Conversation
avedmala
changed the title
| sfn_event = { | ||
| "Execution": { | ||
| "Id": "arn:aws:states:sa-east-1:425362996713:execution:abhinav-activity-state-machine:72a7ca3e-901c-41bb-b5a3-5f279b92a316", | ||
| "Name": "72a7ca3e-901c-41bb-b5a3-5f279b92a316", |
There was a problem hiding this comment.
Updated these cases to match the logs-backend and also the node tests added in DataDog/datadog-lambda-js#598
| state_name = context.get("State").get("Name") | ||
| state_entered_time = context.get("State").get("EnteredTime") | ||
|
|
||
| redrive_postfix = "" if redrive_count == 0 else f"#{redrive_count}" |
There was a problem hiding this comment.
Is it possible that redrive _count is null?
There was a problem hiding this comment.
Execution.RedriveCount will always be in the SFNs context object, defaulting to 0
Only Execution.RedriveTime has a risk of being missing, only appearing when redrive count is 1 or more
https://docs.aws.amazon.com/step-functions/latest/dg/input-output-contextobject.html
There was a problem hiding this comment.
Even though this is supposed to be true, I found some instances in customer logs where RedriveCount is null. Wasn't able to reproduce it myself. I'll add some null-check here and also for the node layer change accordingly
There was a problem hiding this comment.
I guess we'll need to assume all the new fields could be null in the future. There may be some edge cases that AWS does not handle well.
3 participants
What does this PR do?
Adds support for Step Functions trace merging in Redrive cases
We previously used
hash(ExecutionId # StateName # StateEnteredTime)for spanID calculation but these values are identical across redrives for a Lambda task state. The new approach also adds aRedriveCountto the end of the hash but omits this value when it is 0 to have easy backwards compatability.Motivation
https://github.com/DataDog/logs-backend/pull/84188
Testing Guidelines
Redriven Trace with Lambda and Trace Merging
Types of Changes
Check all that apply