#haraka

/

      • gitgud is now known as bitbud
      • bitbud is now known as gitgud
      • knutix joined the channel
      • last1
        most likely you would need a custom setup... that toaster looks sweet but I'm not sure how easily it is to grow
      • we are also in the process of moving out filtering away from SA to rspamd with the help of Haraka
      • s/out/our/
      • __smf__ joined the channel
      • __smf__ has quit
      • __smf__ joined the channel
      • I use the rspamd plugin and my own plugin, after rspamd, makes use of transaction.data_bytes
      • shouldn't rspamd update the data_bytes variable with the new data it added in the headers ?
      • or even more generally, shouldn't the add_header function automatically adjust the data_bytes ?
      • say that my downstream mail server accepts X data_bytes, I add 5-10 headers which cause me to go over that limit, but when I check data_bytes to make sure I'm good, it shows I'm ok when in fact I'm not
      • _smf_
        last1: see Base64 encoding overhead.
      • Unless of course the headers you are adding are huge.
      • data_bytes is the size of the input, not the output FWIW
      • Trying to keep data_bytes in step with the modified headers would be a waste of time, and it would be unnecessary overhead.
      • last1
        that's what I figured
      • it's just that I am writing a copy of all inbound e-mails
      • and I am also writing the file name as S=<size>
      • which I was getting from data_bytes and it turned out to be incorrect
      • the only other way I can think of is to write the file to disk, get its size and then rename the file
      • _smf_
        Why is the actual size so important to you?
      • last1
        well, in Maildir it saves dovecot the trouble of finding it rewriting the filename
      • the files I write will be ready by dovecot directly
      • read
      • _smf_ shrugs.
      • _smf_
        Deliver it to Dovecot over LMTP. Problem solved.
      • last1
        I do that too. This is an archive copy
      • before it goes through filtering
      • I guess I'll just not write the S parameter and let dovecot figure it out
      • _smf_
        If you save it *before* anything else runs at hook_data_post, then the size should be correct no?
      • last1
        indeed
      • but I save after rspamd and after my own plugin which writes its own headers
      • _smf_
        I suggest it's not worth it. Let dovecot work it out and rename it. It's hardly expensive to do a rename.
      • last1
        cool, thanks
      • I was surprised in some cases how big the difference was
      • failed: Cached message size smaller than expected (458 < 1417
      • _smf_
        Yeah - that seems huge.
      • bmonty joined the channel
      • __smf__ has quit
      • gitgud has quit
      • gitgud joined the channel
      • knutix_ joined the channel
      • knutix joined the channel
      • knutix_ has quit
      • knutix_ joined the channel
      • knutix has quit
      • gitgud has quit
      • gitgud joined the channel
      • EyePulp has quit
      • EyePulp joined the channel
      • gitgud has quit
      • gitgud joined the channel
      • nedry_ joined the channel
      • nedry has quit
      • nedry_ has quit
      • nedry joined the channel
      • DoubleMalt joined the channel
      • nedry has quit
      • nedry joined the channel