api

README

Remote execution

Currently remote execution only applies to execution of modules in container. Theoretically it could also be used e.g. to execute modules on other machines via SCP/SSH.

Sequence diagram overview

The following sequence diagram illustrates the remote protocol:

sequenceDiagram
    client ->>+ executor: StartTaskRequest
    client ->> client: listen for executor events
    executor ->> executor: Start FS watcher
    executor ->> executor: Unmarshal task and execute
    
    loop stream task output & logs
        executor -->> client: TaskLog
        executor -->> client: TaskOutput
    end

    executor -->>- client: TaskResult - as soon as execution completes

Execution steps for container tasks

The overall execution looks like this:

1. create container network
2. create container
3. copy buildr binary, tools and content into container
   • if input mapping is configured, content is copied as it would be mapped
   • if no input mapping is configured complete repository is copied
4. start container with buildr as entrypoint (gRPC server)
5. connect to gRPC server
6. send StartTaskRequest specifying what to do - although it's called TaskRequest it could be any module
7. server starts a FS watcher to monitor modified files
8. buildr CLI listens for logs and output from executed task
9. as soon as the execution completes, modified files are collected into a .tar.s2 file and error message + location of the archive are sent as result to buildr CLI
10. CLI copies archive back and extracts modified files to output directory

Documentation

Index

• type ProtoMessage

Types

type ProtoMessage

type ProtoMessage interface {
	UnmarshalVT(dAtA []byte) error
	MarshalVT() (dAtA []byte, err error)
}