Context:
- we use
aws:executeAutomation
inside another SSM automation
- we use
Targets
and TargetParameterName
in combination with MaxConcurrency
to execute on one instance at a time
- the automation erroneously executes the first instance twice and then exits with:
Overall status: Failed
even though none of the steps returned any failures
In other words, instead of executing: Instance A (Step#1) -> Instance B (Step#2), it executes Instance A (Step#1) -> Instance B (Step#2) -> Instance A (Step#1)
If the child automation is not executed from a parent (meaning that it is executed from the cli directly), then the behavior is as expected: just 2 steps are executed, one for each of the 2 instances and the result is successful
Also, a thing of note is that the cli command aws ssm get-automation-execution --automation-execution-id
shows only the first two steps of the erroneous execution. So, the management console shows 3 steps in total, but the CLI only the first 2.
If you query the third step (the one who should not exist) using the cli, it shows the parent correctly, but the parent does not see it
Please fix this bug :D
Please provide the following information to help with investigation. And the cases happening in the recent 1 month would be better.
We get another data point from our customer. We will work on code fix to address this.