XSF Discussion - 2023-08-04


  1. emus

    I will open and communicate to collect all out information today

  2. moparisthebest

    where do I report bugs and/or ask questions about connect.xmpp.net

  3. moparisthebest

    (I'm guessing that's a question for jonas’ )

  4. MattJ

    The frontend, me, the backend, jonas’

  5. moparisthebest

    uh, probably backend, Server-to-server connectivity both said "Connected" even though Direct TLS was broken

  6. moparisthebest

    actually prosody using people could join the MUC but ejabberd folk could not, it's a seperate bug I'll need to report to them :D

  7. moparisthebest

    but I fixed my server so either need a badxmpp for this or I can set up another broken one

  8. jonas’

    moparisthebest, broken in which way exactly?

  9. moparisthebest

    well it connected to a web server that didn't speak XMPP at all

  10. moparisthebest

    the only difference in the output was it didn't say "SASL mechanisms: EXTERNAL"

  11. jonas’

    hm

  12. jonas’

    that's sad

  13. jonas’

    myeah, reproduces when using xmpp:https.badxmpp.eu as a test case

  14. Zash

    http or https.badxmpp.eu

  15. jonas’

    quite sad indeed

  16. jonas’

    moparisthebest, I filed https://github.com/horazont/xmpp-blackbox-exporter/issues/24 for tracking, not super inclined to work on this right now

  17. moparisthebest

    the ejabberd bug is it doesn't proceed to other SRV records, just "remote timeout"

  18. moparisthebest

    thanks jonas’ , just wanted to get it written down somewhere

  19. moparisthebest

    oh no, new prosody has the same bug as ejabberd

  20. moparisthebest

    gonna have to set up another broken testcase

  21. Menel

    Not moving on srv records? I think I encountered that once

  22. moparisthebest

    Yep

  23. Zash

    Wait do we have Direct TLS s2s in a release already?

  24. Zash

    Oh, I thought that was trunk only still.

  25. Zash

    But Happy Eyeballs aren't in a Prosody just yet, which does a lot for avoiding all those serial per-connection timeouts adding up to the point where it gives up on the whole s2s attempt.