Add GCP dev instructions (#1720)

Author: deliahu

CONTRIBUTING.md

@@ -134,6 +134,8 @@ git diff  # there should be no diff
 
 ### Cluster configuration
 
+These instructions assume you'll be creating clusters on AWS and GCP. You may skip some of the steps and configuration if you'll only be developing / testing on a single cloud provider.
+
 Create a config directory in the repo's root directory:
 
 ```bash
@@ -146,9 +148,14 @@ Create `dev/config/env.sh` with the following information:
 # dev/config/env.sh
 
 export AWS_ACCOUNT_ID="***"  # you can find your account ID in the AWS web console; here is an example: 764403040417
-export AWS_REGION="us-west-2"  # you can use any AWS region you'd like
+export AWS_REGION="***"  # you can use any AWS region you'd like, e.g. "us-west-2"
 export AWS_ACCESS_KEY_ID="***"
 export AWS_SECRET_ACCESS_KEY="***"
+
+export GCP_PROJECT_ID="***"
+export GOOGLE_APPLICATION_CREDENTIALS="***"
+export GCR_HOST="gcr.io"  # must be "gcr.io", "us.gcr.io", "eu.gcr.io", or "asia.gcr.io"
+
 # export NUM_BUILD_PROCS=2  # optional; can be >2 if you have enough memory
 ```
 
@@ -158,41 +165,68 @@ Create the ECR registries:
 make registry-create-aws
 ```
 
-Create `dev/config/cluster-aws.yaml`. Paste the following config, and update `cortex_region` and all registry URLs (replace `XXXXXXXX` with your AWS account ID, and update the region):
+Create `dev/config/cluster-aws.yaml`. Paste the following config, and update `region` and all registry URLs (replace `<account_id>` with your AWS account ID, and replace `<region>` with your region):
 
 ```yaml
 # dev/config/cluster-aws.yaml
 
 cluster_name: cortex
 provider: aws
-region: us-west-2
+region: <region>  # e.g. us-west-2
 instance_type: m5.large
 min_instances: 1
 max_instances: 5
 
-image_operator: XXXXXXXX.dkr.ecr.us-west-2.amazonaws.com/cortexlabs/operator:latest
-image_manager: XXXXXXXX.dkr.ecr.us-west-2.amazonaws.com/cortexlabs/manager:latest
-image_downloader: XXXXXXXX.dkr.ecr.us-west-2.amazonaws.com/cortexlabs/downloader:latest
-image_request_monitor: XXXXXXXX.dkr.ecr.us-west-2.amazonaws.com/cortexlabs/request-monitor:latest
-image_cluster_autoscaler: XXXXXXXX.dkr.ecr.us-west-2.amazonaws.com/cortexlabs/cluster-autoscaler:latest
-image_metrics_server: XXXXXXXX.dkr.ecr.us-west-2.amazonaws.com/cortexlabs/metrics-server:latest
-image_inferentia: XXXXXXXX.dkr.ecr.us-west-2.amazonaws.com/cortexlabs/inferentia:latest
-image_neuron_rtd: XXXXXXXX.dkr.ecr.us-west-2.amazonaws.com/cortexlabs/neuron-rtd:latest
-image_nvidia: XXXXXXXX.dkr.ecr.us-west-2.amazonaws.com/cortexlabs/nvidia:latest
-image_fluentd: XXXXXXXX.dkr.ecr.us-west-2.amazonaws.com/cortexlabs/fluentd:latest
-image_statsd: XXXXXXXX.dkr.ecr.us-west-2.amazonaws.com/cortexlabs/statsd:latest
-image_istio_proxy: XXXXXXXX.dkr.ecr.us-west-2.amazonaws.com/cortexlabs/istio-proxy:latest
-image_istio_pilot: XXXXXXXX.dkr.ecr.us-west-2.amazonaws.com/cortexlabs/istio-pilot:latest
+image_operator: <account_id>.dkr.ecr.<region>.amazonaws.com/cortexlabs/operator:latest
+image_manager: <account_id>.dkr.ecr.<region>.amazonaws.com/cortexlabs/manager:latest
+image_downloader: <account_id>.dkr.ecr.<region>.amazonaws.com/cortexlabs/downloader:latest
+image_request_monitor: <account_id>.dkr.ecr.<region>.amazonaws.com/cortexlabs/request-monitor:latest
+image_cluster_autoscaler: <account_id>.dkr.ecr.<region>.amazonaws.com/cortexlabs/cluster-autoscaler:latest
+image_metrics_server: <account_id>.dkr.ecr.<region>.amazonaws.com/cortexlabs/metrics-server:latest
+image_inferentia: <account_id>.dkr.ecr.<region>.amazonaws.com/cortexlabs/inferentia:latest
+image_neuron_rtd: <account_id>.dkr.ecr.<region>.amazonaws.com/cortexlabs/neuron-rtd:latest
+image_nvidia: <account_id>.dkr.ecr.<region>.amazonaws.com/cortexlabs/nvidia:latest
+image_fluentd: <account_id>.dkr.ecr.<region>.amazonaws.com/cortexlabs/fluentd:latest
+image_statsd: <account_id>.dkr.ecr.<region>.amazonaws.com/cortexlabs/statsd:latest
+image_istio_proxy: <account_id>.dkr.ecr.<region>.amazonaws.com/cortexlabs/istio-proxy:latest
+image_istio_pilot: <account_id>.dkr.ecr.<region>.amazonaws.com/cortexlabs/istio-pilot:latest
+```
+
+Create `dev/config/cluster-gcp.yaml`. Paste the following config, and update `project`, `zone`, and all registry URLs (replace `<project_id>` with your project ID, and update `gcr.io` if you are using a different host):
+
+```yaml
+# dev/config/cluster-gcp.yaml
+
+project: <project_id>
+zone: <zone>  # e.g. us-central1-a
+cluster_name: cortex
+provider: gcp
+instance_type: n1-standard-2
+min_instances: 1
+max_instances: 5
+# accelerator_type: nvidia-tesla-k80  # optional
+
+image_operator: /cortexlabs/operator:latest
+image_manager: gcr.io/<project_id>/cortexlabs/manager:latest
+image_downloader: gcr.io/<project_id>/cortexlabs/downloader:latest
+image_istio_proxy: gcr.io/<project_id>/cortexlabs/istio-proxy:latest
+image_istio_pilot: gcr.io/<project_id>/cortexlabs/istio-pilot:latest
+image_google_pause: gcr.io/<project_id>/cortexlabs/google-pause:latest
 ```
 
 ### Building
 
-Add this to your bash profile (e.g. `~/.bash_profile`, `~/.profile` or `~/.bashrc`), replacing the image registry URL accordingly:
+Add this to your bash profile (e.g. `~/.bash_profile`, `~/.profile` or `~/.bashrc`), replacing the placeholders accordingly:
 
 ```bash
-export CORTEX_DEV_DEFAULT_PREDICTOR_IMAGE_REGISTRY_AWS="XXXXXXXX.dkr.ecr.us-west-2.amazonaws.com/cortexlabs"  # set the default image for APIs
-export CORTEX_TELEMETRY_SENTRY_DSN="https://c334df915c014ffa93f2076769e5b334@sentry.io/1848098"  # redirect analytics to our dev environment
-export CORTEX_TELEMETRY_SEGMENT_WRITE_KEY="0WvoJyCey9z1W2EW7rYTPJUMRYat46dl"  # redirect error reporting to our dev environment
+# set the default image for APIs
+export CORTEX_DEV_DEFAULT_PREDICTOR_IMAGE_REGISTRY_AWS="<account_id>.dkr.ecr.<region>.amazonaws.com/cortexlabs"
+export CORTEX_DEV_DEFAULT_PREDICTOR_IMAGE_REGISTRY_GCP="gcr.io/<project_id>/cortexlabs"
+export CORTEX_DEV_DEFAULT_PREDICTOR_IMAGE_REGISTRY="cortexlabs"
+
+# redirect analytics and error reporting to our dev environment
+export CORTEX_TELEMETRY_SENTRY_DSN="https://c334df915c014ffa93f2076769e5b334@sentry.io/1848098"
+export CORTEX_TELEMETRY_SEGMENT_WRITE_KEY="0WvoJyCey9z1W2EW7rYTPJUMRYat46dl"
 
 alias cortex='$HOME/bin/cortex'  # your path may be different depending on where you cloned the repo
 ```
@@ -213,12 +247,16 @@ cortex version  # should show "master"
 Build and push all Cortex images:
 
 ```bash
+# for AWS:
 make images-all-aws
+
+# for GCP:
+make images-all-gcp
 ```
 
 ## Dev workflow
 
-Here is the typical full dev workflow which covers most cases:
+Here is the typical full dev workflow which covers most cases (replace `aws` with `gcp` if desired):
 
 1. `make cluster-up-aws` (creates a cluster using `dev/config/cluster-aws.yaml`)
 2. `make devstart-aws` (deletes the in-cluster operator, builds the CLI, and starts the operator locally; file changes will trigger the CLI and operator to re-build)
@@ -230,7 +268,7 @@ Here is the typical full dev workflow which covers most cases:
 If you want to switch back to the in-cluster operator:
 
 1. `<ctrl+c>` to stop your local operator
-2. `make cluster-configure-aws` to install the operator in your cluster
+2. `make operator-start-aws` to restart the operator in your cluster
 
 If you only want to test Cortex's local environment, here is the common workflow:
 

docs/workloads/dependencies/docker-images.md

@@ -1,6 +1,6 @@
 # Docker images
 
-You can build a custom Docker image for use in your APIs, e.g. to avoid installing dependencies during replica initialization.
+You can build a custom Docker image for use in your APIs, e.g. to avoid installing dependencies during replica initialization or to have smaller images.
 
 ## Create a Dockerfile