Allow large form posts via multipart encoded forms to command API

[agjohnson]

This temporarily puts this under a feature flag, to make debugging a
little easier, but initial tests show things working as expected. This
allows for a large form post to be chunked up and POST, where we were
noticing a number of different issues with body size before.

Additionally, we can remove the chunked encoding PR as this was not the ultimate fix that worked.

Hotfixed on readthedocsinc, not on community version.

[codecov]

Codecov Report

❗ No coverage uploaded for pull request base (master@59d9e3e). Click here to learn what that means.
The diff coverage is 100%.

@@            Coverage Diff            @@
##             master    #5000   +/-   ##
=========================================
  Coverage          ?   76.72%           
=========================================
  Files             ?      158           
  Lines             ?     9962           
  Branches          ?     1246           
=========================================
  Hits              ?     7643           
  Misses            ?     1984           
  Partials          ?      335

Impacted Files	Coverage Δ
readthedocs/restapi/views/model_views.py	95.18% <100%> (ø)
readthedocs/projects/models.py	85.58% <100%> (ø)
readthedocs/doc_builder/environments.py	85.27% <100%> (ø)

[agjohnson]

Also, this might be able to be implemented as a slumber serializer so we're not forking the code more here.

Docs at: https://slumber.readthedocs.io/en/v0.6.0/options.html#serializer

Perhaps we can clean up this hotfix PR with a follow up PR and a better serializer.

[humitos]

Also, this might be able to be implemented as a slumber serializer so we're not forking the code more here.

I like this idea. I'd wait some weeks to see if this solution solves most of these problems and in that case, I can write the serializer. I think it's the way to go.

[humitos]

Looks good!

I've triggered my project at corporate with huge build output and it built properly.

One problem that I noticed is that none of the sphinx-build commands were logged in the build output: https://readthedocs.com/projects/read-the-docs-test-builds/builds/168252/

I'm not sure if it's related to this, but it wasn't happening before this PR.

[humitos]

On the other hand, for a normal build without huge output, it did log those commands: https://readthedocs.com/projects/read-the-docs-test-builds/builds/168251/

[agjohnson]

Strange, i missed this in my testing, i don't see a build that has those steps. I might have solved the request issue but opened a new bug with this. I do think this is at least part of the solution and agree it probably makes sense to let this simmer and figure out if writing a serializer would further help 👍

I'd say we should merge this if it looks okay, as it is feature flagged anyways, and track down why the large command posts are not showing up in a separate issue

[humitos]

I might have solved the request issue but opened a new bug with this

What is strange to me is that all of the requests are done with the MultipartEncoder but only large ones are not logged :/

I'd say we should merge this if it looks okay, as it is feature flagged anyways, and track down why the large command posts are not showing up in a separate issue

Agree.

I'm opening two issues to track them: #5021 and #5022

[ericholscher]

Is this actually hotfixed? If so, we have a conflict, which seems bad. We should get it merged ASAP.

[agjohnson]

This is live on .com yes, there was a bug causing builds to fail with a generic error.