Remove Deprecated Method for Collecting CloudWatch Metrics

[Hamms]

The CloudWatch monitoring scripts are deprecated in favor of the CloudWatch agent, which we started using with #32110

Links

• https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/monitoring-scripts-intro.html

Follow-up work

We may still have dashboards that are referencing the old System/Linux/MemoryUtilization metrics; those should be updated to use CWAgent/mem_used_percent

[sureshc]

I’ve updated the CloudWatch Overview Dashboard app server Memory Utilization widget to use the newer Metric.

[sureshc]

Looks like we need to update the CloudWatch Alarm to reference the new Metric Namespace and Name here https://github.com/code-dot-org/code-dot-org/blob/4f457167cdd18d84306452a87af0ba084fe43fdc/aws/cloudformation/components/alarms.yml.erb#L29C1-L30

[Hamms]

Great, thank you!

Looks like there's also a reference in

code-dot-org/bin/cron/restart_high_memory_frontend_services

Line 45 in 02f7d6e

memory_utilization = cloudwatch_resource.metric('System/Linux', 'MemoryUtilization').get_statistics(

; I'll update both of 'em, even though I'm pretty sure this one isn't used any more

[sureshc]

We might want to change the TreatMissingData property below to breaching so that we act on missing data with more urgency https://docs.aws.amazon.com/AWSCloudFormation/latest/UserGuide/aws-resource-cloudwatch-alarm.html#cfn-cloudwatch-alarm-treatmissingdata

[Hamms]

That seems reasonable; I'll do so in a follow-up PR, though, since that's unrelated to the metrics switch in this PR