Deployment timeout after 10 minutes

Hey,

First many thanks for maintaining this plugin.

Last couple of days we hit some weird errors with deployment. Looks like:

> gatsby-plugin-s3 deploy --yes

- Retrieving bucket info...

Too long with no output (exceeded 10m0s): context deadline exceeded

It times out after 10m. Not sure what times out though. Admittedly we have a lot of pages to upload on each build (35k right now and growing) and the build is running on circleci. Sometimes it finishes after 3 minutes, sometimes after 7 i guess this is circleci related.

I d really like to get a better understanding of this 10 minute timeout and see whether we can increase it.

[JoshuaWalsh]

Ora has actually been causing issues for my testing suite work too. If I make the test suite pass-through console output from the deploy process then the spinner animation causes hundreds of lines to be logged.

It’s worth investigating what Gatsby uses. Another option would be to write a wrapper around Ora and replace it with console.log if we are running in a CI environment.

[jariz]

Yea, I think ora just stops logging to stdout if the terminal doesn’t support updates. Basically, it’s still doing stuff, but CircleCI thinks it’s not.

10 minutes is a long time for the upload to take, but I’ve seen this happen on pretty large (10k+ pages) sites, so it’s not that uncommon. You mention it sometimes does finish, so it’s clearly not stuck or anything.

On my own CI, the maximum execution time can be increased. Is this not possible for CircleCI?

I wonder what kind of library the gatsby CLI itself uses for the progressbars, because the build output in my CI is a lot cleaner there. (it logs after the ending of each phase)