How to disable range requests?

[wintercounter]

I'm using the default docker image. I recently moved from Cloudflare proxy to Bunny.net CDN and seems like they don't support range requests, and they are caching incorrectly because of that. I need to disable it completely, but it seems like there is no option for that.

UPDATE:
Bunny support said my server is not responding correctly to the range requests. So I wonder what could be the issue here. They are requesting 5MB chunks. I'm using default configuration.

The header they are sending looks like content-range: bytes */25314. Is it possible that * is not supported?

UPDATE 2:
I switched to an nginx based container for now, works fine. I was using latest v2 by the way.

[joseluisq]

The header they are sending looks like content-range: bytes */25314. Is it possible that * is not supported?

As far as I understood, content-range: bytes */25314 means a 416 Range Not Satisfiable status response.
So you have to check how are you requesting byte ranges and if they are valid.
However since you have not shown more info I can only guess.

It would be useful some tracing logs of SWS requests and responses.

[phartenfeller]

I ran into a similar issue where LinkedIn does not show any previews for my blog posts. They have this debug site where you can check the previews: https://www.linkedin.com/post-inspector (but it needs a login).

I tracked down the request in my Traefik logs, and the problem is this header: Range: bytes=0-3145727.

static-web-server returns 416:

 curl -o - -I 'http://localhost/blog/oracle-apex-text-messages/' \
   -H 'Range: bytes=0-3145727'
HTTP/1.1 416 Range Not Satisfiable
content-range: bytes */62519
vary: accept-encoding
cache-control: public, max-age=86400
date: Thu, 06 Jul 2023 11:27:03 GMT

This is the trace of static-web-server:

2023-07-06T10:56:57.504069Z TRACE hyper::proto::h1::conn: Conn::read_head
2023-07-06T10:56:57.504131Z TRACE hyper::proto::h1::io: received 3670 bytes
2023-07-06T10:56:57.504220Z TRACE parse_headers: hyper::proto::h1::role: Request.parse bytes=3670
2023-07-06T10:56:57.504272Z TRACE parse_headers: hyper::proto::h1::role: Request.parse Complete(3670)
2023-07-06T10:56:57.504371Z TRACE parse_headers: hyper::proto::h1::role: close time.busy=162µs time.idle=3.64µs
2023-07-06T10:56:57.504412Z DEBUG hyper::proto::h1::io: parsed 25 headers
2023-07-06T10:56:57.504427Z DEBUG hyper::proto::h1::conn: incoming body is empty
2023-07-06T10:56:57.504487Z  INFO static_web_server::handler: incoming request: method=GET uri=/blog/oracle-apex-text-messages/
2023-07-06T10:56:57.504571Z TRACE static_web_server::static_files: dir: base="/public", route="blog/oracle-apex-text-messages/"
2023-07-06T10:56:57.504604Z TRACE static_web_server::static_files: getting metadata for file /public/blog/oracle-apex-text-messages
2023-07-06T10:56:57.504641Z TRACE static_web_server::static_files: file found: "/public/blog/oracle-apex-text-messages"
2023-07-06T10:56:57.504671Z DEBUG static_web_server::static_files: dir: appending an index.html to the directory path
2023-07-06T10:56:57.504725Z TRACE static_web_server::static_files: file found: "/public/blog/oracle-apex-text-messages/index.html"
-----
This line is interesting
-----
2023-07-06T10:56:57.504813Z TRACE static_web_server::static_files: unsatisfiable byte range: 0-3145728/62263
-----
2023-07-06T10:56:57.504981Z TRACE encode_headers: hyper::proto::h1::role: Server::encode status=416, body=None, req_method=Some(GET)
2023-07-06T10:56:57.505031Z TRACE encode_headers: hyper::proto::h1::role: close time.busy=50.1µs time.idle=2.46µs
2023-07-06T10:56:57.505224Z DEBUG hyper::proto::h1::io: flushed 124 bytes
2023-07-06T10:56:57.505258Z TRACE hyper::proto::h1::conn: flushed({role=server}): State { reading: Init, writing: Init, keep_alive: Idle }

Looking at MDN the Range header desc the request seems to be right as it matches Range: <unit>=<range-start>-<range-end>. I don't really understand the HTTP 416 description, as it mentions both the Range and Content-Range headers. Note that there is no Content-Range header present in the LinkedIn or my test curl request.

I validated this error against a local instance without Traefik. I pulled the latest Docker image, so I guess I am using v2.19.0.

[joseluisq]

Those particular issues were because SWS is strict when it comes to rage-end exceeding the max length of a file as reported by #295. So since Rage requests out of bounds is a common behavior, SWS will support it soon. See #306.

[joseluisq]

The range request limitation should be resolved now by v2.26.0. Feel free to try it out.

[joseluisq]

Just to summarize.
It is no possible to disable range requests. However, SWS now accepts Range requests out of bounds (as shown in the description of the discussion) since v2.26.0.

Closing for now. If you have questions or experience other issues please feel free to create another discussion.