Alerts & Thresholds

Configure thresholds for Hive agents to get notified when a server metric crosses a warning or critical level. Alerts fire when a violation persists for a configured duration.

How Thresholds Work

Each threshold monitors a single metric on one or more agents. When the metric exceeds the configured value for the required duration, Hive creates an alert and sends notifications to your configured channels.

Metric value > threshold  AND  violation duration ≥ durationSeconds
                              → fire alert (warning or critical)

When the metric returns to normal, the alert is automatically resolved.

Threshold Fields

Field	Description
Metric Type	Which metric to monitor (see list below)
Warning Threshold	Value that triggers a warning alert
Critical Threshold	Value that triggers a critical alert
Duration (seconds)	How long the violation must persist before alerting (avoids spikes)
Comparison	`greater_than`, `less_than`, or `equals`
Notify on Warning	Send notification for warning-level violations
Notify on Critical	Send notification for critical-level violations
Enabled	Toggle threshold on/off without deleting it

Supported Metric Types

Metric	Description
cpu	CPU utilization (%)
memory	Memory utilization (%)
disk	Disk utilization (%)
network_in	Inbound network throughput (bytes/s)
network_out	Outbound network throughput (bytes/s)
load_1m	1-minute load average
load_5m	5-minute load average
load_15m	15-minute load average

Organization Defaults vs Agent-Level Thresholds

Organization defaults apply to all agents that have no agent-level override. Set them once and every new agent inherits them automatically.

Agent-level thresholds override the organization defaults for a specific agent. Use these when a server legitimately runs hotter than your baseline (e.g. a build server vs. a web server).

Setting Organization Defaults

Go to Hive → Thresholds
Click Edit Defaults
Configure the metric, values, and duration
Save — all agents without agent-level overrides will use this threshold

Setting Agent-Level Thresholds

Go to Hive → Agents → click the agent
Click Thresholds
Click Add Threshold or edit an existing one
Configure values specific to this agent

Alert Severity Levels

Level	Color	Meaning
Info	Blue	Informational, no action required
Warning	Yellow	Approaching a limit, monitor closely
Error	Orange	Exceeding threshold, action recommended
Critical	Red	Severe violation, immediate action required

Alert Counts

Each agent card shows alert counts by severity. The counts update in real time as alerts fire and resolve.

Notification Channels

Alerts are delivered through your configured notification channels. Go to Settings → Integrations → Alerts to connect:

Slack
PagerDuty
Email
Webhook

Thresholds respect the notifyOnWarning and notifyOnCritical toggles — you can send warnings to Slack and critical alerts to PagerDuty simultaneously.

Example: CPU Threshold

Metric:     cpu
Comparison: greater_than
Warning:    70%
Critical:   90%
Duration:   120 seconds

With this threshold:

CPU > 70% for 2 minutes → warning alert
CPU > 90% for 2 minutes → critical alert
CPU drops below threshold → alert auto-resolves

Troubleshooting

Alert fires too often (flapping)

Increase the Duration field. A duration of 60–300 seconds filters out brief spikes and only fires when the condition is sustained.

No notifications received

Check that a notification channel is connected in Settings → Integrations → Alerts and that notifyOnWarning / notifyOnCritical are enabled on the threshold.

Agent shows critical alert but metric looks normal

Alerts resolve when the metric drops below the threshold for the required duration. If the alert persists but the current metric looks fine, the resolution window may not have elapsed yet.