[SLO] SLO ops management page

[kdelemme]

🍒 Summary

The SLO ops management page would provide a single pane of glass for users to manage all their SLOs and get insight into potential issues with the underlying infrastructure powering the SLOs, e.g. pipeline and transforms.
This page would focus on the SLO definitions instead of the SLO instances as the current SLO listing page is showing.

Available bulk actions:

• Bulk delete SLO
• Bulk reset SLO
• Bulk delete stale SLO instances

The page lists the SLO definitions including:

• SLO name
• Ops Health Status
• Number of instances
• Other?

Misc:

• Pagination
• Searching by slo.name
• Searching by slo.id
• Filter by status
• Sorting by number of instances

Questions

• Expectation around search?
  • In order to investigate issues with an SLO, would you search by an SLO id, name, instanceId?
  • Do we expect to surface the problematic SLO first? i.e. sort by Ops Health Status?

Ops Health Status

Note

This field provides the overall SLO health status from an operational standpoint. If possible this should be a red-yellow-green light.

This field is computed using the following checks:

• SLO Version Model is up to date
• Both transforms exist and are healthy
• Both ingest pipelines exist
• Duration between last sli @timestamp and its event.ingested is within reason
• Other?

APIs

Note

Some existing APIs might be reusable.

List SLO definitions

The current GET /_definitions routes uses the SLO Repository directly. On this Ops page, we need to be able to filter by the Ops Status, which requires us to start from the summary index and then merge the result with the SLO Definition retrieved from the SLO Repository.

[elasticmachine]

Pinging @elastic/obs-ux-management-team (Team:obs-ux-management)