idk, a counter api ig?

to be clear i mean an api where a user could hit or get an id to get a count

i was thinking kv but i wasn’t sure if that would propagate globally fast enough

If it was a counter you want to increment, Durable Object Storage. R2 is also strongly consistent but pretty overkill. If it was something like a website counter, you could write to a durable object, which stores locally and also writes to KV, which you could read and have cached as long as if you're fine with the result not being consistent

KV is eventually consistent and limited on writes, so you wouldn't want to use it alone (at least for that use case, ex: if you tried to read and then update the counter from KV in two colos at once, you lose one of them)

thank you for such helpful insight :)

yup that was what i took into consideration and was trying to choose something other than kv

i got another question similar to the previous one and it’s pretty complex/off topic, mind if i ask u in dms?

counting would be very ideal with #981314061268578304 if all you’re doing is “adding”

If it's something workers related/non-private could ask in #1052656806058528849 post and other people can potentially benefit from it, otherwise sure

hm how so? also ideally i’d like to have add and subtract and some other stuff, so it won’t be strictly ++

yeah then probably DO would be best. I was just thinking how cheap it could be if you were just incrementing a count with AE

but Chaika is right, DO and a mix of KV would be nice

Analytics Engine is great for writing append-only statistics / billing / etc, the workers binding is easy to use, but if you need more control / the data to never be sampled, then would need to use something else

Can workers copy or move files of R2? I can't see it in reference https://developers.cloudflare.com/r2/api/workers/workers-api-reference/#bucket-method-definitions

I am not sure why copyObject is missing there, but there is no concept of moving on R2 (or S3), you would do a copy and then delete the original

And if they had copyObject it would be pretty speedy because the server would do it intra-server and not pass the file through your worker

I know there's no move or rename in S3, but I cannot find copy in the docs neither

Yeah it's supported but I am not sure why the Worker API doesn't have it

I tried a naive

bucket.copy(from, to)

just in case someone forgot to put it in the docs, but no, haha

I tested

const got = await bucket.get(from)
await bucket.put(to, got.body)
await bucket.delete(from)

and works fine, can it have problems with big files?

That works but it's not great because the file passes through your worker, yeah

It should be fine if your files are less than a few gigs

Theoretically, it should be able to handle any file in R2, but it gets more unreliable the larger it is...

And execution times, limits and all that stuff could be a problem?

I mean, egress is free.. but we've pushed 100TB ea. for three months via R2 via CF CDN and received a "letter of invitation" to discuss moving onto Enterprise 😁 So I would say it's technically free, up until some point

Not really. If you are just streaming, it doesn't count toward any limits, other than the 30 second minimum runtime cap.

All limits are * with "Hey, you might be Enterprise level. Maybe you should upgrade?" It all just depends on how much you use.

When Business is 200 and Ent is 5K it's a little steep. But we understand it 🙂

Then I'm fuqed O.o