[SLO] Create an embeddable alert summary

[mgiota]

Slack Thread

🍒 Summary

We should allow users to add an alerts summary to their dashboards with predefined filters. This would allow SREs to see a summary of the alerts for the SLOs they are viewing on the dashboard.

🎨 Design requirements

• Select multiple SLOs
• Add KQL query for additional filters the alerts
• @maciejforcone to provide updated design for current alert summary widget
• Empty States:
  • When there are no rules
  • When there are no active or recovered alerts

Acceptance criteria

• Create an embeddable alert summary
• Embeddable should have an initial width and height instead of getting the default size (relates to [SLO] [Embeddables] provide initial width for the slo overview embeddable #170720

Tasks

Beta Give feedback

1. SLO configuration with option to select multiple SLOs
2. Fetch active alerts logic
3. SLO alert widget UI
4. Set initial dimensions to the panel
5. Display selected SLOs in the widget
6. Unit tests with different scenarios
7. Alert Widget empty state
8. Option to edit the configuration
9. Link to Observability alerts page
10. Should honor dashboard's time picker
11. Should honor manual Refresh button

[elasticmachine]

Pinging @elastic/actionable-observability (Team: Actionable Observability)

[mgiota]

According to the [Slack thread](Slack thread) of this issue, untracked alerts shouldn't be included in the calculation of total alerts. Total alerts = Active + Recovered

Update: We will not include Total alerts in the widget, just Active Alerts.

[elasticmachine]

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

[mgiota]

I summarize here the points that were discussed in the Slack Thread with @maciejforcone while refining this issue

• Shall we display total number of alerts or shall we just focus on the number of Active alerts? (leave out recovered and untracked alerts).
• Technically I managed to display the alerts table in the embeddable, so we are thinking to display the table as well.
  • One question we had is if it makes sense to show only active alerts
  • Leave it to the user to show also recovered alerts
  • Does it make sense to the team to display the table as well or shall we focus only to the summary widget as part of this ticket.

[mgiota]

Update:

• We will focus on the Alert Summary widget first and later on we can make it configurable so that user can select what else they might want to see list of alerts, activity chart, possibly recovered alerts.

• We will show only number of Active alerts (no recovered alerts)

• The UI will look very similar to the current SLO Overview widget, but we decided not to use the Metric chart, because it doesn't support badges at the moment and we need badges for displaying the selected SLO names.

• Because the user can select many SLOs and there is of course not much space, we decided to display only the first one (truncated if needed) with a +5 more badge

• That's why we need to support the Edit settings option in the embeddable, which will open the SLO modal configuration with the selected SLOs. _Technically the Edit settings can be achieved by adding triggers and actions to the embeddable (I will use Links and ML Anomaly Detection embeddables as an inspiration on how to do this)

• We need to set an initial width and height to the alert embeddable similar to this
• Clicking on the alert widget leads to Observabilty Alerts pages filtered by sloIds, sloInstanceIds and active status
• Support empty states
• Should honor dashboard's time picker
• Should honor manual refresh button

[mgiota]

@maciejforcone When user selects multiple SLOs, where one can have active alerts and the other one has no active alerts (could be either 0 alerts or no alert rules), how do we display it with current design? What color has the widget, what is the title and what is the badge color of the selected SLOs? Combining the information of multiple SLOs with current design is restrictive. What do you think? Maybe we shouldn't allow the selection of multiple SLOs? Any other suggestion?

[mgiota]

@maciejforcone Does it make sense to still work on this one? Since the Overview embeddable now shows the number of active alerts (for one SLO), I was thinking maybe this is redundant. I can continue working on this one, if you think it will be useful information for the user though.