Workers from X API fails due to invalid response (just "IPv6" plain text)
I have a Cloudflare Workers app and it periodically calls X api every 30 minutes. Since July 3 2025 05:30UTC, the workload began failed due to
";SyntaxError: Unexpected token 'I', "IPv6" is not valid JSON" error. I found the response from api.x.com is just "IPv6" string with 403 http status. I tried to call X API on my local env (WSL Ubuntu) and it worked. So I guess the issue is on Cloudflare Workers. Does someone see same issue?15 Replies
You mean I should ask X support?
Honestly (and personally) I don't expect X support can handle the issue appropriately. As you say, migrating the workload to other platform would be better than asking X support...
But I'm wondering why it happened suddenly today. Does it mean X didn't use Cloudflare until yesterday?
I found same issue is already reported on here:
https://devcommunity.x.com/t/getting-403-ipv6-error-from-cloudflare/245157
Okay thanks
I'll keep an eye on the community thread and consider the workload migration if it takes long time to fix
In case it helps anyone:
I think I fixed our usecase by proxing the request through Netlify. I've deployed a small Netlify function just for X API calls which runs on AWS Lambda and handles IPv4/IPv6 better
is there any better way like forcing IPv4 on cloudflare workers
or setting a rule in cf zones so any request to a domain will be IPv4?
:feelsbadman: that's really painful...
I seems like twitter does not notice this problem
Yeah I think so too. But I don't think Twitter would handle the issue even if Twitter found out. Twitter might not think it is even a problem. After Elon's acquisition, Twitter keeps corrupting
Ugh was trying to find out why twitter logins were failing suddenly good to know it's not a me problem at least..
currently my only way is making a twitter proxy then deploy it to a free hosting like render.com
but with only 100GB bandwidth and a super weak instance
currently that's the best way that I can think of
another way is making use of oracle free tier
My scheduler job (runs every 30 minutes) exited without errors at July 4 11:00 UTC. Refreshing OAuth2 token succeeded. The error seems to be gone. X has fixed??
ooh? I was taking a nap and many happy returns? 😄
Yeah I don't see any errors so far since July 4 11:00UTC. I'm not sure if it's temporary or not. We might want to keep an eye for a while. I hope it could be permanent solution
yup creating simple text tweet also succeeded: https://x.com/ricemountainer/status/1941152147334475867
this guy (from Cloudflare) confirmed that this problem should be fixed by now. It seems like Twitter has misconfigured their networking that accidentally blocks IPv6 requests from cloudflare's cross zone IP.
i was thinking of scheduling Twitter worker to other place to mitigate the problem
I see. Thank you. I'm glad that I didn't need to make a proxy...