Cherry-pick of PR #20875 to 7.x branch. Original message:

What does this PR do?

This PR adds the latency config parameter for aws module. With this config parameter, users can collect CloudWatch metrics with a latency larger than Metricbeat collection period. For example, for S3 request metrics, the latency can be 3 minutes. The recommended collection period is 1min. Without the latency config parameter, no data points will be collected ever.

In this screenshot: current timestamp is 14:21 but we only see the latest data at 14:18 with a 3-minute delay. In this case, without this fix, our Metricbeat module will keep trying to collect data from the last 2-minute time frame, which will always be empty:

(@gbanasiak Thank you for your screenshot!)

Why is it important?

latency config sets CloudWatch API start time and end time with a delay. For example, if the current time is 2020-08-31T18:30:00.000Z with collection period 1m.

Without latency:
startTime = 2020-08-31T18:29:00.000Z
endTime = 2020-08-31T18:30:00.000Z

With latency set to 5min:
startTime = 2020-08-31T18:24:00.000Z
endTime = 2020-08-31T18:25:00.000Z

Checklist

• My code follows the style guidelines of this project
• I have commented my code, particularly in hard-to-understand areas
• I have made corresponding changes to the documentation
• I have made corresponding change to the default configuration files
• I have added tests that prove my fix is effective or that my feature works
• I have added an entry in CHANGELOG.next.asciidoc or CHANGELOG-developer.next.asciidoc.

How to test this PR locally

1. Enable aws module: ./metricbeat modules enable aws
2. Change modules.d/aws.yml to:

- module: aws
  period: 5m
  credential_profile_name: elastic-beats
  latency: 5m
  metricsets:
    - ec2

3. Start Metricbeat with debug level logging: ./metricbeat -e -d "*"
4. You should see the startTime and endTime in debug log with a latency of 5 minutes applied.