Re: Nice easy Spamassassin question - solved

On 16/09/2025 09:40, Frank Leonhardt wrote:
> On 15/09/2025 22:37, Lexi Winter wrote:
>> Frank Leonhardt:
>>> Apparently someone's decided to drop sendmail from the base system in 15.
>>   
>> this is not true, sendmail is still shipped in base in 15.0 and there
>> are no plans to remove it in any future release.  in fact, i just
>> committed a couple of sendmail fixes in the last few days.
>>
>> please tell whoever told you this that they are misinformed.
>
> SirDice, 11th July, 2024:
>
> https://forums.freebsd.org/threads/upgrading-to-release-14-from-13.91161/
>
> "And if you are actively usingsendmail(8) 
> <https://man.freebsd.org/cgi/man.cgi?query=sendmail&sektion=8&manpath=freebsd-release-ports>I 
> suggest switching to_mail/sendmail_ 
> <http://www.freshports.org/mail/sendmail>instead. In 15 sendmail will 
> be completely removed from the base."
>
> If this is not the case then I'm pleased to hear it, but after the 
> Berkeley Internet Name Domain daemon (and most importantly tools) was 
> removed from BSD; and /etc/motd was replaced by something 
> incompatible; and DMA has been the default since 14.0, it sounded 
> quite plausible.SirDice is very well informed so I assume it was 
> correct a year ago, and movements such as this tend to keep going 
> until they get their way. Do you know what's going on in the background?
>
>>> But spamassassin is causing me trouble.
>> sorry, no idea about that, but i wanted to correct the above.
> Yeah, it should be a fairly straightforward thing but the 
> documentation is lacking and I'm trying to do something about that. If 
> you're running a an actual mail server you're going want 
> sendmail+SASL+spamassassin+DKIM, and these days in a jail, and it's a 
> bit of a pain. The handbook now covers SASL, but only the ports 
> version and requires the base sendmail to be recompiled. (Cue Exim and 
> Postfix aficionados, who may have a point).
>
> Regards, Frank.
>
>
This was a problem due to named pipes appearing but not working inside a 
jail. I can probably get around it using a different tweaked devfs 
ruleset, but I opted to have SA components communicate through ports 
instead, which had its own interesting problems.

So it does now work. I'll document this stuff on my blog later.