Reduce sensitivity of job processing rate alert

[chef.git] / cookbooks / prometheus / templates / default / alert_rules.yml.erb

diff --git a/cookbooks/prometheus/templates/default/alert_rules.yml.erb b/cookbooks/prometheus/templates/default/alert_rules.yml.erb
index 3618f1461ff485b666b8eb26c510cca2f95760ce..02c41ce52555b862ed94c33b0e377ecfd3a3497c 100644 (file)
--- a/cookbooks/prometheus/templates/default/alert_rules.yml.erb
+++ b/cookbooks/prometheus/templates/default/alert_rules.yml.erb
@@ -4,19 +4,19 @@ groups:
   - name: amsterdam
     rules:
       - alert: pdu current draw
-        expr: rPDU2PhaseStatusCurrent{site="amsterdam",rPDU2PhaseStatusIndex="1"} / 10 > 10
+        expr: rPDU2PhaseStatusCurrent{site="amsterdam",rPDU2PhaseStatusIndex="1"} / 10 > 28
         for: 6m
         labels:
           alertgroup: "amsterdam"
         annotations:
           current: "{{ $value | humanize }}A"
-      - alert: site current draw
-        expr: sum(rPDU2PhaseStatusCurrent{site="amsterdam",rPDU2PhaseStatusIndex="1"} / 10) > 13
+      - alert: site power
+        expr: sum(avg_over_time(rPDU2PhaseStatusApparentPower{site="amsterdam",rPDU2PhaseStatusIndex="1"}[1h]) / 100) > 3
         for: 6m
         labels:
           alertgroup: "amsterdam"
         annotations:
-          current: "{{ $value | humanize }}A"
+          current: "{{ $value | humanize }}kVA"
       - alert: site temperature
         expr: min(rPDU2SensorTempHumidityStatusTempC{site="amsterdam"}) / 10 < 18 or min(rPDU2SensorTempHumidityStatusTempC{site="amsterdam"}) / 10 > 26
         for: 6m
@@ -90,8 +90,8 @@ groups:
   - name: cpu
     rules:
       - alert: cpu pressure
-        expr: rate(node_pressure_cpu_waiting_seconds_total[5m]) > 0.6
-        for: 15m
+        expr: rate(node_pressure_cpu_waiting_seconds_total[5m]) > 0.75
+        for: 60m
         labels:
           alertgroup: "{{ $labels.instance }}"
         annotations:
@@ -99,28 +99,37 @@ groups:
   - name: database
     rules:
       - alert: postgres replication delay
-        expr: pg_replication_lag_seconds > 5
-        for: 5m
+        expr: pg_replication_lag_seconds > 30
+        for: 15m
         labels:
           alertgroup: database
         annotations:
           delay: "{{ $value | humanizeDuration }}"
+  - name: discourse
+    rules:
+      - alert: discourse job failure rate
+        expr: rate(discourse_job_failures[5m]) > 0
+        for: 5m
+        labels:
+          alertgroup: discourse
+        annotations:
+          failure_rate: "{{ $value }} jobs/s"
   - name: dublin
     rules:
       - alert: pdu current draw
-        expr: rPDU2PhaseStatusCurrent{site="dublin",rPDU2PhaseStatusIndex="1"} / 10 > 13
+        expr: rPDU2PhaseStatusCurrent{site="dublin",rPDU2PhaseStatusIndex="1"} / 10 > 28
         for: 6m
         labels:
           alertgroup: "dublin"
         annotations:
           current: "{{ $value | humanize }}A"
-      - alert: site current draw
-        expr: sum(rPDU2PhaseStatusCurrent{site="dublin",rPDU2PhaseStatusIndex="1"} / 10) > 17
+      - alert: site power
+        expr: sum(avg_over_time(rPDU2PhaseStatusApparentPower{site="dublin",rPDU2PhaseStatusIndex="1"}[1h]) / 100) > 4
         for: 6m
         labels:
           alertgroup: "dublin"
         annotations:
-          current: "{{ $value | humanize }}A"
+          current: "{{ $value | humanize }}kVA"
       - alert: site temperature
         expr: min(rPDU2SensorTempHumidityStatusTempC{site="dublin"}) / 10 < 18 or min(rPDU2SensorTempHumidityStatusTempC{site="dublin"}) / 10 > 26
         for: 6m
@@ -354,14 +363,14 @@ groups:
   - name: network
     rules:
       - alert: interface transmit rate
-        expr: rate(node_network_transmit_bytes_total[1m]) / node_network_speed_bytes > 0.98
+        expr: rate(node_network_transmit_bytes_total[1m]) / node_network_speed_bytes > 0.99
         for: 5m
         labels:
           alertgroup: "{{ $labels.instance }}"
         annotations:
           bandwidth_used: "{{ $value | humanizePercentage }}"
       - alert: interface receive rate
-        expr: rate(node_network_receive_bytes_total[1m]) / node_network_speed_bytes > 0.98
+        expr: rate(node_network_receive_bytes_total[1m]) / node_network_speed_bytes > 0.99
         for: 5m
         labels:
           alertgroup: "{{ $labels.instance }}"
@@ -489,8 +498,8 @@ groups:
         labels:
           alertgroup: "{{ $labels.instance }}"
       - alert: postgresql replication delay
-        expr: pg_replication_lag_seconds > 5
-        for: 1m
+        expr: pg_replication_lag_seconds > 30
+        for: 15m
         labels:
           alertgroup: "{{ $labels.instance }}"
         annotations:
@@ -516,6 +525,13 @@ groups:
           alertgroup: "{{ $labels.instance }}"
         annotations:
           queries: "{{ $value }}"
+      - alert: postgresql idle transactions
+        expr: sum(pg_process_idle_seconds_count{state="idle in transaction"}) by (instance, server) > sum(pg_process_idle_seconds_bucket{state="idle in transaction",le="300"}) by (instance, server)
+        for: 5m
+        labels:
+          alertgroup: "{{ $labels.instance }}"
+        annotations:
+          queries: "{{ $value }}"
   - name: prometheus
     rules:
       - alert: prometheus configuration error
@@ -530,6 +546,16 @@ groups:
           alertgroup: "prometheus"
   - name: raid
     rules:
+      - alert: raid controller battery failed
+        expr: ohai_controller_info{battery_status="failed"} > 0
+        for: 5m
+        labels:
+          alertgroup: "{{ $labels.instance }}"
+      - alert: raid controller battery recharging
+        expr: ohai_controller_info{battery_status="recharging"} > 0
+        for: 4h
+        labels:
+          alertgroup: "{{ $labels.instance }}"
       - alert: raid array degraded
         expr: ohai_array_info{status="degraded"} > 0
         for: 5m
@@ -564,12 +590,21 @@ groups:
         labels:
           alertgroup: "{{ $labels.instance }}"
       - alert: smart ssd wearout approaching
-        expr: smart_percentage_used >= 80
+        expr: smart_percentage_used / 100 >= 0.8
         for: 60m
         labels:
           alertgroup: "{{ $labels.instance }}"
         annotations:
           percentage_used: "{{ $value | humanizePercentage }}"
+  - name: smokeping
+    rules:
+      - alert: packet loss
+        expr: 1 - (rate(smokeping_response_duration_seconds_count[5m]) / rate(smokeping_requests_total[5m])) > 0.02
+        for: 10m
+        labels:
+          alertgroup: smokeping
+        annotations:
+          loss_rate: "{{ $value | humanizePercentage }}"
   - name: snmp
     rules:
       - alert: snmp pdus missing
@@ -606,7 +641,7 @@ groups:
   - name: statuscake
     rules:
       - alert: statuscake uptime check failing
-        expr: statuscake_uptime{status="down",paused="false"} > 0
+        expr: statuscake_paused == 0 and statuscake_up == 0
         for: 10m
         labels:
           alertgroup: statuscake
@@ -618,10 +653,33 @@ groups:
         labels:
           alertgroup: "{{ $labels.instance }}"
       - alert: systemd failed chef client service
-        expr: node_systemd_unit_state{state="failed",name="chef-client.service"} == 1
-        for: 6h
+        expr: sum_over_time(node_systemd_unit_state{state="inactive",name="chef-client.service"}[6h]) == 0
+        for: 0m
         labels:
           alertgroup: "{{ $labels.instance }}"
+  - name: taginfo
+    rules:
+      - alert: taginfo planet age
+        expr: time() - taginfo_data_from_seconds > 129600
+        for: 0m
+        labels:
+          alertgroup: taginfo
+        annotations:
+          age: "{{ $value | humanizeDuration }}"
+      - alert: taginfo database age
+        expr: time() - taginfo_database_update_finish_seconds > 129600
+        for: 0m
+        labels:
+          alertgroup: taginfo
+        annotations:
+          age: "{{ $value | humanizeDuration }}"
+      - alert: taginfo database size
+        expr: abs(delta(taginfo_database_size_bytes[30m])) / taginfo_database_size_bytes > 0.1
+        for: 30m
+        labels:
+          alertgroup: taginfo
+        annotations:
+          size_change: "{{ $value | humanizePercentage }}"
   - name: tile
     rules:
       - alert: renderd replication delay
@@ -639,8 +697,8 @@ groups:
         annotations:
           miss_rate: "{{ $value | humanizePercentage }}"
       - alert: tile render rate
-        expr: sum(rate(renderd_zoom_metatiles_total[5m])) by (instance) < 1
-        for: 5m
+        expr: sum(rate(renderd_zoom_metatiles_total[5m])) by (instance) == 0
+        for: 15m
         labels:
           alertgroup: tile
         annotations:
@@ -669,8 +727,8 @@ groups:
         annotations:
           error_rate: "{{ $value | humanizePercentage }}"
       - alert: job processing rate
-        expr: rate(pg_stat_user_tables_n_tup_del{datname="openstreetmap",relname="delayed_jobs"}[5m]) / rate(pg_stat_user_tables_n_tup_ins{datname="openstreetmap",relname="delayed_jobs"}[5m]) < 0.9 and ignoring(job, name, datname, relname, schemaname, server) chef_role{name="db-master"} == 1
-        for: 15m
+        expr: rate(pg_stat_user_tables_n_tup_del{datname="openstreetmap",relname="delayed_jobs"}[1h]) / rate(pg_stat_user_tables_n_tup_ins{datname="openstreetmap",relname="delayed_jobs"}[1h]) < 0.9 and ignoring(job, name, datname, relname, schemaname, server) chef_role{name="db-master"} == 1
+        for: 1h
         labels:
           alertgroup: web
         annotations: