Nah, it 401's if you don't include the header. Try "manually" sending a request with cURL.

curl -i \
    --aws-sigv4 'aws:amz:us-east-1:s3' --user '<id>:<secret>' \
    -H 'X-Amz-Content-Sha256: e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855' \
    https://e5f0f929bb59b9d5e1bc9b35e9e3bf00.r2.cloudflarestorage.com/test/automatica.mp4

Wait, CURL has a dedicated aws-sigv4 flag?!

Yeah, the later versions do. And I have the latest version, of course, I run Arch btw 😛

The S3 specification allows signing a request with

X-Amz-Content-Sha256: UNSIGNED-PAYLOAD

so you don't have to calculate the hash. But R2 doesn't seem to accept that yet :/

Use

HTTPS_PROXY=http://localhost:12345 rclone --no-check-certificate ...

. I also recommend running mitmproxy with

--set stream_large_bodies=1m

to avoid massive memory usage.

thanks!

mitmproxy stores the entire request/response body in memory by default. So this flag is useful if you're pulling internal d... ahem, I mean, if you're downloading/uploading files from/to R2.

@Vitali are there any plans to support

X-Amz-Content-Sha256: UNSIGNED-PAYLOAD

?

That should already be supported. That was the first thing we did

I think I mentioned. With multipart uploads you want to be very careful to not upload too many parts at once (1-2 is safe, 2-3 is medium risk, 4+ is probably not going to work)

I'll be revamping that post open beta

Working on migrating https://www.github.com/helloimalastair/firenza to R2. Very exciting!

Just checked this against R2, and it totally works. Need to set UNSIGNED-PAYLOAD as both the payload hash in the canonical request, and also as the

x-amz-content-sha256

header value

Oh, yeah, you're right. I forgot cURL doesn't include additional headers when using --aws-sigv4, my bad...

still missing presigned urls though (which also use UNSIGNED-PAYLOAD)

I'm really loving seeing folks and their

r2-sharex

subdomains 😄

is r2 beta only for companies? the form says "work email", but i plan on using it for personal projects

Every once in a while, some employees will drop beta access for whoever wants it, so keep an eye on this channel

when was the last time?

4/11

i implemented s3 myself in sharex without use any library, is something wrong in my implementation?

how so every sharex user here have access to r2, except me as developer of the app

yea it was most difficult uploader i ever implemented in sharex

and i implemented hundreds of uploaders

while debugging i was losing my mind

luckily i could implement it in single day

because s3 was retuning that something is wrong in signature, and i had no clue what is, so it was hours of try and error to figure out what part is causing the issue