Note: Edited to remove several questions as I found the doc that covered changing the domain name.

I set this instance up on my personal/development domain and would like to move it to a new, dedicated one before it picks up any more users.

I’ve got a few questions about how this would affect existing federation:

According to the backup/restore doc, it says that changing the domain name after federating will break federation.

I guess my questions relate to how badly and what is the best way to mitigate any damage? If I have to scrap this instance and lose everything, at this point, it’s not a huge problem.

  1. Will currently peered instances eventually drop my old instance after a period of unavailability? If not, are there any negative implications for having a dead peer?

  2. Would I be better off just to create a brand new instance, alert my users and have them re-register, and then decommission this one?

  3. Would returning an HTTP 301 response at my old instance be beneficial or would that cause problems for existing peers?

  4. Assuming no other instances are subscribing to my single community yet, would that make any difference?

  5. Are there any other gotchas I should know about?

Basically, in my haste to get started exploring running a Lemmy server, I let my development instance get published. I can easily move it to a production state, but I would prefer to do so on a dedicated domain versus my personal/development one.

Thanks in advance.

  • PTZ@lemmy.ptznetwork.orgOPEnglish
    4·
    1 year ago

    For #3, as far as the instance lists and the Lemmy Community Browser are concerned, I believe a 301 redirect should work as long as it redirects with the URI. e.g. old.lemmy.tld/c/Hangout -> new.lemmy.tld/c/Hangout . Eventually my new instance will be picked up by those, but I’m still not sure if/when my old instance would drop off (see question #1)

    That would allow anyone clicking into my old instance from any of the currently published instance lists to land on the new domain. But I’m not sure how currently federated instances would handle that.