`pagerduty_alert_grouping_setting` Unexpectedly Removed or Reverted During Terraform Runs

[nmistry24]

Terraform Version
Terraform v1.11.4

PagerDuty Provider Version
v3.25.0

---

Affected Resource(s)

• pagerduty_alert_grouping_setting

---

Terraform Configuration Files

resource "pagerduty_alert_grouping_setting" "main_alert" {
  name = "Alert Grouping - ${pagerduty_service.main.name}"
  type = "content_based"
  config {
    fields      = var.alert_grouping_fields
    aggregate   = var.alert_grouping_aggregate # Accepts either 'any' or 'all'
    time_window = 86400                        # 24 hours
  }
  services = [pagerduty_service.main.id]
}

---

Expected Behavior

• Terraform should preserve the specified pagerduty_alert_grouping_setting with:
  • type = "content_based"
  • time_window = 86400 (24 hours)
• No unexpected diffs in terraform plan unless the configuration was explicitly changed.

---

Actual Behavior

• Periodically (often after upgrading the PagerDuty provider), the terraform plan shows:
  • pagerduty_alert_grouping_setting being removed from some services.
  • time_window reverting to 300 seconds for others.
• These changes are not reflected in the actual Terraform configuration.
• Manual re-apply restores the correct configuration.

---

Steps to Reproduce

1. Configure any number of services with pagerduty_alert_grouping_setting as shown above.
2. Apply the configuration.
3. Upgrade or reinitialize the PagerDuty provider (e.g., after some time or version bump).
4. Run terraform plan.
5. Observe unexpected diffs in pagerduty_alert_grouping_setting.

---

Likely Cause

• Inconsistencies between Terraform state and the PagerDuty API response regarding alert grouping settings.
• Possibly due to:
  • Internal default values applied by PagerDuty but not surfaced correctly via the API.
  • Changes in provider behavior during upgrades (e.g., handling of time_window defaults).

---

Temporary Workaround

• Manually re-apply the Terraform plan whenever this issue appears.

---

Impact of the Issue

• Loss of expected alert grouping behavior, resulting in:
  • Alert noise.
  • Reduced signal clarity for on-call engineers.
• Repeated manual remediation required to restore desired configurations.

---

Suggested Resolution

• Clarify in provider documentation how pagerduty_alert_grouping_setting is managed and surfaced via the API.
• Fix any inconsistencies in how defaults or deletions are handled during refresh.
• Consider more predictable behavior or flags to maintain time_window persistence.

[cjgajard]

Hi @nmistry24, I was not able to reproduce by following Steps to Reproduce alone.

This probably occurs when theres a conflicting configuration of alert_grouping_parameters in the service used by the alert grouping setting. Can you share the terraform code for pagerduty_service.main?

[nmistry24]

Hi @cjgajard, we have removed the older alert_grouping_parameters around 5-6 months ago and have since migrated to the new pagerduty_alert_grouping_setting. Here is the TF code block that we are using:

resource "pagerduty_service" "main" {
name = local.full_name #eg. Google Analytics (Platform) [Workday]
escalation_policy = var.escalation_policy
auto_resolve_timeout = "null"
acknowledgement_timeout = 3600
alert_creation = "create_alerts_and_incidents"
dynamic "incident_urgency_rule" {
for_each = local.is_low-urgency ? [] : [1]
content {
type = "use_support_hours"
during_support_hours {
type = "constant"
urgency = "high"
}
outside_support_hours {
type = "constant"
urgency = "low"
}
}
}
dynamic "support_hours" {
for_each = local.is_low-urgency ? [] : [1]
content {
type = "fixed_time_per_day"
time_zone = "America/Los_Angeles"
start_time = local.support_tiers[var.support_tier].start
end_time = local.support_tiers[var.support_tier].end
days_of_week = local.support_tiers[var.support_tier].days_of_week
}
}
dynamic "incident_urgency_rule" {
for_each = local.is_low-urgency ? [1] : []
content {
type = "constant"
urgency = "low"
}
}
}

resource "pagerduty_alert_grouping_setting" "main_alert" {
name = "Alert Grouping - ${pagerduty_service.main.name}"
type = var.alert_grouping_type #eg. "content_based"
config {
fields = var.alert_grouping_fields #eg. "summary"
aggregate = var.alert_grouping_aggregate #eg. "all"
time_window = 86400
}
services = [pagerduty_service.main.id]
}