Alerts

Configure alert rules so you are notified when warehouse, cluster, or job metrics cross the thresholds you set.

VeloDB Cloud provides monitoring and alerting at no additional charge (beyond SMS alert notifications).

View alert rules

The Alerts list shows existing alert rules and their current status. A red dot means the rule is firing. A green dot means the rule is not firing. From the list, you can enable, disable, copy, edit, or delete a rule.

Create an alert rule

Go to Observability → Alerts and click New Alert Rule, or copy an existing rule from the list.

An alert rule has four parts:

Part	Description
Rule Name	A name unique within the warehouse.
Target	What the rule monitors. The available metrics depend on the target. When a target is deleted, its rules are kept but become inactive.
Conditions	One or more thresholds. Each condition compares a metric to a value with an operator (>, >=, <, <=, ==, !=). Combine multiple conditions with and / or.
Duration	How long the conditions must hold continuously before the rule fires. For example, alert only after CPU stays above the threshold for 15 minutes. Balance timeliness against noise.

A rule monitors one of three target types:

• A Warehouse target aggregates metrics across the whole warehouse.
• A Cluster target reports metrics for a single compute cluster.
• A Job target reports metrics for a single Routine Load or Warmup job.

You create Warehouse and Cluster rules from the Alerts page as described above. You create Job rules from the job's own page, not from the Alerts page (see Job metrics).

Available metrics

Alert on any warehouse or cluster metric shown on the Metrics page. See that page for each metric's meaning, unit, scope, and minimum version.

Job metrics are specific to a single Routine Load or Warmup job and are listed below.

Job metrics

Job rules are not created from the Alerts page. Create them from the job's own page:

• Routine Load: Go to Ingestion → Import Data, click a job name to open its job page, then click Create Alert.
• Warmup: Go to Compute → Warmup Jobs, click a Job ID to open its job page, then click Create Alert.

Use job metrics to catch problems in a single job, such as a Routine Load job that falls behind or accumulates errors.

Metric	Job type	Unit	Available from
Routine Load Job Error Rows	Routine Load	rows	4.1.8
Routine Load Job Lag	Routine Load	rows	4.1.8
Routine Load Job Aborted Transactions	Routine Load	tasks	4.1.8
Warmup Job Trigger Gap	Warmup	ms	4.1.8
Warmup Job Pending Size	Warmup	bytes	4.1.8

Recommended alerts

Click Enable Recommended Alerts to create a recommended set of alert rules automatically. The set applies to the warehouse and all of its compute clusters, including any clusters you add later.

It includes the following rules. Warehouse rules apply to the whole warehouse. Cluster rules apply to each compute cluster.

Scope	Metric	Condition	Duration
Warehouse	CPU Utilization	> 60%	15 min
Warehouse	Memory Utilization	> 80%	15 min
Warehouse	Active Connections	> 150	15 min
Warehouse	Unreachable Node Count	> 0	1 min
Warehouse	Inactive FE Service Node Count	> 0	1 min
Cluster	CPU Utilization	> 80%	15 min
Cluster	Memory Utilization	> 80%	15 min
Cluster	Unreachable Node Count	> 0	1 min
Cluster	Disk Cache Hit Rate	< 90%	15 min
Cluster	Average Query Latency	> 5000 ms	5 min
Cluster	P99 Query Latency	> 60000 ms	5 min
Cluster	Query Success Rate	< 90%	5 min
Cluster	Compaction Score	> 1500	15 min

Notification channels

Every alert rule can push to one or more channels. Each channel pushes the alert message independently.

For the In-site notification and Email channels, select the users to notify.

For the SMS channel, select users or enter phone numbers directly.

The Webhook channel forwards alert payloads to any external system that accepts HTTP callbacks. VeloDB Cloud sends a POST request with a JSON body when an alert fires. Your endpoint must respond with HTTP 200. Otherwise VeloDB Cloud retries the delivery.

The PagerDuty channel routes alerts into PagerDuty incidents using a service integration key. To set up this channel:

1. In PagerDuty, open Services, then Service Directory, and open the target service or create a new one.
2. Open the Integrations tab, then click Add Integration.
3. Search for Events API V2, select it, then click Create Service.
4. Copy the Integration Key from the integration detail page.
5. Paste the integration key into the alert channel configuration.

The Slack Group channel forwards alerts to Slack incoming webhook URLs. To use more than one URL, separate them with a semicolon (;). To set up this channel:

1. In Slack, open Apps, search for Incoming WebHooks, then click Add to Slack.
2. Choose the channel that should receive alerts, then click Add Incoming WebHooks Integration.
3. Copy the webhook URL into the alert channel configuration.

The Lark channel forwards alerts to a Lark custom bot. To set up this channel:

1. In the target group, open Settings, choose BOTs, click Add Bot, then select Custom Bot.
2. Give the bot a name and description, then click Next.
3. Copy the webhook URL into the alert channel configuration.

The DingTalk channel forwards alerts to a DingTalk custom robot. See DingTalk's guide for the full procedure. In summary:

1. In the target DingTalk group, open Group Settings, then Group Assistant, then Add Robot, and choose Custom.
2. Set a profile picture, name, and security settings. For security, use Custom Keywords and enter alert.
3. Accept the terms, then click Finished.
4. Copy the webhook URL into the alert channel configuration.

Note The Slack, Lark, and DingTalk channels let you restrict message sources with a webhook IP allowlist. The VeloDB Cloud server IP is 3.222.235.198.

Alert history

View the alert firing history on the Alerts page. Filter it by time, rule, or cluster.

Next steps

• Metrics: review the metrics you can alert on.
• Metrics API: scrape metrics into your own monitoring stack.