home.social

#spamassassin — Public Fediverse posts

Live and recent posts from across the Fediverse tagged #spamassassin, aggregated by home.social.

fetched live
  1. 🚀 Deploy #Virtualmin on #AlmaLinux #VPS (5 Minute Quick-Start Guide)

    This article provides a how-to guide to deploy Virtualmin on AlmaLinux VPS.
    What is Virtualmin?
    Virtualmin is a powerful open-source web hosting control panel built on top of Webmin.

    It supports:

    Multi-domain virtual hosting
    Apache or Nginx
    PHP ...
    Continued 👉 blog.radwebhosting.com/deploy- #selfhosting #bind #controlpanel #postfix #mariadb #certbot #spamassassin #clamav #letsencrypt #phpfpm #dovecot #selfhosted

  2. 🚀 Deploy #Virtualmin on #AlmaLinux #VPS (5 Minute Quick-Start Guide)

    This article provides a how-to guide to deploy Virtualmin on AlmaLinux VPS.
    What is Virtualmin?
    Virtualmin is a powerful open-source web hosting control panel built on top of Webmin.

    It supports:

    Multi-domain virtual hosting
    Apache or Nginx
    PHP ...
    Continued 👉 blog.radwebhosting.com/deploy- #letsencrypt #clamav #selfhosting #mariadb #certbot #dovecot #controlpanel #selfhosted #postfix #spamassassin #bind #phpfpm

  3. 🚀 Deploy #Virtualmin on #AlmaLinux #VPS (5 Minute Quick-Start Guide)

    This article provides a how-to guide to deploy Virtualmin on AlmaLinux VPS.
    What is Virtualmin?
    Virtualmin is a powerful open-source web hosting control panel built on top of Webmin.

    It supports:

    Multi-domain virtual hosting
    Apache or Nginx
    PHP ...
    Continued 👉 blog.radwebhosting.com/deploy- #controlpanel #phpfpm #letsencrypt #selfhosted #mariadb #dovecot #spamassassin #selfhosting #bind #postfix #clamav #certbot

  4. Бесплатная защита от спама на почте с помощью ИИ фильтрации без VPN: многоуровневый метод с BERT и 550 МБ RAM

    Вас тоже достаёт спам и реклама? Рекламу я блокирую через свой DNS сервер и локальными CSS фильтрами, а вот для почты пришлось придумать что-то другое. Схема многоступенчатая. Первыми срабатывают фильтры Яндекса и Mail — что-то они отсеивают сами, ещё до пересылки. То, что прошло через них, падает на мой сервер, где стоит SpamAssassin. Ловит ещё часть. Но после двух уровней всё равно что-то просачивается, спамеры же не сидят без дела. И вот этот остаток доезжает до Gmail и что-то оседает в папке Спам, а что-то попадает во входящие и приходит раздражающее уведомление. Хотелось, чтобы со временем не накапливался мусор в папках, который надо разгребать вручную. Особенно важно заблокировать то, что не является полностью спамом: приглашения на конференции, партнёрские предложения, кредиты — формально не нарушение, поэтому байесовский фильтр такие вещи плохо ловит. Локальная BERT-модель закрыла обе проблемы. Взял ruBert-base-antispam с HuggingFace — файн-тюн на базе DeepPavlov/rubert-base-cased-conversational. 177 миллионов параметров, 12 слоёв трансформера, 768 hidden size. Физически не принимает больше 512 токенов на вход. В памяти занимает около 550 МБ, ответ приходит за 100-200 миллисекунд. Бинарный классификатор — текст на входе, 0 или 1 на выходе, никаких промптов и reasoning. Идеально!

    habr.com/ru/articles/1055518/

    #спам #email #BERT #AI #exim #dovecot #spamassassin #фильтрация #python #почта

  5. 🚀 Deploy #Virtualmin on #AlmaLinux #VPS (5 Minute Quick-Start Guide)

    This article provides a how-to guide to deploy Virtualmin on AlmaLinux VPS.
    What is Virtualmin?
    Virtualmin is a powerful open-source web hosting control panel built on top of Webmin.

    It supports:

    Multi-domain virtual hosting
    Apache or Nginx
    PHP ...
    Continued 👉 blog.radwebhosting.com/deploy- #dovecot #clamav #selfhosting #postfix #certbot #selfhosted #letsencrypt #phpfpm #mariadb #bind #spamassassin #controlpanel

  6. Yep. It's a SpamAssassin/Email/character set thing. Because character sets aren't reliable then the default behaviour is to leave it as-is and do byte matching.

    stackoverflow.com/questions/49

    Edit: Except that the docs say the default in 4.x is to enable the conversion. So apparently I'm just dealing with matching UTF-8 bytes rather than character codes 😑

    #SpamAssassin #Regex

  7. I can't even get `[\x{0400}-\x{058F}]` to match (Cyrillic and Armenian characters) 🤨

    #SpamAssassin #Regex #WTF

  8. As a member of the #ASF #SpamAssassin Project Management Committee, I have an important announcement…

    Today, in honor of Julius Caesar and Pope Gregory and fools across Europe, we will join the Foundation in a special celebration: we will do nothing. No joke rules added to block MS365 for being a threat to the email ecosystem. No new rules marking all Google Mail spam. No flamewars on spamassassin-users over the precise definition of spam.

    The RuleQA server continues to toil along unthanked.

  9. I need help understanding #SpamAssassin
    I have a mail here that got scored with RCVD_IN_VALIDITY_CERTIFIED_BLOCKED=0.408

    Looking at github.com/apache/spamassassin this seems to mean that Validity told me to fuck off (fair, I don't pay them), but why does that increase the score of the mail?
    The score matches github.com/apache/spamassassin and I see no score for RCVD_IN_VALIDITY_CERTIFIED there.
    There is a 0 score in github.com/apache/spamassassin which implies that check should be disabled anyway?

  10. I guess it's time for me to finally get around to setting up #SpamAssassin on my #Email server. A while ago I had set up auto rejecting of emails that didn't meet basic levels of authentication and grey listing and that seemed to take care of any #Spam I had been getting for a bit, but now I'm getting a handful a day that are making it through. Seems the more sophisticated spammers have discovered me.

  11. I'm getting the same spam over and over to a self hosted account. The subject is always the same, the body is pretty short and nearly identical. The only thing that changes are the sending address and the parameters on a long URL at the end of the email.

    No matter how many I pass to spamassassin it still never scores more than 3.7

    Any ideas?

    Subject: "Let's start getting to know each other."

    #email #spam #spamassassin

  12. Toiling in the #SpamAssassin dungeons today...
    It's ridiculous how many bots are still hitting our RuleQA, mostly with referrer-free “detail” URLs, which I made trivial to answer (403) many months ago. Not enough to be a problem, just about 1k-hits per hour. Spread all around the net: latest 1000 hits came from 967 different IPs with 2 IPs accounting for all ofthe repeat visitors. 959 different /24s, 958 different /16s.
    Asking for random info from many different years.
    #InfoSec

  13. If you’re in LACNIC-allocated space and can’t get to the #SpamAssassin infra, blame your immediate network neighbors. The DDoS’er who has been trying to destroy our RuleQA seems to be preferring that space. I've been blocking mostly by /22 as that's how much of the space is split up.
    I know it's whack-a-mole, but I have no better approach. The attack is blended: SYN flood with those that get answered asking for high-cost URLs.
    #InfoSec

  14. I'm still at ~20% of TLSv1.3 for SMTP connections to deliver me mail with ~80% still being TLSv1.2.

    However, for non-spam mail delivery it's 92% TLSv1.3 and 8% TLSv1.2.

    Perhaps disabling TLSv1.2 is a good replacement of #rspamd or #spamassassin ? :)

  15. It’s a rhetorical Q.
    It’s almost a miracle that it was found.

    It’s also why I do a paranoid level of consistency checks on every #SpamAssassin release. Our definitive repo is in Apache SVN, so we don’t have precisely the same vulnerability as libxz had, but I still verify that no matter how one gets the source, it is identical to what we've checked in.

    #FOSS #Sysadminnery #InfoSec @mjg59 nondeterministic.computer/@mjg

  16. Someone out there really wants to down the #ASF #SpamAssassin RuleQA web server.
    Generating almost a GB/min of log.
    Also generating a LOT of blockage.
    #DigitalOcean and #Huawei customers are going to have a hard time using SA. Damn shame.

  17. Spam wird immer „besser“.

    Ich hätte da fast draufgeklickt, aber der gute #SpamAssassin schrie seine Warnungen aus und er hatte recht.

    Das Logo ist etwas falsch, aber ansonsten sah es für mich plausibel aus.

  18. I have an actual idea (ASF oak leaf, rotated and repeated into a shuriken) for a new #SpamAssassin logo, but no time to knock the rust off of my graphics skills and do it.
    This is where generative graphic tools could make sense, were they not toxic for the planet and built on theft. I am controlling the urge.