operation insight dashboard incorrect graph

[whoDoneItAgain]

Describe the bug

The pie chart (for atleast ec2 instances controlled) is not displaying by instance type like depicted in the implementation guide(https://docs.aws.amazon.com/solutions/latest/instance-scheduler-on-aws/monitor-the-solution.html)

To Reproduce

basic deployment

Expected behavior

the insight dashboard shows instances controlled by instance type

Please complete the following information about the solution:

• Version: v3.0.11

To get the version of the solution, you can look at the description of the created CloudFormation stack. For example,
"(SO0030) instance-scheduler-on-aws v1.5.1". You can also find the version from
releases

• Region: us-east-1
• Was the solution modified from the version published on this repository? yes
• If the answer to the previous question was yes, are the changes available on GitHub? only modification was changing the deletion and update replace policy from retain to delete. I have a heavily modified version i just deployed but deployed the vanilla version to confirm the results i'm seeing in my other deployment.
• Have you checked your service quotas for
  the sevices this solution uses? n/a
• Were there any errors in the CloudWatch Logs? no
  Troubleshooting

Screenshots
If applicable, add screenshots to help explain your problem (please DO NOT include sensitive information).

Additional context
Add any other context about the problem here.

[gockle]

Thanks for reaching out, we will investigate this issue and provide further updates.

[whoDoneItAgain]

thanks @gockle. I'd also like to add that the bar graphs in the screenshot are depicting similar behavior.

[CrypticCabub]

hi @whoDoneItAgain

Have you modified any of the Hub Stack CloudFormation parameters from their default values? the first picture you sent looks like the graph is using the wrong underlying metric (there are both total and per-instance-type metrics)

I've been working on some updates to the dashboard recently which includes a lot of testing but have not come across the issue you are describing. Sometimes spikes like in your second picture can occur in RDS due to transitional states that do not fall cleanly into Instance Scheduler's categorization of running/stopped but I've not typically observed them with EC2 instances (those spikes are also typically down rather than up).

[whoDoneItAgain]

No. that screenshot was from a vanilla deployment except for changing the deletion and updatereplace polices to delete instead of retain. I did some more testing this morning and discovered that with only a single instance, it shows like the screenshots. I've since added 3 more EC2s and i am now showing it grouping with the expected labels. For the 'By Schedule' charts, i still only have one. I suspected, based on my findings of going from 1 managed ec2 to multiple, once i add another assigned schedule to the mix those 2 charts will populate correctly. This behavior could be a caveat to mention in documentation.

Note, i haven't onboarded any RDS instances yet.

[CrypticCabub]

Ah, very interesting. my test account runs ~ 100 instances of various types so that would explain why I'm not seeing it.

on a related note. do you feel that the by-schedule chart provides any useful value? I am strongly considering removing the metrics that power that chart to reduce metrics cost as the data doesn't seem to me to be of much value but still costs ~$0.60/schedule (billed separately for EC2 and RDS)

edit: billing is monthly*

[whoDoneItAgain]

confirmed. i assigned a second schedule to an instance and it immediately started showing correct.

[whoDoneItAgain]

to answer your question, yes i find value. Its a good way to help audit provisioned schedules vs assigned schedules

[CrypticCabub]

Interesting. that's good to know.

Here's a preview of what I'm been experimenting with as a more condensed dashboard view that includes ASG support:

Presently I don't have any per-schedule metrics on there since "running by schedule" is rather redundant, though I could potentially see a "managed by schedule" widget being rather useful. If you could share a little more about how you would intend to use the dashboard and what widgets are/are not useful it would be very much appreciated.

[whoDoneItAgain]

i do like the updated dashboard. i was going to ask about ASG support but see i don't need to now. This preview is much better. More info is available without having to scroll around. Consolidating the Running Resources into a single graph makes sense.

Additional info on my bug.

i thought the bug was when only 1 instance was managed. it looks like it might be when only one instance type is scheduled. both those RDS instances are the same instance type. I'm hoping to move another RDS over to the updated scheduler today that is a different type to see if it starts showing the correct information

[CrypticCabub]

I see -- I wonder if it has to do with the default behavior of how the widget shows search-expression data. The instance-type is a metric dimension with the metric name being ManagedInstances like you are seeing there. Basically if it doesn't need to distinguish between two different dimensions the label is defaulting to metric.name() instead of metric.dimension()

I'll have to see if there's a way to force the label to use the dimension name even in that scenario.

[whoDoneItAgain]

confirmed. the update cycle after adding the rds instance of another instance type, its showing correctly.