#haraka

/

      • ultimatt2
        On that note, I think we should stick with grunt.
      • :)
      • oh, and in case you missed yesterdays xkcd: http://xkcd.com/1520/
      • Since Steve is here, I won't post a link to todays xkcd.
      • baudehlo
        I enjoyed today's xkcd :)
      • kpsullivan has quit
      • _smf_
        baudehlo: a quick yay or nay on #951 would be appreciated when you have a minute.
      • donspaulding joined the channel
      • baudehlo
        Given the length of the discussion, I doubt a quick yay or nay will be possible :)
      • _smf_
        baudehlo: honestly - you can skip the discussion entirely; it's mostly irrelevant to the actual proposed change except for Matt's initial comment.
      • ultimatt2
        FTR, I don't object to it.
      • So long as it's not being proposed *instead* of having a connect_pre hook (which it's not, AFAICT), then this helps scratch one of Steve's itches
      • it adds a little extra control to a presently difficult problem
      • baudehlo
        Does it have docs? :)
      • ultimatt2
        A part of me wishes that instead of serial, each hook was run with async.parallel, and then a final callback decided what to do
      • baudehlo
        I don't :)
      • Last thing I'd want is uribl running in parallel with spamassassin.
      • ultimatt2
        that already happens, just not to the same message at the same time
      • baudehlo
        that's the point
      • I want to reject for uribl before SA runs.
      • _smf_
        Yeah - concurrent hooks would be a nightmare.
      • baudehlo
        I can see that ultimatt2 would want them for karma.
      • but I don't use karma.
      • _smf_
        Me neither.
      • But I'm getting close to looking at it - although I'll probably rip a lot of it out and just implement it as a generic way to score stuff that I don't want to reject outright.
      • e.g. something like 'host was greylisted' AND 'has an attachment' then 'reject'
      • And I couldn't do that if things were concurrent.
      • Just checked - and I've got a whole bunch of stuff that wouldn't work or code that I'd have to duplicate if hooks were run in parallel. example: my MX lookup code runs and sets a whole bunch of notes for other plugins that run on the same hook then use.
      • I wouldn't want to have to run the MX lookup code in all of those plugins - it would be really inefficient which is why I did it like that in the first place.
      • ampzamp joined the channel
      • ampzamp has quit
      • ampzamp joined the channel
      • teknix__ joined the channel
      • teknix_ has quit
      • teknix_ joined the channel
      • teknix__ has quit
      • ampzamp has quit
      • ultimatt2 has quit
      • thezeroth has quit
      • thezeroth joined the channel
      • ultimatt2 joined the channel
      • thezeroth has quit
      • thezeroth joined the channel
      • erick2014 joined the channel
      • erick2014
        I'm getting a message on one of the plugins I'm running stating "delaying response for 1 seconds". It's an SMTP Auth plugin which inherits auth/auth_base'. My code doesn't include any message like this. Is this something built into Haraka that I'm missing?
      • teknix_ has quit
      • nevermind - figured it out
      • erick2014 has quit
      • jbraun joined the channel
      • jbraun has quit
      • EyePulp has quit
      • PixelPaul joined the channel
      • PixelPaul
        has anyone stress tested haraka with the queued messages?
      • just wondering if it could handle 1,000,000 + messages queued at once
      • or would be better to send it to postfix to be queued