r2 not working in safari. cannot access links. does anyone have a solution to this situation?
r2 not working in safari. cannot access links. does anyone have a solution to this situation?
CThat's really vague, what's not working? The dashboard, the r2.dev public bucket, a public bucket on a custom domain? What error?
KI am sorry. I'm posting a video stream. https://cdn.anizium.com/demon-slayer-1-1-2160-jp.mp4 is my sample link. It works fine in all browsers, but I get an error on iOS devices.
KI read in a few places that it might be related to cors policy.
CIf you're going to the mp4 directly and not trying to load it on an external site it's not related, it doesn't load for me in Firefox either, looks like the mp4 is somehow corrupt or just unreadable by mozilla's mp4 as well
EIts 1.47 GB and cacheable, I wonder if its the cache bug mentioned a couple times in this channel
Etry bypassing cache for that route
Kbut it works for me and it's ok. I don't think this has anything to do with caching. I couldn't find the problem either.
UIm assuming safari/firefox do range requests on mp4 files which would make it a cache issue
Ein chromium it works but the range requests dont, I suspect^ the other browsers are relying on this behaviour
KCan you try from chrome? I'm working too. I do not think that there will be a problem due to the file or the installation. If you wish, I can download bunny.mp4 and try it.
EIts because the file is too big to be cacheable, please bypass cache for it
Khow do i bypass the cache?
EUse Cache Rules to customize cache properties of your HTTP requests. For example, create a rule to specify how long to cache a resource in the …
KWe can find out about the problem here. I'll skip the cache by the way, that's on my mind. https://cdn.anizium.com/Big_Buck_Bunny_1080_10s_30MB.mp4 there is a small image in this link. i am getting the same error.
EThat second link works for me on both Firefox and Safari iOS without any errors
Ethis is because its small and fits within the cache
Eas seen by
< CF-Cache-Status: HITyes you were right. thanks a lot. but this page is very detailed... and I don't know that much.. can I get some more help?
VMy tests using
unstable_dev()24ms
Sdoes caching occur if the contents of R2 are accessed through a worker with a route?
KNope
KYou can add the Cache API yourself but you'd still be billed for the Worker invocation
Sah ok
KI am writing this with my apologies as I have no knowledge of it. what i want is to turn off cache and play high gb files in r2. but I don't understand what to do here. Do I connect with api or in the form of a subdomain?
CUnder incoming requests match, Field:
Cache Status: Bypass
Give it some name and save
Hostnameequalscdn.anizium.comCache Status: Bypass
Give it some name and save
CYou could make a more complex rule, perhaps only bypassing for a specific subfolder (i.e Path Starts With /videos) or even specific files with a full path match, so that you only bypass cache for large videos. Although if everything you have is large videos it won't matter
KThank you. but my problem still persists. it does not open or stream large videos in any way.
example:https://cdn.anizium.com/demon-slayer-1-1-2160-jp.mp4 its not working safari and firefox.
example:https://cdn.anizium.com/demon-slayer-1-1-2160-jp.mp4 its not working safari and firefox.
Kbut there is no problem in chrome.
Ki think
content-type: video/mp4your linkexpected (working one)So, yes, it does expose my bucket name and the account ID. Is that safe? Bucket name doesn't seem to be a big deal as it's same as my domain. But is ACCOUNT_ID safe to be exposed?
Yes, it's fine.
Note that I do have Custom Domains setup on this bucket. Can I just replace the BUCKET_NAME.ACCOUNT_ID.r2.cloudflarestorage.com part of the presigned url with my custom domain? Would that still validate the signature?
Nope, custom domains don't implement the S3 API.
ACan you get an object's specific version?
UThere’s no versions
AObjects do have versions :)
AThe in-Worker R2 API is accessed by binding an R2 bucket to a Worker. The Worker you write can expose external access to buckets via a route or …
AAvailable under the
versionHe’s not wrong.
KIt’s a placeholder that isn’t recorded, it’s just there to support versioning in the future
AAgain, this is also wrong, versions change when you replace an existing file.
AI suppose I could mimic object versions using the Cache API though.
ZWhat do you want the version information for? Only the latest upload under a key is stored.
Kthat isn’t recorded
KThere is no storage, there is no history, there is no rollbacks and there is no log.
KOnce you upload a new version, the string changes and you can do nothing with the old one.
KThe Cache API is ephemeral, per-colo and can be evicted at any time so it isn’t going to work for versioning.
AA specific use case, I upload an set of objects, and an index for a particular release that may occassionally change. On request for the initial index, I set a version cookie so all next requests will get the respective file, in case between requests the file was changed, to avoid mismatches.
ACache only needs to be accurate within the next few requests.
ZThat's not possible with R2. Like kian pointed out, the moment a new object is uploaded, the old one is completely gone
AAnd so, as I said, the cache API will suffice, or I could just add an arbitrary version to an objects key.
