| 1 |
gross - Greylisting of suspicious sources |
| 2 |
|
| 3 |
INTRODUCTION |
| 4 |
|
| 5 |
Gross is a greylisting server. The features that make gross |
| 6 |
stand out from other greylisters are: |
| 7 |
|
| 8 |
- it's blazingly fast |
| 9 |
- it's amazingly resource efficient |
| 10 |
- it can be configured to query DNSBL databases, and |
| 11 |
enforce greylisting only for hosts that are found on |
| 12 |
those databases |
| 13 |
- it can block hosts that match multiple DNSBL's |
| 14 |
- it can be replicated and run parallel on two servers |
| 15 |
|
| 16 |
It supports Sun Java System Messaging Server and Postfix, |
| 17 |
Exim and Sendmail. |
| 18 |
|
| 19 |
AUTHORS |
| 20 |
|
| 21 |
Gross was originally written by Eino Tuominen <[email protected]> |
| 22 |
and Antti Siira <[email protected]> |
| 23 |
|
| 24 |
Since 2023 maintained by Dmitry Mikhirev <[email protected]> |
| 25 |
|
| 26 |
MOTIVATION |
| 27 |
|
| 28 |
We decided to write a greylister of our own because |
| 29 |
traditional greylisting was not acceptable due to it's |
| 30 |
impact on legitimate emails. Likewise, we have not used |
| 31 |
DNSBLs to directly refuse mail because of the possible |
| 32 |
false positives. One of the cornerstones of our spam |
| 33 |
fighting policy is to minimize the possibility of false |
| 34 |
positives. |
| 35 |
|
| 36 |
Because greylisting is found to be a very effective way |
| 37 |
to reduce spam, and most of spam is originated from |
| 38 |
hosts on various DNSBLs, I suggested that greylisting just |
| 39 |
those servers found on DNSBLs would reduce the amount |
| 40 |
of spam received. This software has been written to test |
| 41 |
if that hypothesis was correct. |
| 42 |
|
| 43 |
DESIGN AND OPERATION |
| 44 |
|
| 45 |
Gross consists of grossd, the greylisting daemon, and |
| 46 |
a client library for SJSMS. The server also implements |
| 47 |
Postfix's content filtering protocol. |
| 48 |
|
| 49 |
Upon receiving a request from a client, grossd first |
| 50 |
validates it. The request includes a triplet (smtp-client-ip, |
| 51 |
sender-address, recipient-address). A hash is then |
| 52 |
calculated and matched against the Bloom filter. If a |
| 53 |
match is found, server sends an OK message. |
| 54 |
|
| 55 |
If the triplet has not been seen before, grossd then |
| 56 |
queries configured DNSBL databases for the smtp-client-ip. |
| 57 |
If a positive match is found, grossd sends client a |
| 58 |
GREY response. If DNSBL tests are negative, an OK response |
| 59 |
is sent. Database is then updated; currently there are |
| 60 |
two updating styles: either to always update or to only |
| 61 |
update if the result was GREY. |
| 62 |
|
| 63 |
As mentioned above, the data store is implemented with |
| 64 |
Bloom filters. A Bloom filter is a very efficient way |
| 65 |
to store data. It's a probabilistic data structure, |
| 66 |
which means that there is a possibility of error when querying |
| 67 |
the database. False positives are possible, but false |
| 68 |
negatives are not. This means that there is a possibility |
| 69 |
that grossd will falsely give an OK response when a |
| 70 |
connection should be greylisted. By sizing the bloom |
| 71 |
filters, you can control the error possibility to meet your |
| 72 |
needs. The right bloom filter size depends on the number |
| 73 |
of entries in the database, that is, the retention time |
| 74 |
versus the number of handled connections.[1] |
| 75 |
|
| 76 |
DNS queries are done asynchronously using c-ares library |
| 77 |
(http://daniel.haxx.se/projects/c-ares/). |
| 78 |
|
| 79 |
DOCUMENTATION |
| 80 |
|
| 81 |
See INSTALL for installation instructions and for more |
| 82 |
information about configuration. |
| 83 |
|
| 84 |
See doc/examples/grossd.conf for an example configuration |
| 85 |
along with additional information about each feature. |
| 86 |
|
| 87 |
There is a mailing list linked from the project's home page. |
| 88 |
Please, don't hesitate to contact the authors directly |
| 89 |
if you have any questions! |
| 90 |
|
| 91 |
PERFORMANCE AND STABILITY |
| 92 |
|
| 93 |
Gross is currently doing greylisting for the mail gateways |
| 94 |
of the University of Turku and the University of Wisconsin. |
| 95 |
|
| 96 |
Gross has proven to be very stable on Solaris (9 and 10) |
| 97 |
on Sun Sparc. It has also been run on a production Red Hat |
| 98 |
server with minor stability issues. |
| 99 |
|
| 100 |
When we enabled Gross on our servers, we saw a tremendous |
| 101 |
drop on our internet originated mail flow. Also, |
| 102 |
the load on our spam filtering servers was cut in half! |
| 103 |
|
| 104 |
DEVELOPMENT AND LICENSING |
| 105 |
|
| 106 |
Gross is published on a BSD-like license. |
| 107 |
|
| 108 |
See https://codeberg.org/bizdelnick/gross for more information. |
| 109 |
|
| 110 |
If you have any suggestions regarding the software or wish |
| 111 |
to take part on the development, feel free to contact the authors. |
| 112 |
We'll be happy to hear if you dare to try out the software. |
| 113 |
|
| 114 |
Should it be grey or gray?-) |
| 115 |
|
| 116 |
Eino Tuominen |
| 117 |
|
| 118 |
|
| 119 |
[1] http://en.wikipedia.org/wiki/Bloom_filter |