Gentoo Archives: gentoo-dev

From: Niels Dettenbach <nd@××××××××.com>
To: gentoo-dev@l.g.o
Subject: Re: [gentoo-dev] Anti-spam changes: proposal to drop spammy mail
Date: Mon, 11 May 2015 10:09:25
Message-Id: 1626925.WQM6IekEy6@gongo
In Reply to: [gentoo-dev] Anti-spam changes: proposal to drop spammy mail by "Robin H. Johnson"
1 Am Montag, 11. Mai 2015, 04:26:01 schrieb Robin H. Johnson:
2 > This was a good early policy, as Gentoo was a much more reliable host than
3 > email providers a decade ago. This isn't true anymore, with the meteoric
4 > rise
5 > and success of gmail.
6 This is not true at all - but email service "reliability" was and is not a
7 primarily question of the hosts OS nor some kjind of basic standard
8 configuration of a mailer "package" (or ebuild).
9
10
11 > As past long-standing practice, @Gentoo.org system-level mail handling for
12 > incoming mail was officially to tag everything, and delete nothing.
13 This is - for a public internet Mailer / MX - a VERY bad option - at least
14 mail not fulfilling basic email standards should be blocked (as usual by the
15 very most professional level mail services), because it could be (used)
16 abusive by thirds.
17
18
19 > A LOT of developers forward their mail now, to systems that
20 > refuse/temporarily blacklist the forwarding system because there is a lot
21 > of spam. Gmail is particularly strict in this regard, throttling mail to
22 > any recipient from the forwarding source.
23 Gmail is crap (from my opionion) - at least for really professional email
24 users or at least users which need a reliable email service (includes the
25 possibility to recieve or send out higher frequent emails or emails from more
26 "exotic" sources).
27
28 We - as a email provider - recognized the development of the rate limiting
29 policy by Google as well and it seems that Google is adapting that limits by
30 the amount of mail which is send from google users into that "domains" (as far
31 as this is correctly locatable for them, because by specs it is not
32 really...). This works OK for source mail servers / domains which still have
33 "typical" email users and their regular traffic (or what googles thinks about
34 - i.e. what some kind of user deletes without reading or is (i.e. false)
35 marking as "spam") going to different recipients at google too, but not very
36 well for more specialized email systems (like mailing list weighted mail
37 systems / MTAs or systems handling some kind of notifications for a larger
38 amount of users / customers - (widely) independent from what type of mail they
39 transport and how high the part amount of spam within is).
40
41 For mail ISPs there is a contact where they can "complain" or "discuss" about
42 limits with a mail server, but i did not know any case wher any answer was
43 coming from them (does not mean that google did not recognizes it).
44
45 But if someone decied for google as his primarily email provider - he has to
46 live with that policies from google, which might work OK for most peoples and
47 their "most peoples traffic".
48
49
50 > Unless there are any major objections, as of May 17th, Infra will start
51 > dropping mail that scores more than 10.0 points in Spamassassin.
52 >
53 > If that is successful, I propose to drop the score point by 1 point every
54 > month until it hits a score of 5.0 (so by mid-October, it will be dropping
55 > mail that scores more than 5.0).
56 This will work (depending form some of your SA setup details and how far you
57 use all of the features, channels and possible extensions / third party
58 services - i.e. DCC, Razor, Pyzor, "all" the different update channels, Bayes
59 - while disabling DNSBLs and doing that still before in your mailer) until you
60 go down 5.
61
62 On 6 you typically will still get a lot of spam through (enough to make users
63 work to sort it out regularly) while on 5 you definitely will loose ham - even
64 if you still use the "usual" DNS based spam blocking lists (would let do this
65 the mailer before SA usually because of performance/ressource reasons).
66
67 So typical "working" values are between 5 and 6.
68
69 But even then you will get spam, as long if you won't loose any ham "from time
70 to time" - installing a filter solution like SA alone is just one part of a
71 modern and working anti spam gateway with a zero false positive target policy
72 (but might be enough to help your moderators here ß).
73
74 Most important tasks are to set up the MTA/MDA far enough to block email which
75 are not fulfilling the specs and/or coming from machines which doesnt it - the
76 very most of spam should be catched here usually, but is is a dynamic tasks to
77 know and/or find out which parts of the standards are important and which
78 less, because there are still many used mail systems out which are not
79 configured properly, but have a significant user base.
80
81 But there are very good reasons why more and more admins of smaller otherwise
82 focussed admins decide to route their email traffic over a gateway mail
83 service. Setting up a "just working" email system is a simple job today, but
84 running a reliable internet email service on a more professional level is a
85 real job which takes time and knowledge in maintaining that service - far over
86 the setup details of a SMTP daemon (i.e. a proper DNS setup with PTR, DKIM,
87 SPF, abuse contacts and so on) which often are out of scope of a typical
88 admin.
89
90
91
92 hth a bit.
93 cheerioh,
94
95
96 Niels.
97
98
99
100 --
101 ---
102 Niels Dettenbach
103 Syndicat IT & Internet
104 http://www.syndicat.com
105 PGP: https://syndicat.com/pub_key.asc
106 ---

Replies

Subject Author
Re: [gentoo-dev] Anti-spam changes: proposal to drop spammy mail "Robin H. Johnson" <robbat2@g.o>