Cloudflare Developers•4y ago

How to generate the URL directly uploaded by the browser?

Yyinxingmaiming6409 How to generate the URL directly uploaded by the browser?

SakuraFC•10/28/22, 8:18 AM

https://kian.org.uk/configuring-cors-on-cloudflare-r2/

should be a good starting point

Kian's Blog

Configuring CORS on Cloudflare R2

Wondering how to add CORS support to your R2 buckets? We'll go over how to format a CORSRule and allow presigned PutObject requests.

kev-ac•10/28/22, 12:35 PM

I'm sure I've missed that here somewhere, but: Are there any plans to restrict api tokens to specific buckets in the near future?

kev-ac•10/28/22, 12:36 PM

Just knowing that all data of all buckets is in danger if one app has a vuln which may leak a keyset is a bit worrying.

Kkev-ac I'm sure I've missed that here somewhere, but: Are there any plans to restrict a...

Sid•10/28/22, 1:56 PM

Yep! This is pretty high on our list

Zonimi•10/28/22, 1:58 PM

I am getting this error VERY often. get: We encountered an internal error. Please try again. (10001)get: We encountered an internal error. Please try again. (10001) NotLikeThis

kev-ac•10/28/22, 2:09 PM

500 error usually means that either your request is malformed that R2 can't respond correctly or some bug, etc.

kev-ac•10/28/22, 2:11 PM

R2's model is strong consistency so any mutating action (POST, PUT) is reflected immediately (except deleting a bucket) and there is no replication lag across regions (as data is stored in one region)

kian•10/28/22, 2:27 PM

There is no replication currently - your bucket's data is kept in one location.

kian•10/28/22, 2:27 PM

The 500 sounds like an issue with R2 more than anything else

JJamesXXV the file itself is 3 bytes (text/plain), you do raise a good point about immedia...

kev-ac•10/28/22, 2:27 PM

The moment R2 responds with a "successful" response code (20x) the mutating action is done. Any GET after POST / PUT are immediately available (independent of size)

kev-ac•10/28/22, 2:29 PM

One exception will be Multipart uploads. There the change happens when you complete the multipart upload.

kev-ac•10/28/22, 2:31 PM

With S3 there is one edge case where GET requests can respond with the old or new value when upload is done but has not yet been acknowledged to the client. But I don't know how R2 handles this as it's so unlikely to encounter.

JustinNoel•10/28/22, 4:52 PM

It doesn't look like there's a way to detect if an object exists using a Worker.

For example, I'd like to do something like this just to see if an object exists or not.

await env.MY_BUCKET.list("my_key")await env.MY_BUCKET.list("my_key")

Right now, it seems the only way to do that is

await env.MY_BUCKET.get("my_key")await env.MY_BUCKET.get("my_key")

This seems wasteful as I don't actually need the object's data or want to wait around for a 10MB file to be sent across the wire. I just want to know if it exists.

Is there any way to do this?

JJustinNoel It doesn't look like there's a way to detect if an object exists using a Worker....

Sid•10/28/22, 5:01 PM

Yep, have you considered a .head().head()? That will not respond with your object's body. Even for .get().get()s, you get back a stream with your object's body that you should be able to discard, you shouldn't need to buffer the entire body in memory.

CCorin Ah came here to ask, getting the same thing here

Sid•10/28/22, 5:02 PM

Yeah this is being investigated!

Erisa•10/28/22, 6:00 PM

Are there any plans to re-enable the CopyObject API after the issues causing it to be disabled a month ago?
https://www.cloudflarestatus.com/incidents/894125897dd7

Customer impacting issue related to R2 Storage's CopyObject API

SSid Yep, have you considered a `.head()`? That will not respond with your object's b...

JustinNoel•10/28/22, 8:18 PM

headhead did the trick. Thanks!

albert•10/28/22, 8:49 PM

What happened with my cors settings

I see a blank page instead of the 404 error page.
https://cdn.albert.lol/3d273fe7
https://cdn.albert.lol/a7cd22bc

albert•10/28/22, 8:50 PM

whats wrong smh

kian•10/28/22, 8:58 PM

That isn't CORS related

kian•10/28/22, 8:58 PM

You visiting the page directly isn't cross-origin

albert•10/28/22, 9:01 PM

Uhm okay but I still don't understand what went wrong and how to fix it to see the error page instead of this blank nothingness

Sid•10/28/22, 9:10 PM

The default 404 page with the robots only shows up for public buckets. Is cdn.albert.lol configured as a custom domain to your bucket?

SSid The default 404 page with the robots only shows up for public buckets. Is cdn.al...

albert•10/28/22, 9:16 PM

It is yes

albert•10/28/22, 9:17 PM

but the public access url is returing the same blank page too (https://pub-02657f86241d4812aa81f9e57cc3b6ae.r2.dev)

Unsmart•10/28/22, 9:20 PM

Looks like only the home page is like that

Sid•10/28/22, 9:20 PM

Maybe you actually have an object with the name /

Unsmart•10/28/22, 9:21 PM

yeah my bucket still 404s on /

albert•10/28/22, 9:21 PM

I dont have an object with the name

UUnsmart Looks like only the home page is like that 🤔

albert•10/28/22, 9:22 PM

yeah the 404 works for non-existent objects but not for the / page lol what have i done

Unsmart•10/28/22, 9:23 PM

I think you do indeed have an object but I dont think those ones show up in the dash or list tbh... @James I feel like you probably have done this before do you know how to confirm if theres an empty named object

Unsmart•10/28/22, 9:24 PM

Or maybe Sid can also lol

albert•10/28/22, 9:24 PM

I searched for that in S3 browser, found nothing

Sid•10/28/22, 9:27 PM

Empty object names should now show up in the dashboard!
You’re right actually, objects with the name / will show up on the // pathname. @albert.lol you might want to check if you have an object with an empty name in your bucket

UUnsmart Or maybe Sid can also lol

Sid•10/28/22, 9:28 PM

FWIW you can create objects with empty names using the bindings/S3 API haha

How to generate the URL directly uploaded by the browser?

Similar Threads

Similar Threads

Similar Threads