The issue seems to be over here, unencrypted HTTP2 connections seem to be broken. You can try with

curl --http2 http://image.tmdb.org/t/p/original/kXEERWLX5QxxtBJKWkMFT1sp75Q.jpg

Cloudflare reported two issues within the last 24 hours:

• A problem in LAX
• A problem issuing new challenges (which could affect the image service)

Other than that I am not seeing any issues identified by our external monitoring, or our internal logging.

Is it still happening?

Scraping artwork with Kodi is stil not working and broke about 24 to 36 hours ago.

Is there a body being returned on a request that is failing? Cloudflare might be returning a useful error.

There's been no other reported issues so this seems like something unique to Kodi.

The error we see in curl and in kodi is (sorry, forgot to mention): ``` (56) Recv failure: Connection reset by peer

So no, no body, no error message.

I am pretty sure the same issue happens in all applications that use libcurl (which are many) and activate http2 by default.

Sounds like an issue with Cloudflare's HTTP/2 then. I'll file a ticket and let you know when I hear back.

Had the problem here (UK - using Kodi on a Firestick). Did a full reinstall etc (should have checked here first - DOH!) Anyhoo - swapped to using the "The Move Database Python" scraper and that's working perfectly for me. Hope that helps, even if it's only temporary.

Alright, well I have our answer as to what changed. Cloudflare has started rejecting HTTP/2 requests over non TLS (plain HTTP) requests. On the surface this might seem like kind of a strange decision but it's not completely unfounded. It turns out there isn't a single popular web browser that has implemented HTTP/2 without TLS. They all banded together and decided that HTTP/2 will be used with TLS only. So while it's not officially part of the spec, it's how the market has decided to operate itself.

So, @KarellenX, we have 2 easy-ish options:

1. You don't use HTTP/2 if you aren't using SSL. Unencrypted requests should go over HTTP 1.1 only.
2. You make sure to just use TLS for every request, and then you can use HTTP/2.

I'd lean towards options #2 since you'll get the benefits of HTTP/2 and the security of having all requests be encrypted.

Thank you @travisbell

Let me pass this onto pkscoout, olympia and other team members and we can discuss.

We'll be in touch if we have more questions.

@pengipete said:

Had the problem here (UK - using Kodi on a Firestick). Did a full reinstall etc (should have checked here first - DOH!) Anyhoo - swapped to using the "The Move Database Python" scraper and that's working perfectly for me. Hope that helps, even if it's only temporary.

While waiting the issue to be solve, how do you set the Python version as default one? The normal one , default as it is right now, seems that cannot be disable. Many thanks

Go into Kodi settings, got to MEDIA>Videos. Right click (or equivalent) on your Movies" folder and select "Change Content". Click on "Choose information provider" and when the "Movie Information" window opens, select "Get More..." then scroll down and select the scraper you want. Click any "Okay" window to exit and allow Kodi to update the data held for your movies.

FWIW - if you're also using TMDB for TV Shows, I find that The TV Database works better anyway so do the same as above to select TVDB if necessary.

Ya, it's kind of weird Cloudflare has decided to just terminate the request. If you look at what CloudFront is doing (where we host the website and API), they just downgrade the request to HTTP 1.1:

curl --http2 -v http://www.themoviedb.org
*   Trying 99.84.238.85:80...
* Connected to www.themoviedb.org (99.84.238.85) port 80 (#0)
> GET / HTTP/1.1
> Host: www.themoviedb.org
> User-Agent: curl/7.72.0
> Accept: */*
> Connection: Upgrade, HTTP2-Settings
> Upgrade: h2c
> HTTP2-Settings: AAMAAABkAAQCAAAAAAIAAAAA
>
* Mark bundle as not supporting multiuse
< HTTP/1.1 301 Moved Permanently
< Server: CloudFront
< Date: Fri, 16 Oct 2020 15:28:34 GMT
< Content-Type: text/html
< Content-Length: 183
< Connection: close
< Location: https://www.themoviedb.org/
< X-Cache: Redirect from cloudfront
< Via: 1.1 f33529eebdae9f360e1e83d3ee6348f6.cloudfront.net (CloudFront)
< X-Amz-Cf-Pop: SFO5-C3
< X-Amz-Cf-Id: STMNDBsCnaNsQ8in2Y2YBN27molGUadTlakWPR_Ty0RE71l9CyvYuQ==

The image service is the only thing running on Cloudflare, which is why the issue only shows up there.

thanks @pengipete !

@travisbell is it something what will be eventually fixed ? Thanks for all explanations

@FOX_ITA Based on the reply from @KarellenX, it sounds like you can expect to see an update to the scraper at some point. There isn't anything we can do on our end.

Seems to be working again. Thanks folks :)