Runner StdOut Logging Configuration

[JohnYoungers]

I may have missed it but I couldn't find the answer to this question in the self-hosted runner documentation: what options are available as far as customizing how the runner logs to standard out?

So far in my research I've only found the ACTIONS_RUNNER_PRINT_LOG_TO_STDOUT flag, but are there any other controls? e.g. limiting the types of messages, updating the format (json), etc

[jgresc]

Would also be nice to be able to control the loglevel such as i was done here actions/actions-runner-controller#541

[husseinmimiHarri]

any updates here ?
i need to change log format to json instead of text

[nebuk89]

ACTIONS_RUNNER_PRINT_LOG_TO_STDOUT was initially added for a specific usecase we had in action-runner-controller :) I <3 the idea of expanding it, if you have a full list of controls/ideas you would like to see I would love to get them listed out here. (if there are more than those you have already added).

[JohnYoungers]

I think at the time I was primarily interested in formatting the logs as JSON. I believe at the controller level this can be done with the logFormat flag, but I'm not sure if it can be done at the runner level currently.

At the moment we have ACTIONS_RUNNER_PRINT_LOG_TO_STDOUT set to 0 to cut down on the size of our logs, but if it was formatted as JSON, that might be something we'd want to enable again

[HenrikDK]

Having workflow id, job id, and repository available as fields in the json would also be required for any meaningfull data analysis.

[hae-andrew-young]

JSON would definitely be nice. At the least, could we not have every line begin with something like [RUNNER 2024-10-07 22:01:18Z INFO Runner]? Even multiline logs have lines that begin like this which makes it impossible to grok or parse from a log agent like AWS CloudWatch or DataDog Logs.

[hae-andrew-young]

Found it. It has to do with these lines of code.

runner/src/Runner.Common/StdoutTraceListener.cs

Lines 27 to 36 in 4d8402c

	if (!string.IsNullOrEmpty(message))
	{
	var messageLines = message.Split(Environment.NewLine);
	foreach (var messageLine in messageLines)
	{
	WriteHeader(source, eventType, id);
	WriteLine(messageLine);
	WriteFooter(eventCache);
	}
	}

Why do you have to split them on new line and prepending the header before each line? Seems a bit redundant to me.