Identify improvements to our onboarding docs

[choldgraf]

Summary

Now that we've onboarded @sgibson91 in #116 , this is a good opportunity to identify any updates that are needed to the process. Here are a few suggested steps we could take:

Actions

• Look for comments that showed any accounts we needed to provide access to but weren't in the issue template
• Look for confusion around "how to do step XXX" and document it
• Create issues for things to finish in the future
• Finish up any documentation changes that we can make quickly

Issues opened

• Create a "sandbox project" on AWS #102
• Create a sandbox account on GCP #126
• Document tools required to work in this repo and deploy hubs successfully infrastructure#476
• Document how to practice deploys #127
• Document how and when to add a cluster #128
• Document how to add somebody to our Google Workspace account #129
• Document how to access / use the 2i2c Twitter handle #130
• Document how to provide GCP access #133

[sgibson91]

Some ideas from me

• The Google group to be added to in order to see all the GCP projects was gcp-organization-admins, rather than gcp-billing-admins. Thanks to @yuvipanda for debugging that one!
• I'd like a "How to do practice deploys" reference. Is there a "safe space" project I can deploy to where it doesn't matter if I mess up? (i.e. separated from production hubs)
• There's a "how to add a hub" guide, but no "how to add a cluster" guide, which should be accompanied with a "when it's appropriate to add a new cluster" discussion as I imagine we'd like to keep clusters to a minimum but sometimes it's needed. E.g. I'm wondering if Pangeo will involve setting up a new cluster? If so, that'll be a good opportunity to draft the guide
  • Could also be accompanied by a "new cluster" issue template/checklist in pilot-hubs repo
• I also opened Document tools required to work in this repo and deploy hubs successfully infrastructure#476 which is relevant to this discussion

[damianavila]

I'd like a "How to do practice deploys" reference. Is there a "safe space" project I can deploy to where it doesn't matter if I mess up? (i.e. separated from production hubs)

There is a related issue where I asked about some 2i2c AWS playground: #102

There's a "how to add a hub" guide, but no "how to add a cluster" guide, which should be accompanied with a "when it's appropriate to add a new cluster" discussion as I imagine we'd like to keep clusters to a minimum but sometimes it's needed. E.g. I'm wondering if Pangeo will involve setting up a new cluster? If so, that'll be a good opportunity to draft the guide

Both are great points!
Btw, a preexisting issue that may contain other points to improve as well: 2i2c-org/infrastructure#351

[sgibson91]

There is a related issue where I asked about some 2i2c AWS playground: #102

Ah cool! I think a similar project in GCP would be good as well, since it's 2i2c's preferred platform :)

[choldgraf]

For "sandboxes" - is that just a matter of creating a project that is separate from our pilot hubs and then documenting these projects as safe spaces? That sounds like a good idea to me as well.

[sgibson91]

I think so, yes. I think the local deployment instructions detail how to just update one hub on a specific cluster, so point to those docs and then that should be a totally separate deployment.

[choldgraf]

Cool - see my latest comment in #102 , would anybody be willing to help me set up a project on AWS connected to my CC that we can use to run the infrastructure until we find a longer-term way to pay for it?

[choldgraf]

OK, I took a pass through the thread in #116 and created issues from the major points there (see top comment). There was one thing that I think we could do quickly (see below) but other than that I think we can close this one out unless there's something major that I'm missing.

hey @sgibson91 - you mentioned this:

The Google group to be added to in order to see all the GCP projects was gcp-organization-admins, rather than gcp-billing-admins.

but I'm having a hard time figuring out where the documentation was that suggested gcp-billing-admins. Could you point me there so that we can change this?

[sgibson91]

@choldgraf I did a quick, org-wide search and the only instance I can find is in the org-ops repo which mentions needing access to gcp-billing-admins to use that repo. So I suggest that it's actually not documented. We worked this out on Slack.

[choldgraf]

ahhh gotcha thanks :-)

[choldgraf]

OK I've created an issue for documenting how to provide GCP access: #133

can we close this one and follow up in the new issues?

[choldgraf]

closing this issue since I think it's addressed, feel free to re-open if there's something else to discuss! 🎉