Responding to alerts
Published 17 August 2020
SQL Monitor raises an alert whenever the threshold conditions for the alert are breached. For example, it raises a high severity alert if a machine is unreachable for more than 30 seconds, a medium severity alert if CPU utilization is more than 90% for longer than 600 seconds, a low severity alert if a query runs longer than 10 minutes, and so on. It also raises alerts when certain SQL Server events occur, such as a deadlock or a job failure.
The interactive activity graph at the top of the Server Overview page shows when alerts were raised, within the context of server activity around that time, and you can review all the details of each alert raised using the Alert Inbox, as shown and described below.
Using the Alert Inbox
A user who is logged in to the primary Base Monitor can see the list of alerts in the Alert Inbox in the GUI, and they are usually also sent via email or over Slack messaging. SQL Monitor automatically groups alerts that relate to the same problem, such as the same long-running query, to prevent recipients being overwhelmed with too many individual alerts in the inbox. From the Alert Inbox, you can also perform several actions such as marking an alert as read, clearing it, or adding a comment. You can also apply an action to multiple alerts, by selecting them in the inbox.
For full details, see: Using the Alert Inbox.
You can filter which alerts appear in the Inbox by monitored object (such as by instance or database), or by any of their attributes:
- severity (Low, Medium, High)
- status (Event, Active, Ended, Cleared)
- type (e.g. Job failed, Backup overdue, or any selection)
- action taken (Read, Unread)
For full details, see: Filtering alerts.
Viewing the details of raised alerts
Each alert provides all the information required to determine an appropriate response. The alert details provide specialist diagnostic information for that alert type. For example, the alert details for a blocked query alert will contain the full blocking chain, and a Deadlock lock will provide the deadlock graph. All alerts provide a range of performance data that includes a snapshot of machine and SQL Server metrics, system processes, and the "Top 10 queries" running around the time of the alert. So, for example, for a CPU utilization alert, we can quickly find the most CPU-intensive queries that were running, their query plans, any associated waits, and so on.
For full details, see: Viewing details of an alert.
Alert configuration actions
The ACTIONS section in the Alert Inbox will take you to the Configuration menu in the SQL Monitor Web interface, to allow you to customize your alerting strategy and alert settings. You can perform actions such as:
- Change alert settings – the Alert settings page allows you to disable an alert or change its level and thresholds.
- Temporarily suspend alerts – suppress alerts for specific periods, during which time alerting, or just alert notifications, is suspended on selected servers.
- Create custom metrics and associated alerts – the Create custom metric wizard allows you to create a metric and add a custom alert based on the collected metric values.
All these actions are described in the section: Metrics, alerts and notifications, where you'll also find a full description of how SQL Monitor alerts work, the different types of alert that can be raised, and their settings.