XMPP Service Operators - 2024-01-28


  1. Menel

    Sam: seems they have some stats on the matrix resource issue: @praveen@social.masto.host 🔗 https://social.masto.host/users/praveen/statuses/111833599737644731 - We found volunteers to continue #diaspora and #xmpp, but had to drop #matrix, which also made it possible to scale down the server and reduce monthly cost to 1/3rd.

  2. MSavoritias (fae,ve)

    tbh this myth of things are going to be fixed soon has been happening in matrix for years. sliding sync, clients, dendrite that never comes. its probably hard to convince people that are deep into the matrix ecosystem of other things. not that its not true for xmpp either

  3. MSavoritias (fae,ve)

    tech can be pretty "team" oriented

  4. Sam

    Menel: thanks, I'll ask Pirate!

  5. Arya Kiran

    Used to run a ~600-700 user matrix instance (well only a fraction were active of course), and to have it running smoothly it took easily 15-20 GB on postgres alone

  6. Arya Kiran

    And then above that a few gigs of ram on synapse

  7. Arya Kiran

    And the way matrix stores media is very, very dumb

  8. Arya Kiran

    (Every server stores a copy of the media, and it will be re-fetched to the cache even if cleared if a user tries to access it)

  9. Arya Kiran

    Which leads to a few 100 gigabytes of media_store for no reason usually :))

  10. edhelas

    Looks like the Matrix guys have some shares at Seagate and Kingston

  11. klaudie

    serious or joke?

  12. Polarian

    Why would they need shares at Seagate and Kingston anyways

  13. Polarian

    they are a protocol organisation, not a storage media manufacturer

  14. edhelas

    I'm always 100% serious

  15. Polarian

    Arya Kiran, Matrix does that to "fix" a "flaw" in XMPP, which is if a XMPP server goes down, you can't view the messages or media on the server, XMPP servers store the data for their respective users. Matrix instead has all connected peers download the media, spreading it across all the peers therefore meaning if a server goes down, you can still see the media they sent without needing to go fetch it from then.

  16. Polarian

    Its meant to resist disruption from losing peers

  17. Polarian

    _or that is what I heard at least_

  18. Polarian

    But in reality its just spreading your data across half the internet... single points of failure sometimes is a good thing...

  19. A4 paper

    good place to report users for RTBL addition?

  20. edhelas

    In the end, because of this behavior, the server requirement for Matrix is higher... leading to maybe less nodes on the network

  21. Licaon_Kter

    A4 paper: what did they do and where?

  22. A4 paper

    > A4 paper: what did they do and where? just being rude in a small group chat, but I know who it is (templeos), and they're almost certaintly going to use those accounts to spam in other rooms. Dunno if that's enough to add them

  23. A4 paper

    >being rude his standard diaper/scatalogical stuff

  24. array

    > Arya Kiran, Matrix does that to "fix" a "flaw" in XMPP, which is if a XMPP server goes down, you can't view the messages or media on the server, XMPP servers store the data for their respective users. Matrix instead has all connected peers download the media, spreading it across all the peers therefore meaning if a server goes down, you can still see the media they sent without needing to go fetch it from then. > > Its meant to resist disruption from losing peers > > _or that is what I heard at least_ this is correct