Alerts & Check-Up
To help you discover potential problems, pganalyze automatically performs a number of checks on your database.
Checks run regularly, at different frequencies depending on the check type. When a problem is detected, an issue is created for the server or database where the problem is found.
Issues may have severity “info”, “warning”, or “critical”. Critical issues may indicate a serious performance problem or potential database outage within a matter of hours. Warnings are less serious, but should still be investigated as they can often escalate to critical severity. Info-level issues encourage best practices and indicate opportunities for performance improvements.
Alerting
When a check creates or updates an issue, pganalyze can notify you through one or more alert destinations. A destination is a named, reusable target that can send alerts to any combination of:
- E-mail — selected organization members
- Slack — a channel, via the Slack integration
- PagerDuty — a service, via the PagerDuty integration
You manage destinations on the Alert Destinations tab of the Alerts & Check-Up page. Each organization has a default destination that new members are added to automatically.
Connecting integrations
Slack and PagerDuty destinations require a connected integration. Connect them on the Integrations tab of your organization Settings page:
Click the “Connect to …” button for a provider and complete its authentication flow. Once connected, use the “Test” button to confirm the setup: for Slack this sends a message to the selected channel, and for PagerDuty it opens an incident on the selected service (which you should then resolve yourself). A destination’s Slack and PagerDuty targets are chosen from the channels and services these integrations make available.
Alert rules
Alert rules decide which issues reach which destinations. A rule applies to a specific server (or all servers), matches a set of checks, and routes matching issues to one or more destinations. You manage rules on the Alert Rules tab of the Alerts & Check-Up page. Whether and when an issue reaches a given severity is still governed by each server’s per-check configuration; alert rules only control where matching issues are sent.
Alerts are delivered according to severity: Slack destinations receive matching issues
of any severity, while PagerDuty and e-mail destinations are notified only for critical
issues — either issues that start at critical severity (like a disabled fsync setting)
or issues that escalate to critical (like a long-running query).
Slack notifications
The initial alert is sent as a new message, and subsequent changes to the issue update that message and start a thread summarizing issue activity.
Note: the pganalyze app must be invited to the channel you select in order to post there. You can do so from within Slack by typing “@pganalyze” in the channel: you’ll be prompted to invite the app.
PagerDuty alerting
Acknowledging an issue in pganalyze acknowledges the PagerDuty incident; re-opening an issue re-triggers it; and resolving the issue in pganalyze resolves the incident automatically. You may interact with the incident through PagerDuty, though those changes are not reflected back in the corresponding pganalyze issue.
Read more about the PagerDuty integration
Checks
Checks are organized into several different categories:
- queries
- connections
- index_advisor
- schema
- settings
- system
- replication
- vacuum
Couldn't find what you were looking for or want to talk about something specific?
Start a conversation with us →