23 Replies
You didn't think this was from your worker right? Could it be coming from something at heroku?
Not really sure how they work 😅
it definitely could! just find it very weird that it 503's directly from Heroku
If you visit the worker directly, or use a tool like Postman, Insomnia, Httpie, etc. to make a request like heroku would, do you get the error as well?
nope, it works, so I'm thinking either this is something with Heroku, or CF is blocking or rate limiting the heroku instance
If CF was blocking the herku instance, I wouldn't expect a 503 error 🤔
hmmmmmm
looked like 503's often was the result of rate limiting
I'll dig a bit deeper
If the CF ratelimiter did it, you would get a 429
that's good to know!
Do you also have the page that's returned with that request? Maybe that reveals where it's from
this is from the server itself, done on the backend
The heroku server?
A “Error 503: Service Temporarily Unavailable” with “Cloudflare” means you are hitting a connection limit in a Cloudflare data center. When this happens, please contact Cloudflare Customer Support. Include as many of the details mentioned in the Quick Fix Ideas as possible.yeah, request is sent from the backend at heroku to my worker looking like this is a CF issue
Ooohokay, so the worker is requesting heroku? I missed that part 😅
nope
I have a server hosted on Heroku that is making a request to a CF Worker
the request is being done by the backend (on Heroku), not from client-side visiting a URL on Heroku
apologies for my lack of clarity 😅
Ah okay then I understood correctly! Still not sure why it happens though, but it could be that thing you quoted above 🤔
it doesn't even show up as a request on the real-time logging
when I use the worker.dev route it works
these two should, in theory, be the exact same
but as you can see,
https://files.instorier.com/file/1/0VxftDQZL2.json returns a 503, while https://images.instorier.workers.dev/file/1/0VxftDQZL2.json correctly returns 404
Just to confirm, you have the orange cloud enabled in the DNS for files.[yourdomain], right?
yes
and I'm using the discard-way of routing it in the DNS settings
which I'm now seeing is the old way
just updated this to using custom domains, and the errors are still there
