feat: implement centralized alerting with vmalert and Grafana, add alert rules for CPU, disk, inode, RAM usage, and host status
This commit is contained in:
@@ -89,32 +89,73 @@ in
|
||||
};
|
||||
provision = {
|
||||
alerting = {
|
||||
contactPoints.settings = {
|
||||
apiVersion = 1;
|
||||
|
||||
contactPoints = [{
|
||||
orgId = 1;
|
||||
name = "cp_dominik";
|
||||
receivers = [{
|
||||
uid = "dominik";
|
||||
type = "pushover";
|
||||
settings = {
|
||||
security.apiToken = "$__file{${config.sops.secrets.pushover-api-token.path}}";
|
||||
security.userKey = "$__file{${config.sops.secrets.pushover-user-key.path}}";
|
||||
apiToken = "\${PUSHOVER_API_TOKEN}";
|
||||
userKey = "\${PUSHOVER_USER_KEY}";
|
||||
device = "iphone";
|
||||
priority = "2";
|
||||
retry = "30";
|
||||
expire = "120";
|
||||
sound = "siren";
|
||||
okSound = "magic";
|
||||
message = ''
|
||||
{{ template "default.message" . }}
|
||||
'';
|
||||
};
|
||||
contactPoints = {
|
||||
settings = {
|
||||
apiVersion = 1; # As per Grafana provisioning API
|
||||
contactPoints = [{
|
||||
orgId = 1;
|
||||
name = "cp_dominik";
|
||||
receivers = [{
|
||||
uid = "dominik_pushover_cp_receiver"; # Made UID even more specific
|
||||
type = "pushover";
|
||||
settings = {
|
||||
apiToken = "\${PUSHOVER_API_TOKEN}";
|
||||
userKey = "\${PUSHOVER_USER_KEY}";
|
||||
device = "iphone";
|
||||
priority = 2;
|
||||
retry = "30s";
|
||||
expire = "2m";
|
||||
sound = "siren";
|
||||
okSound = "magic";
|
||||
message = ''
|
||||
{{ template "default.message" . }}
|
||||
'';
|
||||
};
|
||||
}];
|
||||
}];
|
||||
}];
|
||||
};
|
||||
};
|
||||
|
||||
policies = { # Corrected from notificationPolicies to policies
|
||||
settings = {
|
||||
apiVersion = 1; # As per Grafana provisioning API
|
||||
|
||||
# Grafana's new unified alerting expects a single policy tree per org.
|
||||
# For OrgID 1 (default), this defines the root of that tree.
|
||||
# The NixOS module should translate this into the correct YAML structure.
|
||||
# The `policies` attribute within `settings` usually takes a list of policy trees.
|
||||
# For a single default organization, we define one policy tree.
|
||||
# Grafana's own YAML examples show a top-level 'route' for the default policy,
|
||||
# or a list under 'policies' if you're managing multiple policy sets (less common for basic setup).
|
||||
# Given the NixOS option `services.grafana.provision.alerting.policies.settings.policies`,
|
||||
# it's likely expecting a list here.
|
||||
policies = [{ # This outer list corresponds to the `policies` option
|
||||
# orgId = 1; # Usually implicit for the default policy file, but can be specified
|
||||
receiver = "cp_dominik"; # This sets the default receiver for the root route
|
||||
|
||||
# The actual routing tree starts here.
|
||||
# For a simple setup where all alerts go to one receiver,
|
||||
# just setting the top-level 'receiver' is often enough.
|
||||
# If more complex routing is needed, 'routes' would be defined here.
|
||||
# Example:
|
||||
# route = {
|
||||
# receiver = "cp_dominik";
|
||||
# group_by = [ "alertname", "job" ];
|
||||
# # ... other root route settings
|
||||
# routes = [
|
||||
# {
|
||||
# matcher_re = { severity = "critical" };
|
||||
# receiver = "critical_alerts_receiver"; # Another contact point
|
||||
# continue = false;
|
||||
# },
|
||||
# # ... other specific routes
|
||||
# ];
|
||||
# };
|
||||
# For the simplest case, just defining the receiver at this level should work
|
||||
# as the root policy for the default organization.
|
||||
}];
|
||||
# resetPolicies = false; # Default, set to true to remove existing policies not in this config.
|
||||
};
|
||||
};
|
||||
};
|
||||
};
|
||||
|
||||
Reference in New Issue
Block a user