You are viewing a plain text version of this content. The canonical link for it is here.
Posted to users@spamassassin.apache.org by Lorenzo Milesi <ma...@ufficyo.com> on 2015/05/25 09:42:06 UTC

Block mailing lists

hi. 

We're receiving a lot of unsolicited mail which is not spam, but I'd like block or considerable limit it. Most of those mails come from official mailing systems, like mailchimp or similar, to which I never subscribed but who probably picked the address from our website. That said common SA rules don't work with this kind of stuff, because comes from official servers and has proper signing and all.

I thought something like, for example, rising the score of mails which contains X-List-Id, but this applies only to a limited set of mailing. 
Did anyone ever made a collection of mailing list tag headers, which can be used to raise the score of such mails?
Or any better idea, rather than obfuscate or remove the info@ address from the website?

thanks
-- 
Lorenzo Milesi - lorenzo.milesi@yetopen.it

YetOpen S.r.l. - http://www.yetopen.it/

Re: Block mailing lists

Posted by Lorenzo Milesi <ma...@ufficyo.com>.
> why don't you just hit the unsubscribe link in case of mailchimp?
> if the same mailchimp-customer after that *really* imports your address
> again you can write to mailchimp-abuse and they are *really* acting

Because we're receiving so many mailing lists that it would be too cumbersome to deal with every single unsubscribe. Or at least very annoying.
Also, keeping track of what unsubscribe went successful or not would be a dedicated job. Which is not my job.

> to be honest: by naming mailchimp in that context you sound like one of
> the people not remembering where they subscribed, too lazy to
> unsubscribe and/or confusing the "spam" with the "delete" button which
> are responsible for a ton of collateral damage at Razor/Pyzor und RBL's
> every single day and the top winners of that users even forward their
> electronic bill of a local supplier as spam to their provider

I was naming that just to make it clear the mails come from mailing list provider, I have nothing against MC or anyone else.
To be honest your comment is very offensive, made to someone you have no idea who he is.

-- 
Lorenzo Milesi - lorenzo.milesi@yetopen.it

YetOpen S.r.l. - http://www.yetopen.it/

Re: Block mailing lists

Posted by Reindl Harald <h....@thelounge.net>.
Am 25.05.2015 um 09:42 schrieb Lorenzo Milesi:
> We're receiving a lot of unsolicited mail which is not spam, but I'd like block or considerable limit it. Most of those mails come from official mailing systems, like mailchimp or similar, to which I never subscribed

why don't you just hit the unsubscribe link in case of mailchimp?

if the same mailchimp-customer after that *really* imports your address 
again you can write to mailchimp-abuse and they are *really* acting

to be honest: by naming mailchimp in that context you sound like one of 
the people not remembering where they subscribed, too lazy to 
unsubscribe and/or confusing the "spam" with the "delete" button which 
are responsible for a ton of collateral damage at Razor/Pyzor und RBL's 
every single day and the top winners of that users even forward their 
electronic bill of a local supplier as spam to their provider


Re: Block mailing lists

Posted by David Jones <dj...@ena.com>.

>From: Lorenzo Milesi <ma...@ufficyo.com>
>Sent: Monday, May 25, 2015 11:16 AM
>To: users@spamassassin.apache.org
>Subject: Re: Block mailing lists

>> I have built an extensive list of safe senders in the whitelist_from_* that will
>> use the SHORTCIRCUIT (DKIM, SPF, RCVD) enabled above.

>I didn't know about this feature, I will dig more into it and see how it works. Thank you very much for your suggestion!

>But if I got it right this implies the BAYES filter has been extensively trained. Is this just to speed up scanning?
>thanks again

Train your Bayes properly like normal.  That is a separate issue from the SHORTCIRCUIT'ing recommendation.  It's not safe to whitelist/blacklist based on just email addresses since they could be spoofed or forwarded.  The safe way to whitelist/blacklist is to use other facts like SPF, DKIM, RCVD (Received header) which goes back to my point about the reputation of the sending mail server.  Once you have reliable whitelist/blacklist entries, then you can enable the SHORTCIRCUIT'ing of those rules to speed up SA's scanning based on the "priority".  Look at your rules for "priority" and "shortcircuit" to learn how they work together.  What I am suggesting is just further tuning/extending of those rules to cover more safe senders.

Re: Block mailing lists

Posted by Lorenzo Milesi <ma...@ufficyo.com>.
> I have built an extensive list of safe senders in the whitelist_from_* that will
> use the SHORTCIRCUIT (DKIM, SPF, RCVD) enabled above.

I didn't know about this feature, I will dig more into it and see how it works. Thank you very much for your suggestion! 

But if I got it right this implies the BAYES filter has been extensively trained. Is this just to speed up scanning?
thanks again 
-- 
Lorenzo Milesi - lorenzo.milesi@yetopen.it

YetOpen S.r.l. - http://www.yetopen.it/

Re: Block mailing lists

Posted by David Jones <dj...@ena.com>.
>________________________________________
>From: Reindl Harald <h....@thelounge.net>
>Sent: Monday, May 25, 2015 7:23 AM
>To: users@spamassassin.apache.org
>Subject: Re: Block mailing lists

>Am 25.05.2015 um 14:17 schrieb Lorenzo Milesi:
>>> There are inherit dangers of what you want to do, but if you're the only mail
>>> user, then you know and accept the risks, if you host mail for others, under no
>>> circumstances should you do any of this with 100% agreement from all mail
>>> users.
>>
>> Indeed I know it's not the best, but this domain is collecting so many unwanted non-spam mails that it would really take >too much effort to get rid of them, and they're continuously increasing so I don't think that just removing the email >address from the website will do any better.
>> I (well, they) just want that in their situation mailing lists are moved to the spam folder.

>sounds more like nobody trains bayes or the wrong bayes
>killing mailing-lists would also hit this list

>what you want to do is just plain wrong

Accurate spam detection should take into consideration the reputation of the sending mail server along with message content.  Mailchimp, Constant Contact, and other major mass mailing companies are responsible senders that do a good job of preventing spam coming out of their mail servers.  If they didn't, they would simply go out of business because all of us would block their messages.  They have reliable unsubscribe links and processes so I let them through to the end user so they can unsubscribe to put the control back in the end user's hands.  I block senders with invalid unsubscribe links/processes that just harvest/validate the email address.

I agree with Reindl on this one, I would not try to block these legit senders.  If you spend a little time training your bayes and whitelisting safe senders using SHORTCIRCUIT, your spam detection accuracy will become very reliable and you won't have to spend a lot of time playing whack-a-mole on new spam.  It pays off in the end and lowers your SA scanning time.

This works pretty well for me and should for most:
shortcircuit USER_IN_WHITELIST on
shortcircuit USER_IN_DEF_WHITELIST on
shortcircuit USER_IN_BLACKLIST on
shortcircuit USER_IN_DKIM_WHITELIST on
shortcircuit USER_IN_DEF_DKIM_WL on
shortcircuit USER_IN_SPF_WHITELIST on
shortcircuit USER_IN_DEF_SPF_WL on

shortcircuit RCVD_IN_MSPIKE_H5 on
shortcircuit RCVD_IN_RP_CERTIFIED on
shortcircuit RCVD_IN_RP_SAFE on
shortcircuit RCVD_IN_DNSWL_HI on
shortcircuit RCVD_IN_IADB_LISTED on

I have built an extensive list of safe senders in the whitelist_from_* that will use the SHORTCIRCUIT (DKIM, SPF, RCVD) enabled above.

P.S. Reindl usually comes across pretty harsh so don't take it personally.

Re: Block mailing lists

Posted by Reindl Harald <h....@thelounge.net>.

Am 25.05.2015 um 14:17 schrieb Lorenzo Milesi:
>> There are inherit dangers of what you want to do, but if you're the only mail
>> user, then you know and accept the risks, if you host mail for others, under no
>> circumstances should you do any of this with 100% agreement from all mail
>> users.
>
> Indeed I know it's not the best, but this domain is collecting so many unwanted non-spam mails that it would really take too much effort to get rid of them, and they're continuously increasing so I don't think that just removing the email address from the website will do any better.
> I (well, they) just want that in their situation mailing lists are moved to the spam folder.

sounds more like nobody trains bayes or the wrong bayes
killing mailing-lists would also hit this list

what you want to do is just plain wrong


Re: Block mailing lists

Posted by Lorenzo Milesi <ma...@ufficyo.com>.
> Often you will see multiple of these used in a post, so if scoring I'd suggest
> using a regex, and not use a rule per hit else it might be scored so high as to
> delete.

Yes I'd like to make a single tag which is triggered by at least one of these tags, not summing them.
I made a quick survey and collected some more of them, like X-Campaign-Id.

> There are inherit dangers of what you want to do, but if you're the only mail
> user, then you know and accept the risks, if you host mail for others, under no
> circumstances should you do any of this with 100% agreement from all mail
> users.

Indeed I know it's not the best, but this domain is collecting so many unwanted non-spam mails that it would really take too much effort to get rid of them, and they're continuously increasing so I don't think that just removing the email address from the website will do any better.
I (well, they) just want that in their situation mailing lists are moved to the spam folder.

thakns
-- 
Lorenzo Milesi - lorenzo.milesi@yetopen.it

YetOpen S.r.l. - http://www.yetopen.it/

Re: Block mailing lists

Posted by Noel Butler <no...@ausics.net>.
 

On 25/05/2015 17:42, Lorenzo Milesi wrote: 

> hi. 
> 
> We're receiving a lot of unsolicited mail which is not spam, but I'd like block or considerable limit it. Most of those mails come from official mailing systems, like mailchimp or similar, to which I never subscribed but who probably picked the address from our website. That said common SA rules don't work with this kind of stuff, because comes from official servers and has proper signing and all.
> 
> I thought something like, for example, rising the score of mails which contains X-List-Id, but this applies only to a limited set of mailing. 
> Did anyone ever made a collection of mailing list tag headers, which can be used to raise the score of such mails?
> Or any better idea, rather than obfuscate or remove the info@ address from the website?
> 
> thanks

X-list-id and X-list are older, rarely used, they are used if configured
to be, on ecartis and its predecessor listar, common official headers
are List-Id, List-post, X-BeenThere (most common with mailman). 

Often you will see multiple of these used in a post, so if scoring I'd
suggest using a regex, and not use a rule per hit else it might be
scored so high as to delete. 

There are inherit dangers of what you want to do, but if you're the only
mail user, then you know and accept the risks, if you host mail for
others, under no circumstances should you do any of this with 100%
agreement from all mail users. 

 

Re: Block mailing lists

Posted by RW <rw...@googlemail.com>.
On Mon, 25 May 2015 09:42:06 +0200 (CEST)
Lorenzo Milesi wrote:

> hi. 
> 
> We're receiving a lot of unsolicited mail which is not spam, but I'd
> like block or considerable limit it. 
> ...
> headers, which can be used to raise the score of such mails? Or any
> better idea, rather than obfuscate or remove the info@ address from
> the website?

If you're are saying that most of this stuff comes on the info@
address, then the sensible thing to do would be to treat such well-know
addresses as special cases that don't receive automated mail - if
possible rejecting with a meaningful message.