[Metricbeat] Add googlecloud compute overview dashboard

[kaiyan-sheng]

This PR is to add googlecloud compute overview dashboard:

How to test it

1. enable googlecloud module

./metricbeat modules enable googlecloud

2. make sure in the google cloud account you are using for testing, there are VMs running. Note on what are the regions they belong to and remember to change region in step 3 configuration.

3. change modules.d/googlecloud.yml to:

- module: googlecloud
  metricsets:
    - compute
  region: "us-central1"
  project_id: elastic-observability
  credentials_file_path: "/Users/kaiyansheng/Downloads/elastic-observability-d17781618202.json"
  exclude_labels: false
  period: 300s
- module: googlecloud
  metricsets:
    - compute
  region: "europe-west1"
  project_id: elastic-observability
  credentials_file_path: "/Users/kaiyansheng/Downloads/elastic-observability-d17781618202.json"
  exclude_labels: false
  period: 300s

4. Start metricbeat and you should see data points in the compute dashboard shortly.

closes #16534

[sorantis]

• The filters can be combined into one control instead of having 3 separate ones.
• instance.uptime can be added to the dashboard as a label between the filters and CPU graph.

[kaiyan-sheng]

• The filters can be combined into one control instead of having 3 separate ones.

Good point, done!

• instance.uptime can be added to the dashboard as a label between the filters and CPU graph.

Thanks for your suggestions @sorantis ! I just added uptime as a separate visualization. The problem is instance/uptime metric from GCP is delta kind, which means it measures the change in a value over a sample period. So with this uptime visualization, you won't be able to see numbers going up to larger than 60.

[sorantis]

The problem is instance/uptime metric from GCP is delta kind, which means it measures the change in a value over a sample period. So with this uptime visualization, you won't be able to see numbers going up to larger than 60.

@kaiyan-sheng I see the issue. I was thinking about it as a True/False indicator(sorry for not making it clear), e.g. <green>Running 34<green>, <red>Down 19</red>. If there's no other way of interpreting this metric, then we should probably not show it at all.

[kaiyan-sheng]

@sorantis Thank you for the suggestion. I changed uptime to show 100% Running 34 and less than 100% Running 2. With how googlecloud metrics are collected, this is still not accurate . Because when you set period=5m in gcp compute metricset, you get metrics every minute still for 3 times in each 5 minute.

I will put this PR into work in progress till I figure out how to accurately display this uptime. Thanks!

[kaiyan-sheng]

@sorantis What do you think about changing uptime using a gauge to show percentage here? I'd like to show True/False indicator but the data we are collecting here doesn't give me a good way of displaying the exact number of instances. I'm also working on figuring out how we can improve data collection shown as screenshot above. After figuring that out, I can come back and revisit the uptime visualization. Thanks!

[sorantis]

looks much better, @kaiyan-sheng, thanks for revisiting it!

[sorantis]

nit: perhaps we can align all dropdowns vertically, and place the gauge between the filters and cpu utilization? this should save space, improve symmetry (and help my OCD :D)

[kaiyan-sheng]

@sorantis Totally understand. I just made the change. Thanks!

[sorantis]

LGTM