Monitoring¶

Servala monitors your service instances and exposes their health so you can keep track of how they are running. This page describes metrics and alerting in general. What is available depends on the service and offering. See the relevant page in the Service Catalog for service-specific details.

Metrics¶

Service instances expose metrics about their state and resource usage, such as:

• CPU and memory usage
• Storage usage
• Service-specific indicators (for example connections or query rates for databases)

Some metrics are visible on the instance detail page in the portal. See Monitor an Instance for how to read CPU, memory, and disk usage there.

Alerting¶

Servala alerts on service level indicators (SLIs) and capacity thresholds so that problems are noticed early:

• Availability and health: alerts when an instance is unhealthy or unreachable.
• Capacity: alerts when resources such as storage approach their limits.

When an alert fires, Servala follows the operations runbooks for the service to resolve the incident. Alerting coverage depends on the service and the chosen service level.

Alerts for your instance¶

Each instance also has its own set of alert rules. These rules watch the instance itself, for example a data volume that is filling up or memory that runs short. The set of rules depends on the service.

The Alerts card on the instance detail page shows these rules and their current state. See Monitor an Instance for how to use it. The card shows two lists:

• Firing now: alerts that are active at the moment.
• Watched and quiet: all other alert rules of the instance. Each row shows what the rule watches for and how long the condition must last before the alert fires (for example "after 15 minutes").

The list of quiet rules tells you what is being watched, even when nothing is wrong. If the card says that no alerts are set up, no user facings rules are available for this instance.

Each alert has a severity:

When at least one alert fires, a banner at the top of the instance detail page shows how many alerts are firing. The banner has the colour of the highest severity.

For what to do when an alert fires, see the Alert Runbooks.

Email notifications¶

To receive alerts by email, set an alert email address in the Monitoring section of the service configuration. This option is available for selected services only.

Muting an alert¶

If you already know about a problem, you can mute the alert. A muted alert stops sending email notifications, and it no longer shows the banner on the instance detail page. The alert stays in the Firing now list with a Muted label and the time when the mute ends.

Rules for muting:

• Only organization Administrators and Owners can mute and unmute alerts. All members can see the Alerts card.
• You can only mute an alert that is firing now. You cannot mute an alert in advance, and you cannot mute all alerts of an instance at once.
• You choose a duration: 1 hour, 4 hours, 24 hours, or 7 days. A mute cannot last longer than 7 days.
• You must give a reason (up to 200 characters).
• When the mute ends and the alert still fires, notifications start again.
• You can unmute an alert before the mute ends.

Muting only affects the notifications to you. It does not stop the Servala operations team from being alerted about problems on the platform.

An alert with the label Muted by support was muted by Servala staff outside the portal. You cannot unmute it in the portal. Contact support if you want to change it.

Every mute and unmute is recorded in the organization audit log, together with the duration and the reason.

The portal does not keep a history of past alerts. It only shows the current state.

Further reading¶

• Service Levels
• Compute Plans & SLAs