#Configure Health Alerts

:::tip

Don't want to write alert configs manually? Alerts Automation uses AI to suggest alerts, generate the configuration, and test it against historical data — so you can validate thresholds before deployment without learning the syntax. You can also use the visual Alerts Configuration Manager.

:::

This page covers manual alert configuration — editing config files directly. For a comparison of all configuration methods, see How alert configuration works in Netdata.

#Quick Start Guide

:::tip

What You'll Learn

In 5 minutes, you'll know how to edit existing alerts, create new ones, and reload your configuration without downtime.

:::

#Get Started in 3 Steps

Step 1: Find Your Config Directory

Navigate to your Netdata config directory

Step 2: Edit an Alert

sudo ./edit-config health.d/cpu.conf

Step 3: Apply Changes

sudo netdatacli reload-health

:::note

Key Concept

You can highly configure Netdata's health watchdog with support for dynamic thresholds, hysteresis, alert templates, and more. You can customize any existing alerts based on your infrastructure's topology or specific monitoring needs or create entirely new entities.

You can use health alerts with any of Netdata's collectors (see the supported collector list) to monitor your systems, containers, and applications in real time.

While you can view active alerts on both the local dashboard and Netdata Cloud, you configure all health alerts per node via individual Netdata Agents. If you want to deploy a new alert across your infrastructure, you must configure each node with the same health configuration files.

:::

Next Steps: Jump to Common Tasks for specific workflows or continue reading for comprehensive guidance.

#Common Tasks

:::tip

What You'll Learn

Step-by-step workflows for the most frequent alert configuration tasks.

:::

#Task 1: Modify Alert Thresholds

Why This Matters: Default thresholds may not fit your specific environment or requirements.

Quick Example:

# Change CPU warning from 85% to 75%
warn: $this > (($status >= $WARNING) ? (60) : (75))
crit: $this > (($status == $CRITICAL) ? (75) : (85))

Step-by-Step:

1. Find the alert file: sudo ./edit-config health.d/cpu.conf
2. Locate the alert (e.g., 10min_cpu_usage)
3. Modify warn and crit lines
4. Save and reload: sudo netdatacli reload-health

#Task 2: Disable Unwanted Alerts

#Task 3: Create a Simple Alert

Real-World Example: Monitor RAM usage above 80%

alarm: ram_usage
   on: system.ram
lookup: average -1m percentage of used
units: %
every: 1m
 warn: $this > 80
 crit: $this > 90
 info: RAM usage monitoring

Next Steps: See How-To Guides for detailed explanations.

#How-To Guides

:::tip

What You'll Learn

Detailed instructions for configuring, managing, and troubleshooting health alerts.

:::

#How to Reload Health Configuration

Why This Matters: You don't need to restart your Netdata Agent when making changes, preventing gaps in monitoring.

You don't need to restart your Netdata Agent when making changes to health configuration files, such as specific health entities. Instead, you can use netdatacli with the reload-health option to prevent gaps in metrics collection.

sudo netdatacli reload-health

Alternative Method: If netdatacli doesn't work on your system, you can send a SIGUSR2 signal to the daemon, which reloads health configuration without restarting the entire process.

sudo killall -USR2 netdata

#How to Edit Health Configuration Files

Configuration Locations:

Configuration Locations:

Navigate to your Netdata config directory and use edit-config to make changes to any of these files.

Edit Individual Alerts:

For example, to edit the cpu.conf health configuration file, run:

sudo ./edit-config health.d/cpu.conf

Understanding Alert Structure:

Each health configuration file contains one or more health entities, which always begin with alarm: or template:. Here's the first health entity in health.d/cpu.conf:

 template: 10min_cpu_usage
       on: system.cpu
    class: Utilization
     type: System
component: CPU
   lookup: average -10m unaligned of user,system,softirq,irq,guest
    units: %
    every: 1m
     warn: $this > (($status >= $WARNING)  ? (75) : (85))
     crit: $this > (($status == $CRITICAL) ? (85) : (95))
    delay: down 15m multiplier 1.5 max 1h
  summary: CPU utilization
     info: Average cpu utilization for the last 10 minutes (excluding iowait, nice and steal)
       to: sysadmin

To customize this alert to trigger warning and critical alerts at lower CPU utilization levels, you can change the warn and crit lines to values of your choosing. For example:

    warn: $this > (($status >= $WARNING)  ? (60) : (75))
    crit: $this > (($status == $CRITICAL) ? (75) : (85))

Save the file and reload Netdata's health configuration to apply your changes.

#How to Disable or Silence Alerts

Why This Matters: Different situations require different approaches to managing alerts - permanent removal, temporary silencing, or selective filtering.

You can disable alerts and notifications permanently via configuration changes, or temporarily via the health management API.

#Disable All Alerts

Use Case: System maintenance or testing

In the netdata.conf [health] section, set enabled to no, and restart your Agent.

#Disable Specific Alerts

Use Case: Remove known noisy or irrelevant alerts

In the netdata.conf [health] section, use pattern exclusion with enabled alarms = !oom_kill * to load all alerts except oom_kill.

To exclude multiple specific alerts, list all exclusions before the wildcard and restart the Agent to apply the netdata.conf change:

[health]
    enabled alarms = !oom_kill !disk_space_usage *

Restart your Netdata Agent after changing netdata.conf (netdatacli reload-health reloads health configuration files, but does not reload netdata.conf).

:::warning

Do not place * between exclusions (!alert1 * !alert2 * is incorrect — patterns are evaluated in order and the first match wins). When using exclusions, put all ! patterns first, followed by a single trailing *.

:::

You can also edit the file where the alert is defined, comment out its definition, and reload Netdata's health configuration.

#Silence Individual Alert Notifications

Use Case: Keep monitoring active but stop notifications

You can stop receiving notifications for an individual alert by changing the to: line to silent in the alert's configuration file.

      to: silent

:::tip

This action requires that you reload Netdata's health configuration.

:::

#Temporary Runtime Control

Use Case: Scheduled maintenance or dynamic control

:::tip

You can use the health management API to temporarily control alert behavior without changing configuration or restarting your Agent. The API allows you to:

• Disable all or some alerts from triggering during certain times (for instance, when running backups)
• Suppress notifications temporarily while keeping health checks running and alerts triggering

:::

#How to Write a New Health Entity

Why This Matters: While tuning existing alerts may work in some cases, you may need to write entirely new health entities based on how your systems, containers, and applications work.

Prerequisites: Read the Alert Configuration Reference for a complete listing of the format, syntax, and functionality of health entities.

Step-by-Step Process:

Step 1: Create the Configuration File

Navigate to your Netdata config directory, then use touch to create a new file in the health.d/ directory. Use edit-config to start editing the file.

As an example, let's create a ram-usage.conf file:

sudo touch health.d/ram-usage.conf
sudo ./edit-config health.d/ram-usage.conf

Step 2: Write Your Alert

Here's a health entity that triggers a warning alert when your node's RAM usage rises above 80%, and a critical alert above 90%:

 alarm: ram_usage
    on: system.ram
lookup: average -1m percentage of used
 units: %
 every: 1m
  warn: $this > 80
  crit: $this > 90
  info: The percentage of RAM being used by the system.

Step 3: Understand Each Component

:::note

Understanding This Example

This health entity, named ram_usage, watches the system.ram chart. It looks up the last 1 minute of metrics from the used dimension and calculates the average of all those metrics in a percentage format, using % units. The entity performs this lookup every minute.

If the average RAM usage percentage over the last 1 minute is more than 80%, the entity triggers a warning alert. If the usage is more than 90%, the entity triggers a critical alert.

:::

Step 4: Activate Your Alert

When you finish writing this new health entity, reload Netdata's health configuration to see it live on your local dashboard or Netdata Cloud.

Next Steps: Explore Alert Examples for more complex scenarios, or dive into the Alert Configuration Reference for complete syntax details.

#Alert Configuration Reference

:::tip

What You'll Learn

Complete syntax reference for all alert configuration options. Use this section when you need specific technical details.

:::

#Entity Types Overview

Alerts are attached to specific charts and use the alarm label.

Templates define rules that apply to all charts of a specific context, and use the template label. Templates help you apply one entity to all disks, all network interfaces, all MySQL databases, and so on.

:::note

Precedence

Alarms are processed before templates. If you have alarm and template entities with the same name that both match the same chart, only the alarm will create an active alert for that chart.

For complete details on configuration loading order and precedence rules, see Alert Configuration Ordering.

:::

#Required vs Optional Configuration

:::note

Configuration Requirements

• The alarm or template line must be the first line of any entity
• The on line is always required
• The every line is required if not using lookup
• Each entity must have at least one of the following lines: lookup, calc, warn, or crit

:::

Special Syntax Rules:

• A few lines use space-separated lists to define how the entity behaves. You can use * as a wildcard or prefix with ! for a negative match. Order is important! See our simple patterns docs for more examples
• Lines terminated by a \ are spliced together with the next line. The backslash is removed, and the following line is joined with the current one. No space is inserted, so you can split a line anywhere, even in the middle of a word. This is handy if your info line consists of several sentences

#Complete Configuration Reference

#Configuration Line Details

#Alert Line alarm or template

Purpose: This line starts an alert or template based on the entity type you want to create.

Alert Syntax:

alarm: NAME

Template Syntax:

template: NAME

Naming Rules:

• NAME can be any alphanumeric character
• Only . (period) and _ (underscore) symbols allowed
• Can’t be chart name, dimension name, family name, or chart variable names

#Alert Line on

Purpose: This line defines the chart this alert should attach to.

For Alerts:

on: CHART

The value CHART should be the unique ID or name of the chart you're interested in, as shown on the dashboard. In the image below, the unique ID is system.cpu.

For Templates:

on: CONTEXT

The value CONTEXT should be the context you want this template to attach to.

:::tip

Finding the Context

Need to find the context? Hover over the date on any given chart and look at the tooltip. In the image below, which shows a disk I/O chart, the tooltip reads: proc:/proc/diskstats, disk.io.

You're interested in what comes after the comma: disk.io. That's the name of the chart's context.

If you create a template using the disk.io context, it will apply an alert to every disk available on your system.

:::

#Alert Line class

Purpose: This indicates the type of error (or general problem area) that the alert or template applies to.

Example Use: Latency can be used for alerts that trigger on latency issues on network interfaces, web servers, or database systems.

class: Latency

Available Classes:

:::note

class will default to Unknown if the line is missing from the alert configuration.

:::

#Alert Line type

Purpose: You can use type to indicate the broader area of the system that the alert applies to.

Example: Under the general Database type, you can group together alerts that operate on various database systems, like MySQL, CockroachDB, CouchDB, etc.

type: Database

Available Types:

:::note

If an alert configuration is missing the type line, its value will default to Unknown.

:::

#Alert Line component

Purpose: You can use component to narrow down what the previous type value specifies for each alert or template.

Example: Continuing from the previous example, component might include MySQL, CockroachDB, MongoDB, all under the same Database type.

component: MySQL

:::note

As with the class and type lines, if component is missing from the configuration, its value will default to Unknown.

:::

#Alert Line lookup

Purpose: This line makes a database lookup to find a value. The result of this lookup is available as $this.

Full Syntax:

lookup: METHOD(GROUPING OPTIONS) AFTER [at BEFORE] [every DURATION] [OPTIONS] [of DIMENSIONS]

Required Parameters:

Optional Parameters:

Processing Options:

When anomaly-bit is used, each data point returns the anomaly rate as a percentage from 0 to 100. At native resolution this is typically 0 (normal) or 100 (anomalous), while aggregated or lower-resolution data can yield intermediate values such as 12.5. For more details and practical examples, see the ML anomaly detection documentation.

Example:

lookup: average -10m unaligned of user,system,softirq,irq,guest

This looks back 10 minutes, calculates the average of the specified CPU dimensions, without aligning to time boundaries.

The result of the lookup will be available as $this and $NAME in expressions. The timestamps of the timeframe evaluated by the database lookup are available as variables $after and $before (both are unix timestamps).

#Alert Line calc

Purpose: You can design a calc to apply some calculation to the values or variables available to the entity.

Key Points:

• The result becomes available as $this variable
• Overwrites the value from your lookup
• Can be used without lookup if using other available variables
• Uses expressions for syntax

calc: EXPRESSION

When to Use:

• With lookup: Perform calculation after database retrieval
• Without lookup: When using other available variables
• For complex logic: Mathematical operations, conditions, transformations

#Alert Line every

Purpose: Sets the update frequency of this alert.

every: DURATION

Supported Units:

• s for seconds
• m for minutes
• h for hours
• d for days

Example: every: 30s checks the alert every 30 seconds.

#Alert Lines green and red

Purpose: Set the green and red thresholds of a chart for visualization.

green: NUMBER
red: NUMBER

Important Notes:

• Both values are available as $green and $red in expressions
• If multiple alerts define different thresholds, the first alert's values are used
• For multiple threshold sets, use absolute numbers instead of variables

#Alert Lines warn and crit

Purpose: Define the expressions that trigger warning or critical alerts.

warn: EXPRESSION
crit: EXPRESSION

Key Points:

• Optional (but you need at least one)
• Should evaluate to true/false (or zero/non-zero)
• Uses Netdata's expression syntax
• Can reference variables like $this, $green, $red

Examples:

warn: $this > 80
crit: $this > 95

#Alert Line to

Purpose: Specifies who receives notifications when the alert changes status.

to: ROLE1 ROLE2 ROLE3 ...

How It Works:

• First parameter passed to the exec script
• Default script (alarm-notify.sh) treats this as a space-separated list of roles
• Roles are consulted to find exact recipients per notification method

#Alert Line exec

Purpose: Script to execute when the alert status changes.

exec: SCRIPT

Default Behavior:

• Default script is Netdata's alarm-notify.sh
• Supports all notification methods Netdata supports
• Includes custom hooks

#Alert Line delay

Purpose: Provide optional hysteresis settings to prevent notification floods.

:::important

These settings don't affect the actual alert - only when the exec script is executed.

:::

Full Syntax:

delay: [[[up U] [down D] multiplier M] max X]

Parameters:

Example with Timeline:

delay: up 10s down 15m multiplier 2 max 1h

Starting at 00:00:00 with CLEAR status:

#Alert Line repeat

Purpose: Defines the interval between repeating notifications for alerts in CRITICAL or WARNING mode.

repeat: [off] [warning DURATION] [critical DURATION]

Options:

Why Use This: Overrides default repeat settings from netdata.conf health configuration.

#Alert Line options

Purpose: Special alert behavior options.

options: no-clear-notification

Available Options:

• no-clear-notification - Prevents clearing the alert notification

When to Use no-clear-notification:

• Alerts comparing two time frames (e.g., last 3 minutes vs last hour)
• When newer data might "pollute" the baseline comparison
• When clearing conditions are unreliable due to data characteristics

Example Use Case: HTTP response time alert comparing recent average to historical average - as time passes, the recent slow responses become part of the historical data, making the alert appear "cleared" even though the underlying issue wasn't resolved.

#Alert Line host labels

Purpose: Restricts alerts to hosts with matching labels.

Prerequisites: See our host labels guide for setup instructions.

Example Configuration:

[host labels]
    installed = 20191211
    room = server

Usage in Alerts:

host labels: room = server

Pattern Support:

host labels: installed = 201*  # Matches all hosts installed in 2010s

How It Works:

• Space-separated list
• Accepts simple patterns
• Alert only loads on matching hosts

#Alert Line chart labels

Purpose: Filters alerts based on chart labels.

How to Find Chart Labels: Check http://localhost:19999/api/v1/charts?all

Example Use Case: Each disk_space chart has a mount_point label. To exclude external disk alerts:

chart labels: mount_point=!/mnt/disk1 *

Multiple Label Logic:

chart labels: mount_point=/mnt/disk1 device=sda

This requires BOTH conditions to be true (AND logic).

:::important

• Space-separated list with simple patterns support
• If a specified label doesn't exist on the chart, the chart won't match
• Multiple labels use AND logic
• Alerts based on chart labels require the underlying chart to exist. For example, a disk.space chart is only created when a mount point is present and collected. For example, if a CIFS mount fails to mount after a system reboot, no disk.space chart will exist for that mount point, and the alert will not activate

:::

#Alert Line summary

Purpose: Brief title of the alert used in notifications and dashboard.

summary: Available Ram

Variable Support:

Example with Variables:

summary: 1 minute received traffic overflow for ${label:device}

Renders as: 1 minute received traffic overflow for eth0

:::note

Variable names are case-sensitive.

:::

#Alert Line info

Purpose: Detailed description of the alert for notifications and UI elements.

info: Percentage of estimated amount of RAM available for userspace processes, without causing swapping

Variable Support:

Examples with Variables:

Family Variable:

info: average inbound utilization for the network interface ${family} over the last minute

Renders as: average inbound utilization for the network interface eth0 over the last minute

Label Variable:

info: average ratio of HTTP responses with unexpected status over the last 5 minutes for the site ${label:target}

Renders as: average ratio of HTTP responses with unexpected status over the last 5 minutes for the site https://netdata.cloud/

Next Steps: Continue to Expressions and Variables to understand the calculation syntax, or jump to Alert Examples for practical implementations.

#Expressions and Variables

:::tip

What You'll Learn

How to write calculations and use variables in your alert definitions. Essential for creating custom logic and accessing chart data.

:::

#Expressions Overview

Why This Matters: Netdata has an internal infix expression parser that allows you to create complex alert logic using mathematical operations, comparisons, and conditional statements.

Supported Operators:

Special Functions:

• abs() - Absolute value
• (condition) ? (true_expr) : (false_expr) - Conditional operator

Special Values:

#Conditional Operator for Hysteresis

Why This Matters: The conditional operator (? :) can create "sticky" alert thresholds that prevent alert spam when values fluctuate around a threshold. This is called hysteresis.

Basic Pattern:

warn: $this > (($status >= $WARNING) ? (lower_threshold) : (higher_threshold))

Real Example - CPU Usage Alert:

warn: $this > (($status >= $WARNING)  ? (75) : (85))
crit: $this > (($status == $CRITICAL) ? (85) : (95))

How This Works:

Benefits:

• Quick alerting when issues arise
• Protection against spam when values hover near thresholds
• Single initial notification instead of constant alerts during fluctuation

Example Scenario: If CPU usage fluctuates between 80–90%, you'll receive just one initial warning, rather than constant notifications.

#Variables Reference

How to Find Available Variables: You can find all variables for a given chart using: http://NODE:19999/api/v1/alarm_variables?chart=CHART_NAME

Example: Variables for the system.cpu chart

:::note

Chart vs Template Variables

Although the alarm_variables link shows variables for a particular chart, the same variables can also be used in templates for charts belonging to a given context. All charts of a given context are essentially identical, with the only difference being the family that identifies a particular hardware or software instance.

:::

#Variable Categories

#Chart Local Variables

What's Available:

• All chart dimensions as variables (e.g., $user, $system for CPU chart)
• Values from other configured alerts on the same chart
• Special chart variables (see table below)

Special Chart Variables:

Dimension Value Types:

• Default: Last calculated (interpolated) value as shown on charts
• Raw suffix: $dimension_raw - Last collected value
• Timestamp suffix: $dimension_last_collected_t - Unix timestamp when dimension was last collected

#Host Variables

What's Available: All dimensions of all charts, including all alerts, in fullname format.

Format: CHART.VARIABLE

• CHART can be either chart ID or chart name
• Both formats are supported

Examples:

• $system.cpu.user - User CPU from system.cpu chart
• $disk.sda.reads - Read operations from sda disk chart

#Cross-Chart Variable Examples from Stock Health Entities

Several stock health configurations use host variables to reference dimensions from other charts in their calc, warn, and crit expressions.

#Prometheus Collector Variables

For metrics collected by the go.d prometheus collector, each unique Prometheus label set usually produces a separate chart. The chart ID is built from the metric name followed by -label=value pairs for every label (e.g. kubelet_volume_stats_used_bytes-persistentvolumeclaim=my-pvc); characters in a label value that are not chart-ID-safe, such as ., are replaced with _ in the chart ID, while the chart's label keeps the original value (so addr="10.0.0.1" yields …-addr=10_0_0_1). In the Netdata chart registry, the prefix comes from the go.d job FullName: it is prometheus.<metric_name>-<label_set> only when the job name is literally prometheus; otherwise it is prometheus_<job_name>.<metric_name>-<label_set> (for example, prometheus_local.<metric_name>-<label_set> or prometheus_kubelet.<metric_name>-<label_set>). Summary and histogram families also emit separate _sum and _count charts; the suffix is part of the metric name, so the IDs are <metric_name>_sum-<label_set> and <metric_name>_count-<label_set> (just <metric_name>_sum / <metric_name>_count when the series has no labels), while histogram buckets are dimensions of the base <metric_name> chart. Verify the exact chart ID you want to reference.

Because Prometheus chart IDs typically contain hyphens and = characters, use the ${...} brace form to reference them in calc/warn/crit expressions — the unbraced $var form stops parsing at -. Apply the same rule for both the common prometheus_<job_name> prefix and the special-case plain prometheus prefix, including any _sum or _count chart variants.

Example — PVC volume usage alert using kubelet metrics from a named Prometheus job (name: kubelet):

   alarm: kubelet_pvc_volume_usage
      on: prometheus_kubelet.kubelet_volume_stats_used_bytes-persistentvolumeclaim=my-pvc
   lookup: max -1m unaligned match-names of kubelet_volume_stats_used_bytes
     calc: $this * 100 / ${prometheus_kubelet.kubelet_volume_stats_capacity_bytes-persistentvolumeclaim=my-pvc.kubelet_volume_stats_capacity_bytes}
     warn: $this > 80
     crit: $this > 95
   units: %
    every: 1m
     info: PVC volume usage percentage
       to: sysadmin

:::note

The exact chart ID and dimension names depend on your endpoint's label sets. Use the alarm variables API to discover the correct names:

http://NODE:19999/api/v1/alarm_variables?chart=prometheus_kubelet.kubelet_volume_stats_used_bytes-persistentvolumeclaim%3Dmy-pvc

URL-encode the chart ID only in API query parameters. In alert expressions, use the chart ID as-is inside ${...} braces.

:::

#Special Variables

Alert Status Constants:

Status Comparison Examples:

# Check if alert is in warning or higher
warn: $status >= $WARNING

# Check if alert is specifically critical
crit: $status == $CRITICAL

:::note

Status values increase with severity, so $status > $CLEAR will match both WARNING and CRITICAL states.

:::

#Alert Status Lifecycle

Status Flow: UNINITIALIZED → UNDEFINED/CLEAR → WARNING → CRITICAL

When Status Changes:

• REMOVED - Alert deleted during configuration reload
• UNINITIALIZED - Alert created but not yet calculated
• UNDEFINED - Database lookup failed, division by zero, etc.
• CLEAR - Alert conditions aren’t met (normal state)
• WARNING - Warning expression returned true/non-zero
• CRITICAL - Critical expression returned true/non-zero

Script Execution: The external script (exec line) is called for ALL status changes.

Next Steps: Ready to see these concepts in action? Continue to Alert Examples for practical implementations.

#Alert Examples

:::tip

What You'll Learn

Real-world alert configurations that demonstrate different monitoring scenarios. Use these as templates for your own alerts.

:::

template: ml_5min_node
      on: anomaly_detection.anomaly_rate
  lookup: average -5m of anomaly_rate
    calc: $this
   units: %
   every: 30s
    warn: $this > (($status >= $WARNING)  ? (5) : (20))
    crit: $this >= (($status == $CRITICAL) ? (20) : (100))
    info: rolling 5min anomaly rate for all ML enabled dims

 alarm: service_failure_event
    on: my_service.health_status
lookup: average -10s of health_status
 every: 10s
  warn: $this > 0
   info: any failure detected in the last 10 seconds
    to: sysadmin

 alarm: service_any_downtime
    on: my_service.health_status
lookup: min -5m unaligned
 every: 10s
  crit: $this == 0
   info: metric dropped to 0 at some point in the last 5 minutes
    to: sysadmin

 alarm: service_continuous_outage
    on: my_service.health_status
lookup: max -5m unaligned
 every: 10s
  crit: $this == 0
   info: service was down for the entire last 5 minutes
    to: sysadmin

 alarm: service_failure_rate
    on: my_service.health_status
lookup: average -5m unaligned of health_status
 every: 1m
  warn: $this > 0.1
   crit: $this > 0.5
   info: failure rate exceeded threshold over the last 5 minutes
    to: sysadmin

 alarm: service_current_state
    on: my_service.health_status
  calc: $health_status
 every: 10s
  crit: $this == 0
   info: service is currently down
    to: sysadmin
 delay: down 5m

Next Steps: Having trouble with your alerts? Continue to Troubleshooting for debugging techniques.

#Troubleshooting

:::tip

What You'll Learn

How to debug alert issues, understand why alerts aren't working, and get detailed information about alert processing.

:::

#Find Chart and Context Information

Finding Chart Names: You can find chart information in two places:

Replace NODE with your server's IP address or hostname.

#Analyze Alert Expressions

Why This Matters: Understand how Netdata interprets your expressions and what values are being calculated.

Check Alert Processing: Visit http://NODE:19999/api/v1/alarms?all to see:

• Original expression as written in config
• Parsed expression with added parentheses showing evaluation flow
• Current alert status and values
• Available variables and their values

Expression Evaluation Flow: Netdata adds parentheses to show how it evaluates your expressions:

Your Expression:

warn: $this > 80 and $status >= $WARNING

Netdata's Interpretation:

warn: (($this > 80) and ($status >= $WARNING))

#Troubleshooting Decision Trees

:::note

How to Use These Decision Trees

Follow the flowcharts below to systematically diagnose and resolve alert issues. Each path leads to specific solutions with step-by-step instructions.

:::

#Decision Tree: Alert Not Working

flowchart TD
    A("Alert Not Working") --> B("Check Configuration")
    A --> C("Check Data Source")
    A --> D("Check Thresholds")

    B --> B1("Syntax errors<br/>Reload config<br/>Review logs")
    C --> C1("Chart exists<br/>Data collection<br/>Variable names")
    D --> D1("Adjust values<br/>Review logic<br/>Test expressions")

    %% Style definitions
    classDef alert fill:#ffeb3b,stroke:#000000,stroke-width:3px,color:#000000,font-size:18px
    classDef neutral fill:#f9f9f9,stroke:#000000,stroke-width:3px,color:#000000,font-size:18px
    classDef complete fill:#4caf50,stroke:#000000,stroke-width:3px,color:#000000,font-size:18px
    classDef database fill:#2196F3,stroke:#000000,stroke-width:3px,color:#000000,font-size:18px

    %% Apply styles
    class A alert
    class B,C,D database
    class B1,C1,D1 complete

#Decision Tree: Alert Always Triggering

flowchart TD
    A("Alert Always Triggering") --> B("Check Values")
    A --> C("Check Logic")
    A --> D("Check Variables")

    B --> B1("Current metrics<br/>Threshold comparison<br/>Units verification")
    C --> C1("Operator direction<br/>Expression syntax<br/>Status conditions")
    D --> D1("Variable names<br/>Dimension availability<br/>API verification")

    %% Style definitions
    classDef alert fill:#ffeb3b,stroke:#000000,stroke-width:3px,color:#000000,font-size:18px
    classDef neutral fill:#f9f9f9,stroke:#000000,stroke-width:3px,color:#000000,font-size:18px
    classDef complete fill:#4caf50,stroke:#000000,stroke-width:3px,color:#000000,font-size:18px
    classDef database fill:#2196F3,stroke:#000000,stroke-width:3px,color:#000000,font-size:18px

    %% Apply styles
    class A alert
    class B,C,D database
    class B1,C1,D1 complete

#Decision Tree: Alert Flapping

flowchart TD
    A("Alert Flapping") --> B("Add Hysteresis")
    A --> C("Smooth Data")
    A --> D("Adjust Timing")

    B --> B1("Conditional thresholds<br/>Different up/down values<br/>Status-based logic")
    C --> C1("Increase lookup time<br/>Use averages<br/>Reduce noise")
    D --> D1("Increase intervals<br/>Add delays<br/>Reduce frequency")

    %% Style definitions
    classDef alert fill:#ffeb3b,stroke:#000000,stroke-width:3px,color:#000000,font-size:18px
    classDef neutral fill:#f9f9f9,stroke:#000000,stroke-width:3px,color:#000000,font-size:18px
    classDef complete fill:#4caf50,stroke:#000000,stroke-width:3px,color:#000000,font-size:18px
    classDef database fill:#2196F3,stroke:#000000,stroke-width:3px,color:#000000,font-size:18px

    %% Apply styles
    class A alert
    class B,C,D database
    class B1,C1,D1 complete

#Common Issues and Solutions

#Issue: Alert Not Triggering

Possible Causes:

#Issue: Alert Always Triggering

Possible Causes:

#Issue: Alert Flapping

Possible Causes:

Hysteresis Example:

# Instead of simple threshold
warn: $this > 80

# Use hysteresis
warn: $this > (($status >= $WARNING) ? (75) : (80))

#Issue: Variables Not Found

Debug Steps:

1. Check Available Variables:

   http://NODE:19999/api/v1/alarm_variables?chart=CHART_NAME

1. Verify Chart Context:

   • For alarm: use chart name
   • For template: use chart context

2. Check Variable Syntax:

   • Chart local: $dimension_name
   • Host variables: $chart_name.dimension_name
   • Special variables: $this, $now, $status

#Testing Alert Changes

Safe Testing Process:

1. Create Test File:

   sudo touch health.d/test-alert.conf
   sudo ./edit-config health.d/test-alert.conf

1. Write Simple Alert:

   alarm: test_ram
      on: system.ram
   lookup: average -1m percentage of used
    every: 10s
     war: $this > 50  # Low threshold for testing
     info: Test alert - safe to ignore

1. Reload and Monitor:

   sudo netdatacli reload-health
   # Watch dashboard for test alert appearance

1. Remove When Done:

   sudo rm health.d/test-alert.conf
   sudo netdatacli reload-health

#Performance Considerations

Alert Impact on System:

Recommended Frequencies:

#Getting Help

Information to Provide:

When seeking help, include:

1. Alert Configuration:

   # Your complete alert definition

1. Chart Information:

   http://your-server:19999/api/v1/alarm_variables?chart=chart_name

1. Current Status:

   http://your-server:19999/api/v1/alarms?all

Community Resources:

• Netdata GitHub Issues
• Netdata Community Forum
• Netdata Discord

#Alert Notification Variables

The following variables are available in alert notification templates and custom notification scripts:

#Related Pages

• Alerts Automation - Create and tune alerts using natural language with AI assistance (no manual configuration needed)
• Alerts Configuration Manager - Visual UI wizard for creating alerts
• ML Anomaly Detection - Machine learning based anomaly detection for all metrics
• Alert Troubleshooting - AI-powered alert investigation and root-cause analysis

Next Steps: You now have comprehensive knowledge of Netdata health configuration. Start with the Quick Start Guide for immediate needs or dive into Common Tasks for specific workflows.