When will R2 Public Access work with gzip + http range? Currently, when specifying a

Range

header, e.g.

bytes=0-10240000

, the response is never gzipped, regardless specifying

gzip, deflate, br

. I've solved this using a worker now, but that's more expensive.

For reference, AWS CloudFront supports it.

Hi - I've noticed that I'm getting cached results when viewing an updated file in a bucket. I've tried every combination of Cache-Control header in my fetch call, but it's constant. A shift refresh appears to then pull the updated record back. I was looking at the response headers and I don't see anything there around cache either. Is there a setting in the bucket that I'm just not seeing, or any advice on how to ensure a latest data returned from a JS/TS fetch call?

Public domains follow normal CF cache rules - so your request's Cache-Control header doesn't really do anything

If you want to disable caching on a domain, use a Cache Rule

@kian thanks for the response. I guess the challenge here is that I have a private application to set values in a json file - I want to see the immediate change in that file after it has overwritten the existing one there. I don't want to disable the cache, because the client applications that read that file should in deed get a cached response for the most part.

It's also for one specific bucket at the moment, not the entire domain - not sure if that makes a difference, but reading the docs with respect to the cache mechanic, I was thinking that Purging the caching was a direction worth exploring, but it's not exactly easy reading to understand what is the right way to go

Use the S3 API to read that file w/o using the custom domain?

Not sure if Cache Rules allow you to bypass cache based on a query param - you could also try that

Thanks - some great ideas there for sure. S3 API might be the way to go. I'll do a test. Cheers

@kian - my bad. I just noticed a single non async fetch call in my code. So I was writing the update and running a query on the bucket before the update had completed. Thanks for you quick response, really appreciate it. Cheers

sorry i meant the s3 url

actually neither the Public R2.dev Bucket URL nor the S3 API url are advertising h2 in the alpn, only http/1.1

r2.dev doesn’t matter much. You can’t put it into production because it has very row rate limits.

I don’t know exactly why the S3 API would not have http/2 but I also don’t think you’ll derive much benefit from it there

That did the trick, thank you!

If I set the Custom Domain for R2 to be a subdomain of a domain on Cloudflare, is creating a Page Rule for the subdomain, to set Cache = Cache Everything, the best way to achieve that?

Use a new Cache Rule, but yes

Page Rules are on an undefined chopping block for the future

thanks

I tried to set this CORS policy on my R2 bucket at `storage.thatsnomoon.dev`:

json
[
  {
    "AllowedOrigins": [
      "http://127.0.0.1:1111",
      "https://thatsnomoon.dev"
    ],
    "AllowedMethods": [
      "GET"
    ]
  }
]

but requesting one of the objects in the bucket, like

https://storage.thatsnomoon.dev/charter_regular.woff2

, still doesn't send a

Access-Control-Allow-Origin

header, whether I do it in the browser or

curl

with or without an

Origin

header. what am I missing?

I am getting a CORS response, maybe clear your Cloudflare caches?

ahh, that looks better. thanks!

If you expect globally distributed users, what would be the recommendation for bucket location? Should you create different buckets for each region, and then implement some logic on the worker to decide which one to use? Or is is better to create just one bucket with one location, and Cloudflare optimizes it behind the scenes depending on from where the worker accesses it?

just 1 bucket for 1 location 😐

I would place the bucket where you expect the most traffic to come from, or where it will be closest to your other software that might access it. But it really isn’t going to kill you in the long run.

Hi all, am currently trying to perform a multi file upload using cloudflare workers and R2 but I keep getting this error

EntityTooSmall: Your proposed upload is smaller than the minimum allowed object size.

I previously tried using the workers R2 Runtime error and got the same error too. Any tips on how to debug this?

Hey, is there a way to batch download a folder?

You can use https://rclone.org to copy a folder to your local device