Alert methods
⛭ Close an alert method
The PagerDuty alert method supports an alert resolution feature that allows you to send an all-clear notification from Nobl9 signaling that an alert should be marked as resolved in the PagerDuty UI. For details, see PagerDuty alert. For other alert methods, they must be closed manually.
⛭ Custom headers limit for the Webhook alert method
You can include up to 10 custom headers for the Webhook alert method.
⛭ Notifications on SLO creation or removal based on user alert preference
Currently, you cannot set up notifying users about SLO creation or removal based on their alert preference.
⛭ Access to fired alerts
The only way to get the log of all fired alerts is through sloctl. To do that, use the sloctl get alerts command.
⛭ The PagerDuty alert doesn't resolve upon SLO going back to green
To activate this feature, you need to check the Send a notification after the cooldown period is over box in the PagerDuty alert method wizard or add the optional sendResolution section to your PagerDuty YAML configuration (with a value other than the default of nil).
Read more about PagerDuty alert.
⛭ PagerDuty alert resolution via Terraform
Our Terraform provider doesn't support PagerDuty alert resolution.
⛭ The behavior of the alert mechanism when an alert condition is satisfied
If the condition remains triggered, the alert will only fire once, and it will stay open. It will be closed after the condition stops being triggered for a period of time, called the cooldown period. After the alert is closed, a new one can be triggered if the condition arises again.
Read more about alert methods.
⛭ The condition "Average burn rate is ≥ 0 and this condition lasts for 1 minute" triggers the alert only once
In this example, the alerting condition indicates that an alert is triggered after one minute of the error budget burn.
Once the burn rate equals 0, the error budget is not burning anymore,
but the alerting condition (the burn rate = 0) is still satisfied.
As a result, Nobl9 doesn't start measuring the cooldown period, and no new alerts be triggered.
To avoid satisfying the condition when the error budget isn't burning, set the value to > 0.