Cloonar/nixos
2
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

refactor: Grafana alerting rules: consolidate and reorganize alert definitions

Browse Source 
- Deleted individual alert files for host down, inode usage, and RAM usage.
- Merged service down alerts into a new structure with separate files for each service (Gitea, Postfix, Dovecot, OpenLDAP, WireGuard).
- Introduced a new system alert structure consolidating CPU, disk, host down, inode, and RAM usage alerts.
- Updated alert conditions to use 'D' for thresholds and adjusted expressions accordingly.
- Improved annotations and labels for clarity and consistency across alerts.
This commit is contained in:
Dominik Polakovics Polakovics
2025-05-31 21:14:36 +02:00
parent 39b9726be7
commit b6b90bca7d
15 changed files with 359 additions and 330 deletions
Download Patch File Download Diff File Expand all files Collapse all files

58
hosts/web-arm/modules/grafana/alerting/system/cpu_usage.nix Normal file
View File

@@ -0,0 +1,58 @@
{ lib, pkgs, config, ... }:
{
grafanaAlertRuleDefinitions = [
{
uid = "high-cpu-usage-alert-uid";
title = "HighCPUUsage";
condition = "D"; # Condition is now D
data = [
# Query A: Calculate CPU usage percentage
{
refId = "A";
datasourceUid = "vm-datasource-uid";
queryType = "prometheus";
relativeTimeRange = { from = 60; to = 0; }; # Query over the last minute
model = {
# Calculate average CPU usage over 1m, grouped by instance and job
expr = ''(1 - avg by (instance, job) (rate(node_cpu_seconds_total{mode="idle"}[1m]))) * 100'';
legendFormat = "CPU usage on {{instance}} ({{job}})";
instant = false; # This is a range query
};
}
# Expression C: Reduce Query A to its last value, preserving labels
{
refId = "C";
datasourceUid = "__expr__";
model = {
type = "reduce";
expression = "A"; # Input is Query A
reducer = "last"; # Get the last value of each series in A
};
}
# Expression D: Apply math condition to the reduced values from C
{
refId = "D";
datasourceUid = "__expr__";
model = {
type = "math";
expression = "$C > 90"; # Alert if CPU usage from C is > 90%
};
}
];
for = "5m"; # Duration the condition must be met
noDataState = "NoData";
execErrState = "Error";
annotations = {
summary = "High CPU usage on {{ $labels.instance }}";
description = ''CPU usage on {{ $labels.instance }} (job: {{ $labels.job }}) has been above 90% for more than 5 minutes. Current value: {{ if $values.C }}{{ $values.C | humanizePercentage }}{{ else }}N/A{{ end }}%.'';
};
labels = {
severity = "warning";
category = "performance";
};
}
];
}

10
hosts/web-arm/modules/grafana/alerting/system/default.nix
View File

@@ -1,11 +1,11 @@
{ lib, pkgs, config, ... }:
let
# Import rule definitions from refactored alert files in the parent 'alerting' directory
cpuAlertRules = (import ../cpu_usage.nix { inherit lib pkgs config; }).grafanaAlertRuleDefinitions;
diskAlertRules = (import ../disk_usage.nix { inherit lib pkgs config; }).grafanaAlertRuleDefinitions;
hostDownAlertRules = (import ../host_down.nix { inherit lib pkgs config; }).grafanaAlertRuleDefinitions;
inodeAlertRules = (import ../inode_usage.nix { inherit lib pkgs config; }).grafanaAlertRuleDefinitions;
ramAlertRules = (import ../ram_usage.nix { inherit lib pkgs config; }).grafanaAlertRuleDefinitions;
cpuAlertRules = (import ./cpu_usage.nix { inherit lib pkgs config; }).grafanaAlertRuleDefinitions;
diskAlertRules = (import ./disk_usage.nix { inherit lib pkgs config; }).grafanaAlertRuleDefinitions;
hostDownAlertRules = (import ./host_down.nix { inherit lib pkgs config; }).grafanaAlertRuleDefinitions;
inodeAlertRules = (import ./inode_usage.nix { inherit lib pkgs config; }).grafanaAlertRuleDefinitions;
ramAlertRules = (import ./ram_usage.nix { inherit lib pkgs config; }).grafanaAlertRuleDefinitions;
allSystemRules = cpuAlertRules ++ diskAlertRules ++ hostDownAlertRules ++ inodeAlertRules ++ ramAlertRules;
in

76
hosts/web-arm/modules/grafana/alerting/system/disk_usage.nix Normal file
View File

@@ -0,0 +1,76 @@
{ lib, pkgs, config, ... }:
{
grafanaAlertRuleDefinitions = [
{
uid = "high-disk-usage-alert-uid"; # Optional: provide a stable UID for the rule itself
title = "HighDiskUsage"; # Name of the alert rule (was 'alert' in vmalert)
# Condition for the alert to fire. 'D' refers to the refId of the threshold expression.
condition = "D"; # Condition is now D
# Removed rule-level relativeTimeRange
# Data queries and expressions
data = [
# Query A: Calculate disk usage percentage
{
refId = "A";
datasourceUid = "vm-datasource-uid"; # UID of the VictoriaMetrics datasource
queryType = "prometheus"; # Explicitly set, though often inferred
relativeTimeRange = { from = 60; to = 0; }; # Query-level, integer seconds
model = {
expr = ''
(
node_filesystem_size_bytes{fstype!~"tmpfs|rootfs",mountpoint!=""} - node_filesystem_avail_bytes{fstype!~"tmpfs|rootfs",mountpoint!=""}
) / (node_filesystem_size_bytes{fstype!~"tmpfs|rootfs",mountpoint!=""} > 0) * 100
and node_filesystem_size_bytes{fstype!~"tmpfs|rootfs",mountpoint!=""}
and node_filesystem_avail_bytes{fstype!~"tmpfs|rootfs",mountpoint!=""}
'';
legendFormat = "{{mountpoint}} on {{instance}}"; # Example legend
instant = false; # For range queries, default is false
};
}
# Expression C: Reduce Query A to its last value, preserving labels
{
refId = "C";
datasourceUid = "__expr__";
model = {
type = "reduce";
expression = "A"; # Input is Query A
reducer = "last"; # Get the last value of each series in A
};
}
# Expression D: Apply math condition to the reduced values from C
{
refId = "D";
datasourceUid = "__expr__";
model = {
type = "math";
expression = "$C > 85"; # Check if the last value from each series in C is > 85
};
}
];
for = "15m"; # Duration the condition must be met (same as vmalert)
# How to handle states where data is missing or query errors
noDataState = "NoData"; # Options: NoData, Alerting, OK
execErrState = "Error"; # Options: Error, Alerting, OK
annotations = {
summary = "High disk usage on {{ $labels.instance }} at {{ $labels.mountpoint }}";
description = ''
Disk usage on {{ $labels.instance }} for mount point {{ $labels.mountpoint }}
(fstype: {{ $labels.fstype }}) has been above 85% for more than 15 minutes.
Current value: {{ if $values.C }}{{ $values.C | humanizePercentage }}{{ else }}N/A{{ end }}%.
''; # Using $values.C as it's the input to the math condition D
};
labels = {
severity = "warning";
category = "capacity";
# Grafana automatically adds labels from the query result (instance, mountpoint, etc.)
# and labels from the rule group/folder.
};
# isPaused = false; # Default is not paused
}
];
}

54
hosts/web-arm/modules/grafana/alerting/system/host_down.nix Normal file
View File

@@ -0,0 +1,54 @@
{ lib, pkgs, config, ... }:
{
grafanaAlertRuleDefinitions = [
{
uid = "host-down-alert-uid";
title = "HostDown";
condition = "C";
data = [
{
refId = "A";
datasourceUid = "vm-datasource-uid";
queryType = "prometheus";
relativeTimeRange = { from = 60; to = 0; }; # Query over the last minute
model = {
expr = ''up'';
legendFormat = "{{instance}} ({{job}})";
instant = false; # Changed from true, as relativeTimeRange is used
};
}
{ # New Expression B: Reduce Query A
refId = "B";
datasourceUid = "__expr__";
model = {
type = "reduce";
expression = "A"; # Input is Query A
reducer = "last"; # Get the last value of each series in A
};
}
{ # Modified Expression C: Math condition based on B
refId = "C";
datasourceUid = "__expr__";
model = {
type = "math";
expression = "$B == 0"; # Check if the last value from B is 0
};
}
];
for = "2m";
noDataState = "Alerting";
execErrState = "Error";
annotations = {
summary = "Host {{ $labels.instance }} is down";
description = ''Host {{ $labels.instance }} (job: {{ $labels.job }}) has been down for more than 2 minutes.'';
};
labels = {
severity = "critical";
category = "availability";
};
}
];
}

63
hosts/web-arm/modules/grafana/alerting/system/inode_usage.nix Normal file
View File

@@ -0,0 +1,63 @@
{ lib, pkgs, config, ... }:
{
grafanaAlertRuleDefinitions = [
{
uid = "high-inode-usage-alert-uid";
title = "HighInodeUsage";
condition = "D"; # Condition is now D
data = [
# Query A: Calculate inode usage percentage
{
refId = "A";
datasourceUid = "vm-datasource-uid";
queryType = "prometheus";
relativeTimeRange = { from = 60; to = 0; };
model = {
expr = ''
(
node_filesystem_files{fstype!~"tmpfs|rootfs",mountpoint!=""} - node_filesystem_files_free{fstype!~"tmpfs|rootfs",mountpoint!=""}
) / (node_filesystem_files{fstype!~"tmpfs|rootfs",mountpoint!=""} > 0) * 100
and node_filesystem_files{fstype!~"tmpfs|rootfs",mountpoint!=""}
and node_filesystem_files_free{fstype!~"tmpfs|rootfs",mountpoint!=""}
'';
legendFormat = "{{mountpoint}} on {{instance}}";
instant = false;
};
}
# Expression C: Reduce Query A to its last value, preserving labels
{
refId = "C";
datasourceUid = "__expr__";
model = {
type = "reduce";
expression = "A"; # Input is Query A
reducer = "last"; # Get the last value of each series in A
};
}
# Expression D: Apply math condition to the reduced values from C
{
refId = "D";
datasourceUid = "__expr__";
model = {
type = "math";
expression = "$C > 80"; # Alert if inode usage from C is > 80%
};
}
];
for = "30m"; # Duration the condition must be met
noDataState = "NoData";
execErrState = "Error";
annotations = {
summary = "High inode usage on {{ $labels.instance }} at {{ $labels.mountpoint }}";
description = ''Inode usage on {{ $labels.instance }} for mount point {{ $labels.mountpoint }} (fstype: {{ $labels.fstype }}) has been above 80% for more than 30 minutes. Current value: {{ if $values.C }}{{ $values.C | humanizePercentage }}{{ else }}N/A{{ end }}%.'';
};
labels = {
severity = "warning";
category = "capacity";
};
}
];
}

61
hosts/web-arm/modules/grafana/alerting/system/ram_usage.nix Normal file
View File

@@ -0,0 +1,61 @@
{ lib, pkgs, config, ... }:
{
grafanaAlertRuleDefinitions = [
{
uid = "high-ram-usage-alert-uid";
title = "HighRAMUsage";
condition = "D"; # Condition is now D
data = [
# Query A: Calculate RAM usage percentage
{
refId = "A";
datasourceUid = "vm-datasource-uid";
queryType = "prometheus";
relativeTimeRange = { from = 60; to = 0; };
model = {
expr = ''
(1 - node_memory_MemAvailable_bytes / (node_memory_MemTotal_bytes > 0)) * 100
and node_memory_MemAvailable_bytes
and node_memory_MemTotal_bytes
'';
legendFormat = "RAM usage on {{instance}} ({{job}})";
instant = false;
};
}
# Expression C: Reduce Query A to its last value, preserving labels
{
refId = "C";
datasourceUid = "__expr__";
model = {
type = "reduce";
expression = "A"; # Input is Query A
reducer = "last"; # Get the last value of each series in A
};
}
# Expression D: Apply math condition to the reduced values from C
{
refId = "D";
datasourceUid = "__expr__";
model = {
type = "math";
expression = "$C > 90"; # Alert if RAM usage from C is > 90%
};
}
];
for = "10m"; # Duration the condition must be met
noDataState = "NoData";
execErrState = "Error";
annotations = {
summary = "High RAM usage on {{ $labels.instance }}";
description = ''RAM usage on {{ $labels.instance }} (job: {{ $labels.job }}) has been above 90% for more than 10 minutes. Current value: {{ if $values.C }}{{ $values.C | humanizePercentage }}{{ else }}N/A{{ end }}%.'';
};
labels = {
severity = "warning";
category = "performance";
};
}
];
}