porter-agent

README

Porter In-Cluster Agent

This repository contains the source code for an in-cluster agent that aggregates cluster events and forwards them to a Porter server. This agent forwards three types of events:

• Pod-level events, for example if a Pod terminates due to an application-level error. Ideally, the agent is able to keep a buffer of the previous 100 log lines from stderr and stdout for each Pod, and will bundle the termination event with those logs, so that application developers can view those logs later on.
• HPA-level events, which shows reasons for autoscaling. DevOps engineers (and sometimes application developers) should be able to view the reasons for horizontal pod autoscaling triggers, such as CPU/memory/custom metric.
• Node-level events, in particular when a node is unhealthy.

This agent forms the basis for an events tab on the Porter dashboard, along with notifications for users when deployments/apps scale, restart, or when machines terminate.