You know how uploads.scratch.mit.edu serves as a mirror for scratchr2 pages. It works when the main website is under maintenance mode. Is there anything like it for scratch-www pages? uploads simply gives out an error for scratch-www pages.

What happens when you try to view the front page with uploads.scratch.mit.edu
(posted with uploads.scratch.mit.edu)

The version for scratch-www pages is scratch.mit.edu

scratch-www is running on scratch.mit.edu, and when a page there does not exist it queries uploads.scratch.mit.edu

Naleksuh wrote:

The version for scratch-www pages is scratch.mit.edu

scratch-www is running on scratch.mit.edu, and when a page there does not exist it queries uploads.scratch.mit.edu

i don't think that's how it works.


they've got some kind of config at fastly or their servers, which sends all traffic to scratchr2 except for paths that exist as scratch-www pages.

uploads is for pictures and stuff, and i guess it's meant to be faster at serving them.

Jeffalo wrote:

Naleksuh wrote:

The version for scratch-www pages is scratch.mit.edu

scratch-www is running on scratch.mit.edu, and when a page there does not exist it queries uploads.scratch.mit.edu

i don't think that's how it works.


they've got some kind of config at fastly or their servers, which sends all traffic to scratchr2 except for paths that exist as scratch-www pages.

One of the startup options for scratch-www is “FALLBACK” which is described as “Pass-through location for old site”. You can set this to "https://uploads.scratch.mit.edu“ to have all scratchr2 pages work. Not setting them will cause all pages not on scratch-www to return something like ”Cannot GET /discuss" (as is the standard express error).

Naleksuh wrote:

Jeffalo wrote:

Naleksuh wrote:

The version for scratch-www pages is scratch.mit.edu

scratch-www is running on scratch.mit.edu, and when a page there does not exist it queries uploads.scratch.mit.edu

i don't think that's how it works.


they've got some kind of config at fastly or their servers, which sends all traffic to scratchr2 except for paths that exist as scratch-www pages.

One of the startup options for scratch-www is “FALLBACK” which is described as “Pass-through location for old site”. You can set this to "https://uploads.scratch.mit.edu“ to have all scratchr2 pages work. Not setting them will cause all pages not on scratch-www to return something like ”Cannot GET /discuss" (as is the standard express error).

yeah but thats a hack we can do to get it to work. i might be wrong, but i doubt they use that in their production (or even development/staging) systems.

Jeffalo wrote:

Naleksuh wrote:

Jeffalo wrote:

Naleksuh wrote:

The version for scratch-www pages is scratch.mit.edu

scratch-www is running on scratch.mit.edu, and when a page there does not exist it queries uploads.scratch.mit.edu

i don't think that's how it works.


they've got some kind of config at fastly or their servers, which sends all traffic to scratchr2 except for paths that exist as scratch-www pages.

One of the startup options for scratch-www is “FALLBACK” which is described as “Pass-through location for old site”. You can set this to "https://uploads.scratch.mit.edu“ to have all scratchr2 pages work. Not setting them will cause all pages not on scratch-www to return something like ”Cannot GET /discuss" (as is the standard express error).

yeah but thats a hack we can do to get it to work. i might be wrong, but i doubt they use that in their production (or even development/staging) systems.

I do know that when scratch-www was first in development and scratchr2 was on scratch.mit.edu , scratch-www pages was scratchr2 querying staging.scratch.mit.edu (with the API being at api-staging.scratch.mit.edu).

Now that api-staging.scratch.mit.edu is now at api.scratch.mit.edu, it is not unreasonable to assume the main domain has been reversed as well, with the scratch-www at scratch.mit.edu querying another domain.

As for why the scratchr2 domain would be called “uploads”….I'm not sure. It could be a dual purpose.

If it is being rerouted from Fastly then I'd like to know: 1) What the purpose of FALLBACK is 2) What the deal is with URLs like https://scratch.mit.edu/community_guidelinesasdfoi signing users out (this would be more likely to happen if it was proxying from scratch-www itself and not through Fastly)