feat(aws): configure basic auth for bindle

[vdice]

• Adds basic auth to the Bindle server

Ref #3

[vdice]

@bacongobbler with the current commit, services deploy correctly but I'm unable to deploy an app. Looking at the bindle logs, I see auth errors like the following, presumably from hippo requests:

2022-06-28T20:48:40.830445Z DEBUG bindle::server::filters: Authentication error error=Invalid byte 58, offset 5.
2022-06-28T20:48:40.830464Z DEBUG bindle::server::filters: Handling rejection as authn rejection
2022-06-28T20:48:40.830468Z DEBUG bindle::server::reply: Parsed accept header into list accept_value=application/toml accept_items=["application/toml"]
2022-06-28T20:48:40.830473Z DEBUG bindle::server::reply: Selected a best-fit MIME best_fit=application/toml

The deploy itself fails like so:

$ spin deploy
Error: Unable to create Hippo app

Caused by:
    {"type":"https://tools.ietf.org/html/rfc7231#section-6.6.1","title":"An error occurred while processing your request.","status":500}

I'm not seeing any logs in Hippo, even with Logging__LogLevel__Default = "Debug" set in its job env...

I wonder if the first thing we can double-check is that Hippo is passing along the username and password correctly?

If easier/more convenient to test using the local installer, you can try this wip branch instead: https://github.com/vdice/fermyon-installer/tree/wip/local-bindle-basic-auth

[bacongobbler]

This is probably because Hippo does not propagate bindle credential auth down to the nomad job. Hippo can fetch information from bindle, but the nomad job running Spin cannot pull the bindle because it doesn't have the username/password credentials.

[bacongobbler]

https://github.com/deislabs/hippo/blob/843f7da45edd3acbab95007aa89ba90ba2155a9a/src/Infrastructure/Services/NomadJobService.cs#L179

[vdice]

Thanks @bacongobbler -- I'll have to double-check but I think bindle-server v0.8.0 running with basic auth actually still allows anonymous GETs, so Spin should be able to fetch bindles as usual. I'll need to double-check this. But I can confirm that the Nomad job representing the Spin app hasn't yet been scheduled on Nomad when I hit this behavior; so I think we're encountering issues prior to this.

[bacongobbler]

Hmm, okay. I’ll see if I can reproduce the issue tomorrow. Thanks for looking into this.