I believe someone said a long time ago that R2 public domains are only HTTP/1.1, there's no real benefit to them being HTTP/2

thanks

my stuff is definitely loading over h2

but it uses the public bucket with a custom domain not the r2.dev

maybe that's the difference

Out of curiosity, what is the precision on duration for r2? I know it's billed in gb-months but let's say data is stored for one day-- even hour, does cf round that to anything

Peak daily usage last I recall

So whatever the highest amount of usage was for a given day

Hey, I didn't know this discord existed until CF support gave me the link. I was monitoring the CF status page/twitter but didn't see anything, should I be keeping tabs on this discord for issues in the future or is there a better place?

You'll see status page updates when we declare an incident, and we didn't when this bug was reported (primarily because I didn't see a lot of complaints outside of Discord and our internal alerts didn't fire, indicating that this bug wasn't too prevalent). For major issues you'll definitely see a status page incident

Got it, ty for the quick fix

Seems like a strange thing to think. While R2 wouldn't be a great candidate for all HTTP/2 features, I'd think R2 public buckets would use HTTP request concurrency a lot for a site that is storing static assets on them (think like a photo site where a user is looking at a page of thumbnails coming from R2). HTTP/2 Multiplexing > HTTP/1.1 connection reuse. Also, the ability to compress HTTP headers is nice (a little less network traffic needed for every request isn't a bad thing either).

this is used in the R2 Cache API example:

ts
// Construct the cache key from the cache URL
const cacheKey = new Request(url.toString(), request);

will the

cacheKey

here be different if the

request

is a

GET

vs. a

HEAD

?

I'm guessing yes, just want to be sure 🙂

You can't store anything but a

GET

and then

match

has an

ignoreMethod

option

gotcha! so yes, the cacheKey will be different, but you can tell match to ignore the method?

makes sense. and then I just have to make sure I only put the

GET

requests into cache

In S3 I have restricted some directories to private in R2, can I do that with custom domain enabled? If I'm not mistaken, if using a custom domain will make everything public, then how can I use private?

what could a reason for my bucket reporting as 32.79 GB in the dashboard, but 15.3 GB using

aws s3 ls --summarize

?

I don't have any multipart uploads ongoing

I did add and remove some files in that time, is there a minimum time for which a file's size counts?

looking deeper,

aws s3 ls --summarize

tells me the bucket has 732 objects and the api request the R2 dashboard makes says 1462 (2x the amount)

what gives?

Hey 👋 , newbie question, for R2 migrator, if I run the sync with the same s3 prefix multiple times. will it figure out the difference and only transfer the delta?

usually yes

I was thinking of offloading my zoom cloud recordings to R2 with a webhook + CF worker proxy, but I'm a little unsure how the pricing works out for the whole thing. My recordings total ~1tb/month or 50k-60k minutes/month right now. Most recordings are ~500mb but can occasionally peak to ~4gb.

The R2 pricing seems straight forward @ ~$15 for the 1tb but idk what the pricing for the worker that proxies the recording videos would be?

My particular R2 public domain is not, do you have an R2 public URL that is?

Hey @rovercs, the current behavior would be to overwrite the existing files. More information on the behavior here: https://developers.cloudflare.com/r2/data-migration/r2-migrator/