]> git.openstreetmap.org Git - chef.git/blobdiff - cookbooks/prometheus/templates/default/alert_rules.yml.erb
Set a minimum threshold for error rate alerts
[chef.git] / cookbooks / prometheus / templates / default / alert_rules.yml.erb
index 8c1aeee48bea93bfc9ec168c698b0ac45babdbb5..b78bbaddd51fb8a54cc2834c4815f28a45c85af4 100644 (file)
@@ -3,30 +3,37 @@
 groups:
   - name: amsterdam
     rules:
+      - alert: uplink
+        expr: junos_interface_up{site="amsterdam",name=~"ge-[01]/2/2"} != 1
+        for: 6m
+        labels:
+          alertgroup: "amsterdam"
+        annotations:
+          status: "{{ $value }}"
       - alert: pdu current draw
-        expr: rPDU2PhaseStatusCurrent{site="amsterdam",rPDU2PhaseStatusIndex="1"} / 10 > 10
-        for: 5m
+        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
-        for: 5m
+      - alert: site power
+        expr: sum(avg_over_time(rPDU2PhaseStatusApparentPower{site="amsterdam",rPDU2PhaseStatusIndex="1"}[1h]) / 100) > 3.5
+        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 > 25
-        for: 5m
+        expr: min(rPDU2SensorTempHumidityStatusTempC{site="amsterdam"}) / 10 < 15 or min(rPDU2SensorTempHumidityStatusTempC{site="amsterdam"}) / 10 > 32
+        for: 6m
         labels:
           alertgroup: "amsterdam"
         annotations:
           temperature: "{{ $value | humanize }}C"
       - alert: site humidity
-        expr: max(rPDU2SensorTempHumidityStatusRelativeHumidity{site="amsterdam"}) / 100 < 0.25 or max(rPDU2SensorTempHumidityStatusRelativeHumidity{site="amsterdam"}) / 100 > 0.65
-        for: 5m
+        expr: max(rPDU2SensorTempHumidityStatusRelativeHumidity{site="amsterdam"}) / 100 < 0.08 or max(rPDU2SensorTempHumidityStatusRelativeHumidity{site="amsterdam"}) / 100 > 0.8
+        for: 6m
         labels:
           alertgroup: "amsterdam"
         annotations:
@@ -45,13 +52,13 @@ groups:
           alertgroup: "{{ $labels.instance }}"
         annotations:
           busy_workers: "{{ $value | humanizePercentage }}"
-      - alert: apache low request rate
-        expr: rate(apache_accesses_total[5m]) / rate(apache_accesses_total[1h] offset 1w) < 0.25 and rate(apache_accesses_total[1h] offset 1w) > 2
-        for: 15m
+      - alert: apache connection limit
+        expr: (apache_connections{state="total"} - on (instance) apache_connections{state="closing"}) / on (instance) (apache_server_limit * on (instance) (apache_threads_per_child + on (instance) (apache_async_request_worker_factor * on (instance) apache_workers{state="idle"} / on(instance) apache_processes{state="all"}))) > 0.8
+        for: 5m
         labels:
           alertgroup: "{{ $labels.instance }}"
         annotations:
-          request_rate: "{{ $value | humanizePercentage }}"
+          connections: "{{ $value | humanizePercentage }}"
   - name: chef
     rules:
       - alert: chef client not running
@@ -90,47 +97,70 @@ 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:
           pressure: "{{ $value | humanizePercentage }}"
   - name: database
     rules:
-      - alert: postgres replication delay
-        expr: pg_replication_lag_seconds > 5
+      - alert: active rails queries
+        expr: sum(pg_stat_activity_count{datname="openstreetmap",usename="rails",state="active"}) by (instance) > 50 and on (instance) chef_role{name="db-master"}
         for: 5m
         labels:
           alertgroup: database
         annotations:
-          delay: "{{ $value | humanizeDuration }}"
+          queries: "{{ $value }}"
+      - alert: active cgimap queries
+        expr: sum(pg_stat_activity_count{datname="openstreetmap",usename="cgimap",state="active"}) by (instance) > 30 and on (instance) chef_role{name="db-master"}
+        for: 5m
+        labels:
+          alertgroup: database
+        annotations:
+          queries: "{{ $value }}"
+  - 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: uplink
+        expr: junos_interface_up{site="dublin",name=~"ge-[01]/2/2"} != 1
+        for: 6m
+        labels:
+          alertgroup: "dublin"
+        annotations:
+          status: "{{ $value }}"
       - alert: pdu current draw
-        expr: rPDU2PhaseStatusCurrent{site="dublin",rPDU2PhaseStatusIndex="1"} / 10 > 13
-        for: 5m
+        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
-        for: 5m
+      - 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 > 25
-        for: 5m
+        expr: min(rPDU2SensorTempHumidityStatusTempC{site="dublin"}) / 10 < 18 or min(rPDU2SensorTempHumidityStatusTempC{site="dublin"}) / 10 > 26
+        for: 6m
         labels:
           alertgroup: "dublin"
         annotations:
           temperature: "{{ $value | humanize }}C"
       - alert: site humidity
         expr: max(rPDU2SensorTempHumidityStatusRelativeHumidity{site="dublin"}) / 100 < 0.25 or max(rPDU2SensorTempHumidityStatusRelativeHumidity{site="dublin"}) / 100 > 0.65
-        for: 5m
+        for: 6m
         labels:
           alertgroup: "dublin"
         annotations:
@@ -144,20 +174,30 @@ groups:
           alertgroup: fastly
         annotations:
           error_rate: "{{ $value | humanizePercentage }}"
-      - alert: fastly healthcheck failing
-        expr: count(fastly_healthcheck_status == 0) by (service) > 0
+      - alert: fastly frontend healthcheck warning
+        expr: count(fastly_healthcheck_status == 0) by (service, datacenter) > 2
         for: 15m
         labels:
           alertgroup: fastly
-      - alert: multiple fastly healthchecks failing
-        expr: count(fastly_healthcheck_status == 0) by (service) > 4
+      - alert: fastly frontend healthcheck critical
+        expr: count(fastly_healthcheck_status == 0) by (service, datacenter) == count(fastly_healthcheck_status) by (service, datacenter)
+        for: 5m
+        labels:
+          alertgroup: fastly
+      - alert: fastly backend healthcheck warning
+        expr: count(fastly_healthcheck_status == 0) by (service, backend) > 10
+        for: 15m
+        labels:
+          alertgroup: fastly
+      - alert: fastly backend healthcheck critical
+        expr: count(fastly_healthcheck_status == 0) by (service, backend) == count(fastly_healthcheck_status) by (service, backend)
         for: 5m
         labels:
           alertgroup: fastly
   - name: filesystem
     rules:
       - alert: readonly filesystem
-        expr: node_filesystem_readonly == 1
+        expr: node_filesystem_readonly > min_over_time(node_filesystem_readonly[7d])
         for: 0m
         labels:
           alertgroup: "{{ $labels.instance }}"
@@ -246,20 +286,60 @@ groups:
           alertgroup: "{{ $labels.instance }}"
   - name: juniper
     rules:
+      - alert: juniper red alarms
+        expr: juniper_alarms_red_count > 0
+        for: 5m
+        labels:
+          alertgroup: "{{ $labels.site }}"
+        annotations:
+          alarm_count: "{{ $value }} alarms"
+      - alert: juniper yellow alarms
+        expr: juniper_alarms_yellow_count > 0
+        for: 5m
+        labels:
+          alertgroup: "{{ $labels.site }}"
+        annotations:
+          alarm_count: "{{ $value }} alarms"
+      - alert: juniper cpu alarm
+        expr: junos_route_engine_load_average_five / 2 > 0.5
+        for: 5m
+        labels:
+          alertgroup: "{{ $labels.site }}"
+        annotations:
+          load_average: "{{ $value | humanizePercentage }}"
       - alert: juniper fan alarm
-        expr: sum_over_time(jnxOperatingState{jnxOperatingContentsIndex="4",jnxOperatingState!~"running.*"}[6m]) > 0
+        expr: junos_environment_fan_up != 1
         for: 5m
         labels:
           alertgroup: "{{ $labels.site }}"
       - alert: juniper power alarm
-        expr: sum_over_time(jnxOperatingState{jnxOperatingContentsIndex="2",jnxOperatingState!~"running.*"}[6m]) > 0
+        expr: junos_environment_power_up != 1
         for: 5m
         labels:
           alertgroup: "{{ $labels.site }}"
+      - alert: juniper laser receive power
+        expr: junos_interface_diagnostics_laser_rx_dbm < -12 and on (site, instance, name) junos_interface_admin_up == 1
+        for: 5m
+        labels:
+          alertgroup: "{{ $labels.site }}"
+        annotations:
+          power: "{{ $value }} dBm"
+      - alert: juniper laser transmit power
+        expr: junos_interface_diagnostics_laser_output_dbm < -8 and on (site, instance, name) junos_interface_admin_up == 1
+        for: 5m
+        labels:
+          alertgroup: "{{ $labels.site }}"
+        annotations:
+          power: "{{ $value }} dBm"
   - name: mail
     rules:
+      - alert: exim down
+        expr: exim_up == 0
+        for: 5m
+        labels:
+          alertgroup: "{{ $labels.instance }}"
       - alert: exim queue length
-        expr: exim_queue > exim_queue_limit
+        expr: exim_queue > ignoring(job) exim_queue_limit
         for: 60m
         labels:
           alertgroup: mail
@@ -327,17 +407,45 @@ groups:
           alertgroup: "{{ $labels.instance }}"
         annotations:
           new_oom_kills: "{{ $value }}"
+  - name: mysql
+    rules:
+      - alert: mysql down
+        expr: mysql_up == 0
+        for: 1m
+        labels:
+          alertgroup: "{{ $labels.instance }}"
+      - alert: mysql connection limit
+        expr: mysql_global_status_max_used_connections / mysql_global_variables_max_connections > 0.8
+        for: 1m
+        labels:
+          alertgroup: "{{ $labels.instance }}"
+        annotations:
+          connections_used: "{{ $value | humanizePercentage }}"
+      - alert: mysql connection errors
+        expr: increase(mysql_global_status_connection_errors_total[1m]) > 0
+        for: 0m
+        labels:
+          alertgroup: "{{ $labels.instance }}"
+        annotations:
+          error_count: "{{ $value }}"
   - name: network
     rules:
+      - alert: interface redundancy lost
+        expr: node_bonding_active < 2 and on (instance, master) label_replace(chef_network_interface{bond_mode="802.3ad"}, "master", "$1", "name", "(.*)")
+        for: 5m
+        labels:
+          alertgroup: "{{ $labels.instance }}"
+        annotations:
+          link_count: "{{ $value }}"
       - 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 }}"
@@ -380,6 +488,11 @@ groups:
           alertgroup: nominatim
         annotations:
           delay: "{{ $value | humanizeDuration }}"
+      - alert: nominatim connections
+        expr: sum(nginx_connections_writing and on (instance) chef_role{name="nominatim"}) > 2500
+        for: 15m
+        labels:
+          alertgroup: nominatim
   - name: overpass
     rules:
       - alert: overpass osm database age
@@ -396,6 +509,23 @@ groups:
           alertgroup: overpass
         annotations:
           age: "{{ $value | humanizeDuration }}"
+  - name: passenger
+    rules:
+      - alert: passenger down
+        expr: passenger_up == 0
+        for: 5m
+        labels:
+          alertgroup: "{{ $labels.instance }}"
+      - alert: passenger queuing
+        expr: passenger_top_level_request_queue > 0
+        for: 5m
+        labels:
+          alertgroup: "{{ $labels.instance }}"
+      - alert: passenger application queuing
+        expr: passenger_app_request_queue > 0
+        for: 5m
+        labels:
+          alertgroup: "{{ $labels.instance }}"
   - name: planet
     rules:
       - alert: planet dump overdue
@@ -448,8 +578,8 @@ 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:
@@ -468,8 +598,8 @@ groups:
           alertgroup: "{{ $labels.instance }}"
         annotations:
           new_deadlocks: "{{ $value }}"
-      - alert: postgresql slow queries
-        expr: pg_slow_queries > 0
+      - 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 }}"
@@ -487,8 +617,23 @@ groups:
         for: 10m
         labels:
           alertgroup: "prometheus"
+      - alert: node exporter text file scrape error
+        expr: node_textfile_scrape_error > 0
+        for: 10m
+        labels:
+          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
@@ -514,7 +659,7 @@ groups:
         labels:
           alertgroup: "{{ $labels.instance }}"
         annotations:
-          new_ercrors: "{{ $value }}"
+          new_errors: "{{ $value }}"
   - name: smart
     rules:
       - alert: smart failure
@@ -523,12 +668,30 @@ 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
+        expr: max_over_time(snmp_scrape_pdus_returned[1d]) - snmp_scrape_pdus_returned > 0
+        for: 15m
+        labels:
+          alertgroup: snmp
+        annotations:
+          missing_pdus: "{{ $value }}"
   - name: ssl
     rules:
       - alert: ssl certificate probe failed
@@ -556,7 +719,7 @@ 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
@@ -568,10 +731,33 @@ 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 and on (instance) chef_role{name="taginfo"}
+        for: 0m
+        labels:
+          alertgroup: taginfo
+        annotations:
+          age: "{{ $value | humanizeDuration }}"
+      - alert: taginfo database age
+        expr: time() - taginfo_database_update_finish_seconds > 129600 and on (instance) chef_role{name="taginfo"}
+        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
@@ -588,6 +774,13 @@ groups:
           alertgroup: tile
         annotations:
           miss_rate: "{{ $value | humanizePercentage }}"
+      - alert: tile render rate
+        expr: sum(rate(renderd_zoom_metatiles_total[5m])) by (instance) == 0
+        for: 15m
+        labels:
+          alertgroup: tile
+        annotations:
+          render_rate: "{{ $value }} tiles/s"
   - name: time
     rules:
       - alert: clock not synchronising
@@ -605,15 +798,15 @@ groups:
   - name: web
     rules:
       - alert: web error rate
-        expr: sum(rate(api_call_count_total{status=~"50[0-8]|5[1-9][0-9]"}[5m])) by (instance) / sum(rate(api_call_count_total[5m])) by (instance) > 0.002
+        expr: sum(rate(api_call_count_total{status=~"50[0-8]|5[1-9][0-9]"}[5m])) by (instance) / sum(rate(api_call_count_total[5m])) by (instance) > 0.002 and sum(rate(api_call_count_total{status=~"50[0-8]|5[1-9][0-9]"}[5m])) by (instance) > 0.01
         for: 5m
         labels:
           alertgroup: web
         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: