment applies only to this one:

does it harm? NO
is it easy to achive? YES
can it be used for scoring on a spamfilter? YES

>> anyways, with 2 PTR records for the same IP on servers with 
>> http://www.postfix.org/postconf.5.html#reject_unknown_client_hostname 
>> you play lottery because one time it's logged as unknown and the other 
>> time as matching, the unknown cases would trigger 
>> reject_unknown_client_hostname
> 
> Actually, this would only happen when one of the A/AAAA records didn't 
> exist.
> Having two PTR records with valid A/AAAA would only confuse people because
> they could see different one each time client connects, but doesn't break
> anything (only dns-based acl's)

this NOT true for all cases

FRANKLY i have seen enough *real world* postfix rejects caused by 
"check_reverse_client_hostname_access" because the idot on the other 
side had "mail.example.com" AND the old 
"my-provider-xx.xx.xx.xx-dyn.crap" PTR where one time 
"check_reverse_client_hostname_access" was fine because it dealed with 
the "mail.example.com" and the next mail was rejected by match 
"my-provider-xx.xx.xx.xx-dyn.crap"

in all of these cases just remove the old useless generic PTR would have 
solved the problem from the start

so please inform yourself and do tests.....