Correct scaling for junos load average alerts

[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 5474e7e38b9d824ac9157ef372d26111c9149163..2678f5e1a26aa4ceb2832bdcde1b837b6ffb0fbf 100644 (file)
--- a/cookbooks/prometheus/templates/default/alert_rules.yml.erb
+++ b/cookbooks/prometheus/templates/default/alert_rules.yml.erb
@@ -3,6 +3,13 @@
 groups:
   - name: amsterdam
     rules:
 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 > 28
         for: 6m
       - alert: pdu current draw
         expr: rPDU2PhaseStatusCurrent{site="amsterdam",rPDU2PhaseStatusIndex="1"} / 10 > 28
         for: 6m


@@ -11,7 +18,7 @@ groups:
         annotations:
           current: "{{ $value | humanize }}A"
       - alert: site power
         annotations:
           current: "{{ $value | humanize }}A"
       - alert: site power

-        expr: sum(avg_over_time(rPDU2PhaseStatusApparentPower{site="amsterdam",rPDU2PhaseStatusIndex="1"}[1h]) / 100) > 3
+        expr: sum(avg_over_time(rPDU2PhaseStatusApparentPower{site="amsterdam",rPDU2PhaseStatusIndex="1"}[1h]) / 100) > 3.5

         for: 6m
         labels:
           alertgroup: "amsterdam"
         for: 6m
         labels:
           alertgroup: "amsterdam"


@@ -45,13 +52,6 @@ groups:
           alertgroup: "{{ $labels.instance }}"
         annotations:
           busy_workers: "{{ $value | humanizePercentage }}"
           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
-        labels:
-          alertgroup: "{{ $labels.instance }}"
-        annotations:
-          request_rate: "{{ $value | humanizePercentage }}"

   - name: chef
     rules:
       - alert: chef client not running
   - name: chef
     rules:
       - alert: chef client not running


@@ -116,6 +116,13 @@ groups:
           failure_rate: "{{ $value }} jobs/s"
   - name: dublin
     rules:
           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 > 28
         for: 6m
       - alert: pdu current draw
         expr: rPDU2PhaseStatusCurrent{site="dublin",rPDU2PhaseStatusIndex="1"} / 10 > 28
         for: 6m


@@ -153,13 +160,23 @@ groups:
           alertgroup: fastly
         annotations:
           error_rate: "{{ $value | humanizePercentage }}"
           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: 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
         for: 15m
         labels:
           alertgroup: fastly

-      - alert: multiple fastly healthchecks failing
-        expr: count(fastly_healthcheck_status == 0) by (service) > 4
+      - 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
         for: 5m
         labels:
           alertgroup: fastly


@@ -256,17 +273,19 @@ groups:
   - name: juniper
     rules:
       - alert: juniper cpu alarm
   - name: juniper
     rules:
       - alert: juniper cpu alarm

-        expr: jnxOperatingCPU{jnxOperatingContentsIndex="7"} > 30
+        expr: junos_route_engine_load_average_five / 2 > 0.5

         for: 5m
         labels:
           alertgroup: "{{ $labels.site }}"
         for: 5m
         labels:
           alertgroup: "{{ $labels.site }}"

+        annotations:
+          load_average: "{{ $value | humanizePercentage }}"

       - alert: juniper fan alarm
       - alert: juniper fan alarm

-        expr: jnxOperatingState{jnxOperatingContentsIndex="4",jnxOperatingState!~"running.*"} > 0
+        expr: junos_environment_fan_up != 1

         for: 5m
         labels:
           alertgroup: "{{ $labels.site }}"
       - alert: juniper power alarm
         for: 5m
         labels:
           alertgroup: "{{ $labels.site }}"
       - alert: juniper power alarm

-        expr: jnxOperatingState{jnxOperatingContentsIndex="2",jnxOperatingState!~"running.*"} > 0
+        expr: junos_environment_power_up != 1

         for: 5m
         labels:
           alertgroup: "{{ $labels.site }}"
         for: 5m
         labels:
           alertgroup: "{{ $labels.site }}"


@@ -278,7 +297,7 @@ groups:
         labels:
           alertgroup: "{{ $labels.instance }}"
       - alert: exim queue length
         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
         for: 60m
         labels:
           alertgroup: mail


@@ -363,14 +382,14 @@ groups:
   - name: network
     rules:
       - alert: interface transmit rate
   - 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
         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 }}"
         for: 5m
         labels:
           alertgroup: "{{ $labels.instance }}"


@@ -525,6 +544,13 @@ groups:
           alertgroup: "{{ $labels.instance }}"
         annotations:
           queries: "{{ $value }}"
           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
   - name: prometheus
     rules:
       - alert: prometheus configuration error


@@ -668,7 +694,7 @@ groups:
           age: "{{ $value | humanizeDuration }}"
       - alert: taginfo database size
         expr: abs(delta(taginfo_database_size_bytes[30m])) / taginfo_database_size_bytes > 0.1
           age: "{{ $value | humanizeDuration }}"
       - alert: taginfo database size
         expr: abs(delta(taginfo_database_size_bytes[30m])) / taginfo_database_size_bytes > 0.1

-        for: 0m
+        for: 30m

         labels:
           alertgroup: taginfo
         annotations:
         labels:
           alertgroup: taginfo
         annotations:


@@ -720,8 +746,8 @@ groups:
         annotations:
           error_rate: "{{ $value | humanizePercentage }}"
       - alert: job processing rate
         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:
         labels:
           alertgroup: web
         annotations: