Add to Repo Docs: alternative way to configure Repository Storage using AWS S3 bucket

[stephenakearns]

Suggested Menu navigation/location: (feel free to change)

Repository >> Installation Procedures >> Advanced Topics >> Configure Cloud Storage using S3

Proposed Warning to include:

NOTE: this is Beta functionality. This deployment configuration should only be performed in consultation with a Continuum Implementation Engineer.

OVERVIEW

There are 4 steps to this process

1. Add S3 Bucket
2. Add credentials to Repo Server
3. Configure Storage on Repo Server
4. Validate storage is working correctly

ADD NEW S3 BUCKET

open http://console.aws.com # …. And create a new bucket my-bucket - naming for bucket has to be globally unique

CONFIGURE REPOSITORY - ADD CREDENTIALS

vim .boto # Add ec2 creds to .boto file - on the server in the home directory of binstar user where Anaconda Repo is running, if doesn't exist, create it (BRAD - what is the best practice.

Format of Boto file:
[Credentials] aws_access_key_id = <your_access_key_here> aws_secret_access_key = <your_secret_key_here>

Reference for format of .boto file.

CONFIGURE REPOSITORY - SET STORAGE LOCATION

   anaconda-server-config —set storage_type s3    anaconda-server-config --set PACKAGE_BUCKET_ID my-bucket

(these are case sensitive parameters)

VALIDATE CONFIGURATION

To validate that this is working correctly, perform a conda package upload and validate that you can browse in the WebUI to the newly uploaded file.

NOT FOR DOCS... INTERNAL Reference to related discussion in flowdock: FYI https://www.flowdock.com/app/continuum/anacondaserver/threads/JAeOcAgNnniJKoGJDDtgv31r1bh>

[stephenakearns]

cc: @kerrywatson1 (as discussed earlier this week).
cc: @hsparra - LMK if the docs for your client change at all after they go thru this process
cc: @srossross - FYI - based on the earlier meeting, please chime in with any edits.