Howdy. I migrated to my single user server and just want to verify something, using cloudflare is fine correct? I have not received a single follow/follower update on my instance, and I’m not sure if it is normal for a brand new instance to take more than a few hours for any of these requests to resolve.
LGBTQIA+ and Tech post as seen outside
Akkoma post as seen inside
I manually ran the follow API call via curl targeting an account i knew i was not following. the response was:
{“id”:“AmQVBI8nCtBHXhfO6a”,“following”:false,“note”:“”,“blocked_by”:false,“blocking”:false,“domain_blocking”:false,“endorsed”:false,“followed_by”:false,“muting”:false,“muting_notifications”:false,“notifying”:false,“requested”:true,“requested_by”:false,“showing_reblogs”:true,“subscribing”:false}
when asking the person I “requested” to follow, they do not see any indication of this request.
Edit: this is a red herring turns out-- I am technically following them (show up on their list, and if i check my account on an instance that knows of me, it shows up there.) None of it is showing up on my instance, though.
you say migrated - you wouldn’t happen to have reused a domain or anything would you? is this a brand new instance on a totally fresh (sub) domain?
yes this is totally brand new, i migrated from a mastodon instance.
While I don’t have personal experience with using Cloudflare, I have heard people say that it can cause federation issues by dropping requests it considers potential unwanted traffic or something. Don’t ask me details though, I’m just repeating what I vagually remember others say.
Having said that, the way I understand Cloudflare to typically work is that they proxy “incomming” requests. So the outgoing requests should not be bothered by this 
But maybe they needed to fetch some key and that failed or something. I’m just guessing, hough.
Also note that in such a setup, other instances will still see your server’s IP. If the server is publically accessable through it, it’s trivial to circumvent Cloudflare. I’ve done it in the past as a proof of concept.
do you think i should stop cloudflare proxying for my instance then? I am not sure how to check for failed outgoing requests, my logs show a lot of ‘could not fetch’ for other instances &c,
07:32:53.773 request_id=F_lYE68ZcjYyiC8AAPRy [error] Could not decode user at fetch https://nafo.uk/users/Burito_Bandito_, :closed
07:32:53.773 request_id=F_lYE68ZcjYyiC8AAPRy [error] Could not fetch user https://nafo.uk/users/Burito_Bandito_, {nil, {:error, :closed}}
07:32:53.944 request_id=F_lYE68ZcjYyiC8AAPRy [error] Could not decode user at fetch https://nafo.uk/users/Burito_Bandito_, :closed
07:32:53.944 request_id=F_lYE68ZcjYyiC8AAPRy [error] Could not fetch user https://nafo.uk/users/Burito_Bandito_, {nil, {:error, :closed}}
07:32:54.022 request_id=F_lYE68ZcjYyiC8AAPRy [error] Could not decode user at fetch https://nafo.uk/users/Burito_Bandito_, :closed
07:32:54.022 request_id=F_lYE68ZcjYyiC8AAPRy [error] Could not fetch user https://nafo.uk/users/Burito_Bandito_, {nil, {:error, :closed}}
07:32:54.164 request_id=F_lYE68ZcjYyiC8AAPRy [error] Could not decode user at fetch https://nafo.uk/users/Burito_Bandito_, :closed
07:32:54.164 request_id=F_lYE68ZcjYyiC8AAPRy [error] Could not fetch user https://nafo.uk/users/Burito_Bandito_, {nil, {:error, :closed}}
07:32:55.469 request_id=F_lYFBVJS9xeVSQAAPSi [error] Could not decode user at fetch https://nafo.uk/actor, :closed
07:32:55.539 request_id=F_lYFBVJS9xeVSQAAPSi [error] Could not decode user at fetch https://nafo.uk/actor, :closed
but this may just be an issue with them not support pleroma or something.
Personally I don’t use cloudflare, nor do I think most people I know do. It depends on what you need cloudflare for, but if you can live without it, it’s at least worth a try to drop it, even if it’s just to see if the problems go away. (Don’t forget you’ll need to set DNS records again, so wait at least the TTL before trying again so that everyone has the correct new IP address.)
If cloudflare has some view where you can see what requests it received and what it did with it, maybe that can help you figure out if it blocked things. But again, I have no experience with this, so I’m not even sure if or how or where you could find this.
The errors that you see , are probably not the issue. It’s because fetching of those actors fails, but when I try, it seems like a problem on their end. Tbh, these should probably not even be logged as errors.
I tried just dropping the cloudflare proxy but that doesn’t seem to have done anything. to migrate off cloudflare i’ll have to move a lot of subdomains and shuffle things around so that will take awhile; it’s not just the instance on here.
do you know of any way i could try to diagnose what is going on for some peace of mind on what the actual issue is? the logs don’t seem to give me failed responses when someone asks for a follow. im not sure if another instances admin could see the output then? I’m a bit in the dark here.
fwiw, I see your follow request. You will see it as “request send” because my account is locked, but my instance did receive it.
The issue is when you try to follow back. if you try that, i believe it will just say ‘requested’ and i will never receive any word of it.
just to touch base again:
i have a well circulated post from my instance (~30 boosts),
i am following ~ 20 people (according to tech.lgbt’s “view” of my user)
When I follow someone, the person receives the request immediately. it does not reflect on my own instance or update my instance follow count. I can not see their posts on my timeline, though the public timeline works just fine.
When someone follows me, they see a “Request pending.” When I hit my “pending follow requests” endpoint, it tells me i have none.
I am using SSL on cloudflare and every other site i host with this works just fine. I am going to migrate off cloudflare just for peace of mind, but someone else on here was using it and it worked for them.
I know no way to verify or even diagnose the issues that are being caused here; the logs on both akkoma_akkoma_1 and akkoma_db_1 don’t seem to help.
I migrated from a mastodon instance, tech.lgbt. I imported my ‘following’ and it only managed about 15 the first time, then 16, then 20. it doesn’t seem to be able to finish the full 67 followings.
My friend has joined the instance, we can follow eachother and reply to eachother just fine.
Everything got sorted. Wanted to give a final update for anyone stumbling upon this.
The issue was the nginx header settings, so i re did that entire process. the issue is entirely through my own faults, but if it does happen to you know that the symptom is a feeling of “half federating.” that is, anything that you reach out for works, but anything that reaches to you fails.
unfortunately this means the migration failed to restore all my followers, but now everything is working well.
2 Likes