Could someone enlighten me how do spamd and rspamd compare? AFAIKnew, Rspamd is the fancy new thing, and I see spamd is similarly aged, but I actually haven’t heard about spamd before.
I’m kind of freshly into mailservers, doing some reading to set up or create my own mailserver (https://twitter.com/feilermichal/status/999724341611909120 — I tweeted this (“I really want to write my own mail server for myself.”) on May 24, before hearing about that effort (I only learned about it today from @romanzolotarev’s tweet))
OpenBSD spamd is dead simple, it just whitelists servers that are good actors. It’s not like rspamd at all: rspamd is more like spamassassin, providing a suite of different spam detection strategies along with some end user tooling.
OpenBSD spamd literally just rejects mail from a new IP for a period of time, and then whitelists the server if it properly handles delivery retry, assuming that spam servers won’t do that. The technique is called greylisting. It’s a fairly legitimate assumption, but it’s problematic for small deployments dealing with mainstream mail providers. How many gmails will spamd delay before it “learns” all the gmail IPs?
It’s behavior is a tiny bit more sophisticated than that, but that’s the basic idea. It doesn’t provide anything like using external whitelists or blacklists, checking SPF or DKIM, statistical analyzers, or any of the normal features you might expect from a spam filter. It’s very much an 80/20 tool. Also, rspamd implements greylisting, so you don’t need spamd if you use rspamd.
IME “learning” all the gmail IPs doesn’t take overly long. Just a minor annoyance when the server is new.
Outlook360, though, does retries from a different IP every time, so greylisting basically blocks email from there (easy to special case in config) and first email from a new service can be a bit slow when making new accounts.
Could someone enlighten me how do spamd and rspamd compare? AFAIKnew, Rspamd is the fancy new thing, and I see spamd is similarly aged, but I actually haven’t heard about spamd before.
I’m kind of freshly into mailservers, doing some reading to set up or create my own mailserver (https://twitter.com/feilermichal/status/999724341611909120 — I tweeted this (“I really want to write my own mail server for myself.”) on May 24, before hearing about that effort (I only learned about it today from @romanzolotarev’s tweet))
OpenBSD spamd is dead simple, it just whitelists servers that are good actors. It’s not like rspamd at all: rspamd is more like spamassassin, providing a suite of different spam detection strategies along with some end user tooling.
OpenBSD spamd literally just rejects mail from a new IP for a period of time, and then whitelists the server if it properly handles delivery retry, assuming that spam servers won’t do that. The technique is called greylisting. It’s a fairly legitimate assumption, but it’s problematic for small deployments dealing with mainstream mail providers. How many gmails will spamd delay before it “learns” all the gmail IPs?
It’s behavior is a tiny bit more sophisticated than that, but that’s the basic idea. It doesn’t provide anything like using external whitelists or blacklists, checking SPF or DKIM, statistical analyzers, or any of the normal features you might expect from a spam filter. It’s very much an 80/20 tool. Also, rspamd implements greylisting, so you don’t need spamd if you use rspamd.
IME “learning” all the gmail IPs doesn’t take overly long. Just a minor annoyance when the server is new.
Outlook360, though, does retries from a different IP every time, so greylisting basically blocks email from there (easy to special case in config) and first email from a new service can be a bit slow when making new accounts.
Woah, I hope there will be a possibility to use rspamd with this new project.
How to turn a simple idea into a revolution.
I wish more people were able to brainstorm with such will to listen objections!
Next step (Part 1: OpenSMTPD) : https://github.com/openbsdjumpstart/mailserver/issues/3