https://github.com/slamdev/gcp-serverless
Serverless app on GCP
https://github.com/slamdev/gcp-serverless
Last synced: 11 days ago
JSON representation
Serverless app on GCP
- Host: GitHub
- URL: https://github.com/slamdev/gcp-serverless
- Owner: slamdev
- Created: 2019-12-27T15:19:34.000Z (about 6 years ago)
- Default Branch: master
- Last Pushed: 2023-12-15T14:58:10.000Z (about 2 years ago)
- Last Synced: 2025-06-14T19:09:11.507Z (8 months ago)
- Language: HCL
- Size: 1.05 MB
- Stars: 0
- Watchers: 2
- Forks: 0
- Open Issues: 27
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# GCP Serverless [](https://github.com/slamdev/gcp-serverless/actions?query=workflow%3Adeploy) [](https://dev.gcp-serverless.slamdev.net)
## Project AIM
Build a globally distributed application using serverless approach on Google Cloud platform.
### Result
- GCP **Cloud Firestore** is not globally distributed
* workaround: switch database to FaunaDB
- GCP **Global Load Balancer** doesn't support **Cloud Functions** and **Cloud Storage**
* workaround: deploy **Envoy Proxy** in each region and put GLB in front; proxy requests to functions\buckets to the corresponding region
## Landscape overview
## Development
### Prerequisites
Create projects for terraform state and a project per environment. [setup.sh](/etc/setup.sh) script covers all the setup.
#### Necessary tools
- [terraform](https://github.com/hashicorp/terraform)
- [terragrunt](https://github.com/gruntwork-io/terragrunt)
- [go](https://golang.org/doc/install)
- [gcloud](https://cloud.google.com/sdk/gcloud/)
macOS:
```shell script
brew install terraform terragrunt go
brew cask install google-cloud-sdk
```
### Build/Deploy
All the interaction with project is implemented via [Makefile](Makefile)
By default all make targets are executed for **dev** environment. It can be changed by providing **TF_VAR_env** variable, e.g.:
```shell script
TF_VAR_env=prod make plan
```
**make init**|**make plan**|**make apply**|**make destroy** executes corresponding command for all the components in parallel
**make init/{COMPONENT_NAME}**|**make plan/{COMPONENT_NAME}**|**make apply/{COMPONENT_NAME}**|**make destroy/{COMPONENT_NAME}** executes corresponding command for a specified component, e.g.:
```shell script
make plan/backend/functions-bucket
```
**make clean-local-state** target removes local state files from all the projects
**make lint** target does a static validation for all the terraform files (at the moment it validates only code style rules according to the terraform conventions)
**make format** formats all the terraform files according to the terraform conventions
**make unlock/{COMPONENT_NAME}** command unlocks the remote terraform state for a component, e.g.:
```shell script
make unlock/backend/functions-bucket LOCK_ID=123456789
```
#### Remark about **make init**
Terragrunt suppose to do auto-init so in most cases there is no need to invoke this command manually, but some times it fails to do it. There is an open issue about it: https://github.com/gruntwork-io/terragrunt/issues/388
#### Remark about **make clean-local-state**
If there was a deploy to dev, terraform will fail to deploy to prod, because it has a state from dev locally. So it will try to migrate a dev state to prod and this is not a desire behavior.
Subj make target will clean all the local state files so deploy to prod will succeed.
### CI/CD
Flow:
- if a user pushes to **master** branch then the pipeline will apply changes to the **DEV** environment (by executing `TF_VAR_env=DEV make apply`)
- if a user pushes to **production** branch then the pipeline will apply changes to the **PROD** environment (by executing `TF_VAR_env=PROD make apply`)
- if a user pushes to an arbitrary branch then the pipeline will:
* lint all the files in the project (by executing `make lint`)
* do **plan** on **dev** environment (by executing `TF_VAR_env=dev make plan`)