Real Time Alerts
Get notified the moment a warehouse starts queueing, runs hot, or pins at its max cluster count — with rules you scope, threshold, and route yourself.
Overview
Real Time Alerts watch your warehouses as they run and notify you when one starts to misbehave — queries piling up in the queue, a runaway queue-to-running ratio, or a multi-cluster warehouse stuck at its ceiling. Unlike Anomaly Detection, which looks at daily cost patterns, Real Time Alerts react to live load, within minutes.
You manage them on the Alerts tab of the Real Time Monitoring screen (Monitoring → Real Time → Alerts).
Required role: Admin or higher to create, edit, or delete rules. Anyone can view existing rules and their status.
The alerts table
Every rule appears as a row you can sort, search (by name), and filter (by status). Each row shows:
Name
The rule name, with a toggle to enable or disable it.
Status
OK (active, nothing wrong), Firing (condition currently met — shows a count if more than one warehouse is firing), or Disabled. Hover a firing rule to see how long it's been firing.
Condition
What the rule watches: Queue depth high, Queue / running ratio high, or Max clusters sustained.
Scope
Which warehouses it covers — specific warehouses, a whole account, or all warehouses.
Severity
Critical, Warning, or Info.
Channels
Where notifications are sent (Slack, Email, Microsoft Teams).
Fired (7d)
How many times the rule fired in the last 7 days.
Last fired
When it most recently fired (or Never).
Created at / Created by
When the rule was created and by whom.
Each row also has Edit, Duplicate, and Delete actions. Duplicate is handy for building a family of similar rules across different warehouses.
Creating a rule
Click Create a new rule (or the bell icon on any warehouse tile to pre-scope the rule to that warehouse). The builder walks through four steps.
Alert
Pick what to watch for and set the threshold:
Queue depth high
Queued queries stay ≥ your threshold for a sustained window.
Queue / running ratio high
The queued-to-running ratio stays ≥ your threshold for a sustained window.
Max clusters sustained
A multi-cluster warehouse runs at its max cluster count for the whole sustained window.
You also set:
Threshold — the queue count or ratio that trips the alert (not needed for Max clusters sustained).
For at least — the sustain window the condition must hold before firing: 1, 2, 5, 10, 15, or 30 minutes. This prevents momentary blips from firing an alert.
Severity — Info, Warning, or Critical (a label to help you triage).
Name — auto-generated from the scope and condition; edit it if you prefer your own.
Notifications
Notify channels — choose one or more channels. These come from your configured Communication Channels (Slack, Email, Microsoft Teams) under Settings → Notifications.
Also notify when the alert resolves naturally — send a follow-up when the condition clears, so the channel sees the all-clear.
Remind every (while open) — optionally re-send the alert on an interval (15 min, 30 min, 1 hour, or 4 hours) while it's still firing.
How firing works
Seemore continuously evaluates each rule against live warehouse activity.
A rule fires only after its condition holds for the entire sustain window — brief spikes don't trigger it.
When a rule fires, Seemore sends a notification to every configured channel, the rule's Status flips to Firing, and the matching warehouse tile on the Compute units tab is highlighted.
When the condition clears, the rule returns to OK, and — if you enabled it — a resolve notification is sent.
Tips
Start with a sustain window of a few minutes. Queueing is normal in short bursts; a 5-minute window catches sustained pressure without noise.
Scope broadly, then refine. An account-wide Queue depth high rule is a good baseline; add warehouse-specific rules with tighter thresholds for your critical warehouses.
Pair alerts with automation. Burst Protector can absorb the spikes your alerts are warning you about, automatically and within the same cost envelope.
Related
Real Time Monitoring — the live fleet view these alerts run on top of.
Communication Channels — set up the Slack, Teams, and Email destinations alerts route to.
Anomaly Detection — for daily cost-pattern alerts rather than live load.
Last updated