SLO rules

SLO rules are optional. Out of the box, Hud compares every flow with its own normal, with nothing for you to set. If a flow has a hard target, such as a checkout endpoint that must respond within 2 seconds, you can add an SLO: you set the duration an endpoint, queue or custom flow has to stay under, and Hud checks every 10 minutes and opens an issue when any of its calls in the last 10 minutes took longer.

The rule is called SLO Breach. It has no on/off switch of its own: it runs wherever a detection condition has Issue detection turned on, and Issue detection starts off.

Set an SLO

In the web app, go to Rules and find SLO Breach.
Choose what the SLO applies to: Endpoint configuration, Queue configuration or Custom flow configuration.
Under Issue Detection Conditions, select Detection Condition.
Pick the services it applies to. To narrow it to specific endpoints or queues, pick one service, then choose them.
Turn on Issue detection and set Duration SLO in milliseconds. The default is 5,000 ms for every endpoint, and Reset to Hud recommendations puts it back.
Select Save.

To set one SLO for everything, edit the Default (all) row instead.

Get alerted

Detection conditions decide when Hud opens an issue. Alert conditions decide when Hud alerts you about it, so set both. Under Alerts Conditions, select Alert Condition, pick the same scope, turn on Issue alerts and set:

Setting	What it does	Default
Duration SLO	The duration, in ms, that triggers an alert	5,000 ms
Duration percentile	Which duration Hud compares with the SLO: P90, P95, P99 or P100	P100
Alert on recurring issues	How long Hud waits before alerting on the same issue again: 1, 3 or 7 days	Every 3 days

The alert condition's Duration SLO doesn't have to match the one in the detection condition.

SLO alerts go wherever your other alerts go. See Alerts.

Turn it off for part of your system

Add a detection condition for that scope and leave Issue detection off. Hud stops opening SLO issues there. To keep the issues but stop the alerts, add an alert condition with Issue alerts off.

To see the issues the rule has opened, select View Issues on the rule's page.

When you don't need an SLO

To catch an endpoint that got much slower than its own normal, you don't have to set anything. See Slowdowns over time and Post-Deploy Issues.