Home Original page

[SVLS-8231] fixing log requestId for LMI by jchrostek-dd · Pull Request #973 · DataDog/datadog-lambda-extension

@jchrostek-dd

https://datadoghq.atlassian.net/browse/SVLS-8231

Overview

  • Bug fix - When running a LMI function, function logs don't get attributed with lambda.requestId. As a result, no function logs appear for an invocation. They still show up for the function, just not not on the invocation.
  • We could have alternatively updated the UI to search for lambda.requestId and requestId instead of just searching for lambda.requestId. However, I would prefer for On Demand and Lambda Managed Instance Functions to be consistent. So, both using lambda.requestId is preferred.

Testing

  • Added integration tests that checks for logs per invocation request.
  • Also confirmed in the UI we are seeing the logs for an invocation (before/after screenshot below).

Screenshot 2025-12-30 at 8 27 34 AM Screenshot 2025-12-30 at 8 20 34 AM

@jchrostek-dd

@jchrostek-dd

lym953

Comment on lines +111 to +116

if is_oom_error(&message) {
debug!("LOGS | Got a runtime-specific OOM error. Incrementing OOM metric.");
if let Err(e) = self.event_bus.send(Event::OutOfMemory(event.time.timestamp())).await {
error!("LOGS | Failed to send OOM event to the main event bus: {e}");
}
}

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think we can delete the if is_oom_error(&message) {} block in the TelemetryRecord::Extension case below. I think runtime-specific OOM errors can only come from the function.

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Are we able to validate that? I would have though that extensions can also have an OOM error.

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Will leave for now to be safe.