XMPP Service Operators - 2018-09-27

  1. madmalkav

    hey guys, what and how do you check the health of your xmpp servers? you check the processes, the listeners, the availability of DB, ...

  2. madmalkav

    anything else

  3. madmalkav


  4. pep.

    I guess you can have a client that connects every so often and tries to do some actions

  5. edhelas

    madmalkav I check the complains on Twitter

  6. edhelas

    https://status.conversations.im/ is actually doing that pep. no ?

  7. pep.


  8. pep.

    That checks disco I think

  9. pep.

    I mean it depends how thorough you want to be

  10. madmalkav

    I don't want my company to call me at 4am because some operators are not receiving notifications on their cells

  11. pep.

    Well, if it's not the company it would be some kind of automated notification, you can't really escape that

  12. madmalkav

    and if that happens, I want my scripts to start the incident measures before people start calling in pacnic

  13. madmalkav


  14. madmalkav

    yeah, but if I automate it OK I can i.e. generate a warning to the 24/7 support to make some steps, and, if the steps fails, call me

  15. madmalkav

    ie., what is failing is DB, call DB guys, not me 😃

  16. pep.

    So the important bit is not the 4am, it's that you know before your users :)

  17. madmalkav

    being serious now, the important thing is detecting as soon as possible and as best as possible to minimize the downtime to a minimum

  18. pep.

    So yeah depending on how thorough you want to be, you can have checks on the server, and also emulate client behaviour

  19. madmalkav

    I can monitor with shell scripts, or python, but I will need help for that, and I can generate warnings to the 24/7 team so they take measures , or directly call me or other teams

  20. pep.

    If you do something like that for the client I'm also interested BTW :-°

  21. Holger

    I have Nagios/Icinga anyway and use a simple check plugin to talk to an echo bot.

  22. madmalkav

    If I do something like that I won't be able to share it , probably . I will have to redo it in a different way so company doesn't say I'm stealing their IP

  23. Holger

    (Plus separate database/HTTP checks, but that's about it.)

  24. madmalkav

    fuck big companies

  25. Holger

    I would suggest a proper monitoring tool rather than inventing wheels.

  26. Holger

    reinventing even

  27. madmalkav

    I thought you knew me well enough to know I don't have a word on deceiding what and how it is done

  28. madmalkav


  29. Holger

    Using existing software is bad?

  30. Holger

    Seems you weren't forced to implement an XMPP server from scratch at least :-)

  31. madmalkav

    here there is a lot of internal developed software that is totally obsolete but they bosses are being slow on moving on. I've hearing on moving surveillances to Zabbix for at least 2 years

  32. madmalkav

    My boss is a reasonable guy with lots of technical knowledge and he wants to run away from reinventing the wheel as soon as possible, but he is an exception here

  33. madmalkav

    So, zabbix / nagios / ... already have XMPP surveillance modules?

  34. Holger

    Dunnu about Zabbix, that has stuff built-in, maybe. Nagios/Icinga and others have nothing built-in but call simple stand-alone scripts/binaries to perform the actual checks.

  35. Holger

    You'll find tons of such checks on the web, I wrote the XMPP check I'm using myself though.

  36. madmalkav

    yeah, I'm probably overthinking it

  37. pep.

    Also forgot to mention metrics, but he's gone now

  38. pep.

    (Prometheus and the like, some services export this kind of stuff directly)