Skip to content
Merged
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
6 changes: 5 additions & 1 deletion terraform/alerting-catalog-paging.tf
Original file line number Diff line number Diff line change
Expand Up @@ -1211,6 +1211,10 @@ resource "grafana_rule_group" "catalog_argocd" {
# `sum by (name)` keeps one series per ArgoCD app so the alert fires as a
# separate instance per degraded app (label `name`), surfaced in the page
# via {{ $labels.name }} and the per-app notification route in alerting.tf.
# thundermail-ticket-spike-monitor is excluded: it is a non-critical CronJob
# (Zendesk count poller) whose transient run failures hold the ArgoCD app
# Degraded for up to ~45m and page platform on-call. De-paged per
# thunderbird/platform-infrastructure#611 (durable fix = monitor resilience).
data {
ref_id = "A"
datasource_uid = var.prometheus_datasource_uid
Expand All @@ -1221,7 +1225,7 @@ resource "grafana_rule_group" "catalog_argocd" {
model = jsonencode({
refId = "A"
datasource = { type = "prometheus", uid = var.prometheus_datasource_uid }
expr = "sum by (name) (argocd_app_info{cluster=\"mzla-eks-shared01\",health_status=\"Degraded\"})"
expr = "sum by (name) (argocd_app_info{cluster=\"mzla-eks-shared01\",health_status=\"Degraded\",name!=\"thundermail-ticket-spike-monitor\"})"
instant = true
intervalMs = 1000
maxDataPoints = 43200
Expand Down
Loading