Issues following the docs for deploying feast on GCP using Terraform

[josegpg]

Hello!

I'm trying to test feast but I've found few issues when trying to deploy the platform on GCP through Terraform following the docs. I'll list them below.

1. Helm fails to find the feast chart (since it was changed to a different repository. You can see here a reference to the old path.
2. Once I manually changed variable to point to the right location of the chart, however on the last terraform step (helm release) it's giving me timeout. When doing a small research I found that there's a Pod (feast-serving) entering a CrashLoopBackOff state, when looking at the logs I found the following:

Factory method 'servingServiceV2' threw exception; nested exception is io.lettuce.core.RedisConnectionException: Unable to connect to feast-0-8-redis-master:6379

Doing some more research I found that the host should not be feast-0-8-redis-master since it should be overwritten here and that's not happening since the name of the variables to overwrite (found here) is feast-online-serving instead of feast-serving (see here).

Expected Behavior

The terraform apply works completely.

Current Behavior

Failing on the last step (helm release)

Steps to reproduce

Basically follow this documentation from scratch.

Specifications

• Version: 0.9.3
• Platform: GKE

Possible Solution

I think the solution comes in 3 steps:

1. Update docs to reflect the new feast helm repo structure
2. Update terraform file to point to the new feast helm repo
3. Update overwrite vars with the proper chart name (feast-serving) so the redis ip is properly set and feast-serving pod can connect to redis.

I've tested (2 and 3) locally and finally got feast working on my GCP account. I'll create a sample fix PR since it's not too complicated

[woop]

Hey @josegpg. Thanks for the update and thanks for figuring out the problem! I've left a comment on your PR