I have an instance (mander.xyz) for which the database was deleted, and I have set up a fresh new instance at the same domain.

Federation appears to still be enabled with lemmy.ml, but I am having general issues fetching and interacting.

For example, if I try to access https://mander.xyz/c/linux@lemmy.ml, the error log shows:

2021-12-11T17:53:52.201426Z ERROR lemmy_websocket::handlers: Error during message handling error sending request for url (https://lemmy.ml/.well-known/webfinger?resource=acct:linux@lemmy.ml): error trying to connect: dns error: failed to lookup address information: Try again

I posted a somewhat related issue on GitHub , and it was pointed out by dessalines that “An activitypub actor id must be unique, and you recreated one with new public and private keys”. It is not clear to me whether these ids apply only to specific usernames, or if the actor is a more general concept (community, post, instance as a whole) to the point that federation is irreversibly broken.

Are there steps I can take to fix the federation? Or is starting a new instance using a different domain the only way forward to re-enable federation?

I do want to point out that during certain periods of the day federation suddenly appears to work just fine for a few minutes.

Thank you!

  • bashrc
    link
    fedilink
    arrow-up
    2
    ·
    3 years ago

    Easiest way would be to set up a new instance on a subdomain. Federation problems will be due to keys for the domain being cached on other instances, including for the instance actor.

  • nutomicA
    link
    fedilink
    arrow-up
    2
    ·
    3 years ago

    I replied to your issue, the problem is something different, recreating an instance should be totally fine.

    • Salamander@mander.xyz
      link
      fedilink
      arrow-up
      1
      ·
      3 years ago

      Thank you! I have recorded the log of the curl command and am trying to understand it. I will reply with it in a bit.

      As for the federation… The federation appears to be working well again, at least for the moment. I made no changes, hopefully it stays!

      • nutomicA
        link
        fedilink
        arrow-up
        2
        ·
        3 years ago

        Sending comments from your instance to lemmy.ml doesnt need webfinger, so that is expected. Webfinger is only needed for profiles and mentions iirc.

        • ZalamanderOP
          link
          fedilink
          arrow-up
          1
          ·
          edit-2
          3 years ago

          The problem is that there is an issue with my server’s DNS resolution. I can’t ping lemmy.ml during these periods from the server, but can ping fine from my home computer. I will figure out what’s wrong with the DNS.

          I tried to reply from my instance (https://mander.xyz/post/655), but at the moment the federation is not working again. This is what I wrote:

          During the day the federation will work during a few hours/minutes, and then it stops working again. For example, if I try to navigate to a lemmy community from my instance at this moment, I get the following dns lookup error:

          lemmy_1 | 2021-12-12T12:11:34.411206Z ERROR lemmy_websocket::handlers: Error during message handling error sending request for url (https://lemmy.ml/.well-known/webfinger?resource=acct:linux@lemmy.ml): error trying to connect: dns error: failed to lookup address information: Try again

          Between my and other lemmy instances (forum.purplerabbit.xyz, sopuli.xyz, dissonanz.xyz) federation appears to work fine continuously. I have also tested whether the federation between lemmy and other instances is also not working during the same period time, but they federate just fine. It appears to be a problem specific to my instance.