Ah, probably something to do with multipart then
Ah, probably something to do with multipart then
Pdunno
Puploading to b2 works, s3 works, only r2 is not (with this error)
VIs there a speed difference between accessing the bucket via a public url and from a worker?
VOr rather, should there be a difference? I'm seeing lower latencies across the board when accessing files from inside a Worker
VWhat could be the reason behind it?
EIts expected that Workers will have lower latency but the difference shouldnt be too dramatic - what numbers are you seeing? And is it tested from the same regions?
Vblue: worker
gray: public url
gray: public url
Vit's not too dramatic tbh, i just wanted to know the fastest possible option
Vi'm curious, why is it faster via a Worker? Doesnt the worker itself add like 30ms latency? Or with the worker the request gets on the cf backbone faster?
EWorkers dont add 30ms latency to requests
EIf I recall correctly I believe it was something to do with the binding not having as much auth overhead, like for S3 you have to sign the request and then verify the signature and all that stuff because the request goes through
EPublic URL I'm not so sure on
VSeems like a valid explanation yeah..
How much overhead do workers add to a request? I mean it can't possibly be 0
How much overhead do workers add to a request? I mean it can't possibly be 0
Vhttps://blog.cloudflare.com/cloudflare-workers-the-fast-serverless-platform/ this shows around 20ms TTFB in US regions
VAlso, based on these outcomes should I just serve my files via Worker and not a public URL?
D20ms sounds about right, the worker is loaded while TLS handshake & validation and everything is performed, so you don't really see the cold starts as much
VI just ran an artillery load test too, and files via workers is almost 2x.. guess i'll disable the public url and serve assets via workers
EThat 20ms would be for the whole request, not just the added latency from Workers
Vis
s-maxage supported for R2 public urls? So will cloudflare honor the s-maxage set on the object?VIt does not seem like it 
it does not get stripped and from that i'm assuming it's not getting cached
it does not get stripped and from that i'm assuming it's not getting cachedPi think not, just like pages.dev (iirc)
just bind a custom domain
just bind a custom domain
VSolution was that I needed to add the correct extension to the end of the filenames (was just using UUIDs) so that CF would cache them.
as per: https://developers.cloudflare.com/cache/concepts/default-cache-behavior/#default-cached-file-extensions
as per: https://developers.cloudflare.com/cache/concepts/default-cache-behavior/#default-cached-file-extensions
HHi, if I make an R2 bucket public, can the files in the R2 bucket be seen, or are files only available to clients if they know the file name?
ETheres no listing functionality for public, its only if they know the filename
U(You still shouldnt put private files on a public url even if the file name is probably unguessable though 
)
)
Si mean you could if you enable access for it i guess
Uenabling access means it isnt a public url
H
H
Hbut i dont have any cors policy
CJust use
1_2_3.example.comEYes
AHi,
I'm wondering if I can use presigned url to do streaming upload ?
I tried to use the streaming upload:
it always returns http 411.
souds like it doesn't support ?
I'm wondering if I can use presigned url to do streaming upload ?
I tried to use the streaming upload:
it always returns http 411.
souds like it doesn't support ?
ENot too long - whats the domain?
EI see a vague error on
surveys.helpdesk.support.rydercragie.com - maybe its temporary? Could you try removing and readding the domain to nudge it again?
EMine too somedays lol
EI think that's mainly forr the bucket connection which shows as Active. The problem is the certificate which requires kicking by redoing it
EIf that still doesn't work I might have to try and find out what the error is trying to tell me, I've had no leads so far
EIt shows in
deleted state for me now, try adding it againElemme check
EGah, the same error
EOkay, well thanks for bearing with me at least
CWhat is your redirect rule? I think r2 uses http cert verification, and it looks like the redirect rule is applying to .well-known
curl http://surveys.helpdesk.support.rydercragie.com/.well-known/acme-challenge/example -vvv
< HTTP/1.1 301 Moved Permanently
< Date: Tue, 18 Jul 2023 17:23:47 GMT
< Transfer-Encoding: chunked
< Connection: keep-alive
< Location: https://surveys.helpdesk.support.rydercragie.com/index.html
< CF-Ray: 7e8c6a202e852419-IAD
curl http://surveys.helpdesk.support.rydercragie.com/.well-known/acme-challenge/example -vvv
< HTTP/1.1 301 Moved Permanently
< Date: Tue, 18 Jul 2023 17:23:47 GMT
< Transfer-Encoding: chunked
< Connection: keep-alive
< Location: https://surveys.helpdesk.support.rydercragie.com/index.html
< CF-Ray: 7e8c6a202e852419-IAD
EGood catch Chaika, yes that would get in the way
EPerhaps I might have found that if the CA had given me an error that made any sense 
E
Uyou have a rewrite rule there which acts as a redirect basically
CYou probably just want an extra condition there, and
Path equals /I'd just add a check that the url path does not begin with /.well-known
