#haraka

/

      • lnb_ joined the channel
      • dopesong joined the channel
      • lnb has quit
      • _smf_
        lnb: are you adding banners or anything like that?
      • lnb__ joined the channel
      • lnb_ has quit
      • lnb__ is now known as lnb
      • lnb has quit
      • lnb joined the channel
      • baudehlo1 joined the channel
      • baudehlo has quit
      • dopesong_ joined the channel
      • baudehlo1 is now known as baudehlo
      • ultimatt1
        _smf_, am I misunderstanding you on https://github.com/baudehlo/Haraka/pull/906 ?
      • _smf_
        ultimatt1: no - I don't think so. It's fine if you intended it to be like that, but if you're ever going to put the results into a custom Web UI instead of via Kibana; putting all the results into the root object isn't going to work that well as the front-end then requires knowledge of each plugin to be able to display anything.
      • DragonPunch joined the channel
      • ultimatt1
        "if you're ever going to put the results into a custom Web UI" <--- yes, I certainly will
      • _smf_
        Then aren't you making it difficult on yourself?
      • ultimatt1
        I can't see how, as I won't be presenting the data willy nilly, but in some carefully constructed format
      • I haven't done it yet, so I'm obviously ignorant here
      • _smf_
        If you have a 'plugin_results' container object e.g. { plugin_results: { p0f: { pass: ... fail: .... skip ...}, dnsbl: { pass: .... fail: .... skip: ....} } ....
      • ultimatt1
        right...
      • _smf_
        You can then simply do a 'walk' of the keys and display it nicely in a web interface.
      • As the results all have a standard format.
      • ultimatt1
        why can't I already do that?
      • _smf_
        You can then add extra bits for some of the plugins if you need.
      • Becuase you've put the plugin results in the root.
      • ultimatt1
        I can easily detect objects with "result_store" data in them
      • _smf_
        So you can't 'just' walk the keys.
      • Yeah - but that's a bit crap though isn't it?
      • ultimatt1
        uhhh, if you say so...
      • _smf_
        You're going to have to decend into each key and then check if it's the correct type.
      • If you put it in a results object; you don't have to do that. You save yourself a bit of code and the actual returned JSON looks a bit tidier IMO.
      • ultimatt1
        Is there a way to change how fields are presented in Kibana?
      • _smf_
        I've no idea; I don't use Kibana
      • I rolled my own UI as Kibana's authentication was totally useless for anything that is exposed on the internet.
      • ultimatt1
        true that
      • but for admins
      • _smf_
        (unless you pay $$$ for the module that deals with permissions)
      • ultimatt1
        it's fantastic
      • _smf_
        Yeah - that was no good for what I was writing.
      • I needed that functionality without having to expose ES to the internet.
      • ultimatt1
        I need subsets of those features w/o ES exposure
      • but I'll have extra meta data in the results that will be used to filter user "scope"
      • and constrain what each user is able to see
      • _smf_
        As Kibana 3 required direct access to ES via TCP; and that was puke to do. No idea if that has changed.
      • Yeah - I've already dealt with that in my web app.
      • ultimatt1
        Dunno, I'm using Kibana 4, which is a node.js app
      • it does require access to ES though
      • Kibana will never be exposed to the internet in our use
      • EyePulp joined the channel
      • lnb has quit
      • I think the general use case of log.elasticsearch should be targeted at kibana users
      • _smf_
        Ok
      • ultimatt1
        and in that case, not having plugin data in an object is a win
      • but, we could certainly have an option that shoves PI results into a 'plugin' container
      • for use cases where that helps (such as: I'm building my own web app)
      • that's something I might add a few months down the road. I don't want/need it for my personal Haraka server, but I might just decide you're right when I get around to consuming that data in the $work web app
      • _smf_
        Ok - np; it's was just a comment as it's something I ran into that I thought might help. No biggie.
      • ultimatt1
        If you hadn't mentioned, it, I'd just have done a bunch of "typeof ..." checks. By mentioning this, I'll give it a 2nd thought in the future. :)
      • howitdo joined the channel
      • Currerius joined the channel
      • DragonPunch
        Mail From: and Mail To: do they take multiple inputs? Or just one input?
      • Mail From & Rcpt To, sorry.
      • ultimatt1
        MAIL FROM is singular
      • RCPT TO is multiple
      • more specifically, RCPT TO is 1+
      • lnb joined the channel
      • DragonPunch
        ultimatt1: In the headers from "To" would they then show up as singular or plural?
      • ultimatt1
        different format
      • DragonPunch
        ultimatt1: To one person and CC the others?
      • ultimatt1
      • DragonPunch
        ultimatt1: What page?
      • _smf_
        DragonPunch: see section 3.4
      • ultimatt1
        and 3.6.3. Destination Address Fields
      • and 3.6.2. Originator Fields
      • _smf_
        3.4 is the field format though.
      • ultimatt1
        True, but the oddity of From: being a list deserves his attention
      • well, possibly being a list
      • _smf_
        Yeah; rare as hens teeth though.
      • DragonPunch
        what's the point of Carbon Copying (CC) if you can have multiple RCPT TO's?
      • Or an array of To:
      • ultimatt1
      • DragonPunch
        "secondary recipients"
      • I guess that's good enough.
      • _smf_
        DragonPunch: you're confusing headers with the message envelope.
      • The SMTP MAIL FROM:<foo@bar.com> sets the envelope-from and RCPT TO:<bar@foo.com> adds envelope-to
      • The envelope data is what is used for routing.
      • Just like with snail mail
      • cek
        kibana doesn't support objects in arrays
      • _smf_
        However inside the message the Headers like To: From: Cc: Bcc: are totally informational and not used at all for routing etc. They are displayed only by the MUA.
      • The MAIL FROM: and To: addresses can be totally different.
      • DragonPunch
        Ah, okay.
      • _smf_
        It's a common misunderstanding with email
      • cek
        the might not even exist
      • *they
      • ultimatt1
        informational only when DMARC is not in the picture. :) Then, the From: header domain must be aligned with either SPF or a DKIM signature
      • DragonPunch
        It's just really easy to get confused. Especially when I step back and ask myself from a scientific perspective "Why?" is this happening. Like, "Why does the squirrel eat the acorn from the tree". Sometimes, it's just best not to ask "Why?".
      • ultimatt1
        only if you like "magic"
      • DragonPunch
        Hmmm, I always wanted to be a wizard like Gandalf. Hehehe.
      • lnb
        hmm
      • client company tries to send me mail. they use google and its bouncing back '550 HELO protocol mismatch'
      • Technical details of permanent failure:
      • Google tried to deliver your message, but it was rejected by the server for the recipient domain outbounds7.obsmtp.com by outbounds7.obsmtp.com. [64.18.6.12].
      • it is rejecting itself?
      • anyone?
      • Currerius has quit
      • DragonPunch has quit
      • DoubleMalt joined the channel
      • dopesong joined the channel
      • doublemalt_ joined the channel
      • dopesong_ joined the channel
      • DoubleMalt has quit
      • _smf_
        lnb: it would seem that your Google sender is proxying their outbound via Outblaze and that's doing some weird protocol stuff.
      • Set mismatch=false in helo.checks.ini under the [reject] section to allow it.
      • lnb
        so its our haraka server bouncing it huh?
      • _smf_
        Indeed - that is one of the tests from the helo.checks plugin that I wrote.
      • Basically - they're sending a HELO or EHLO, then later sending the opposite.
      • Which shouldn't happen.
      • lnb
        looking under reject there is mismatch=true now
      • # it out?
      • _smf_
        Really? ^^^ read what I said.
      • lnb
        sorry set it
      • restart haraka?
      • _smf_
        No - you shouldn't need to. Haraka will notice it's changed and will reload it.
      • lnb
        ok
      • maillog shows nothing
      • loglevel=debug
      • i hope it works, thanks _smf_
      • i will send the client email telling them to try again
      • lnb has quit