Start using proper logging on machine executors (node and orchestrator)

[konstan]

slipstream.util module already contains capabilities for switching the logging on and off.

• Make sure stdout/err from the forked CLIs are redirected to the log file.
• Logging should be externally configurable (level, format etc.).
• Remove redirection of the stdout/err to the log file when the executor is started via CLI.
• Ensure that connectors are logging into the log file.

Extra considerations:

• Introducing log rotation between SendingReports states can allow to send less to the server in each iteration of a mutable run. Server can concatenate incoming parts of the log. This allows to save on space on the server side.

Notes:

• Such proper logging should allow for easier restart of the machine executors.

Comments and additions are very welcomed.

[schaubl]

👍

[konstan]

Proposing for sprint.

It now became relevant in automated processing of the node executor logs with filebeat from ELK (in Simon's work).