Inspiration
Where I've worked for companies as an intern, I have done some work in DevOps and resolving incidents. There I have noticed that incident responses are often slow because internal software is not good enough to help engineers identify and resolve the root cause. There is too much manual overhead that takes time and wastes money for a company and its clients. We can't remove all the manual work yet but we can use cloud Site Reliability Engineer agent to alleviate the load.
What it does
It tracks monitored applications for error logs and 5xx errors. Errors trigger the Pub/Sub bus to alert the Cloud Run container which runs the backend service. The service calls Gemini to triage the incident to one of four levels (Critical, High, Medium or Low), analyse the root cause based on the error logs, then post a summary to the relevant Slack channel necessary depending on the triage level. This all happens automatically as the application runs completely
What we learned
It was my first experience with Google Cloud and I learned great cloud development concepts like IAM, alerting, SRE, logging, subscriptions making an app that works fully in the Cloud and independent of my laptop. Also gained a bit more confidence with containerisation and Docker.
Built With
- antigravity
- cloud
- fastapi
- gcp
- python
Log in or sign up for Devpost to join the conversation.