NoteGala is an Android application that allows people to create, edit, and categorize notes. The client is backed by a public GraphQL API that can readily accommodate new clients or integrations.
You can find GIFs of the app in action here.
The back end is a bit over-engineered for the current scale. This reflects the fact that the project is primarily an educational pursuit.
User devices submit requests alongside Google credentials to a load-balanced API. These CRUD requests are satisfied by a backing PostgreSQL instance, which stores the majority of the system data. In addition to typical SQL queries, postgres also handles full-text search for notes.
While handling a request, the API will generate logs with standard severity levels. These logs are published to Redis channels and then picked up by logT. At that point, logs can go to two places:
- Elasticsearch
- All logs will end up here.
- The email inbox of a developer
- All (but hopefully zero) error logs end up here.
Jenkins performs integration and deployment for the API. A multibranch pipeline runs its tests in an isolated Docker Compose environment. If they pass, the code becomes eligible for a PR into master. Should that also succeed, the code is retested and then deployed into production. It is important that, when modifying the API, you also update its semantic version; failure to do so will result in no deployment.
Kubernetes manages the life cycle of Docker containers and their secrets. Any of the configuration files that can be put into Git will go in the cluster directory.
Everything is AWS. Here's a nice list:
- Route 53
- Handles the DNS configuration for the project
- RDS
- Manages the PostgreSQL instance
- EC2 / EBS
- Provides node/storage resources for the cluster
- S3
- Stores kops information for the cluster
- ELB
- Provides classic load balancing for Kubernetes services
- Elasticsearch
- Houses Elasticsearch but also a pretty sweet Kibana instance
- ElastiCache
- Vehemently guards Redis
- SNS
- Sends email alerts containing log data