aurore/ansible
aurore/ansible
11
2
Fork 0
Code Issues 1 Pull requests 13 Releases Wiki Activity

Prometheus: cleanup #90

Merged
jeltz merged 15 commits from prometheus_cleanup into master 2022-01-01 17:58:19 +01:00
Conversation 0 Commits 15 Files changed 8 +19 -176
3 changed files with 19 additions and 176 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files
Showing only changes of commit 5ab3dcdac2 - Show all commits

106
roles/prometheus/templates/django.rules.yml.j2
View file

@ -1,106 +0,0 @@
# {{ ansible_managed }}
{# As this is also Jinja2 it will conflict without a raw block #}
{% raw %}
groups:
- name: django.rules
rules:
- record: job:django_http_requests_before_middlewares_total:sum_rate30s
expr: sum(rate(django_http_requests_before_middlewares_total[30s])) BY (job)
- record: job:django_http_requests_unknown_latency_total:sum_rate30s
expr: sum(rate(django_http_requests_unknown_latency_total[30s])) BY (job)
- record: job:django_http_ajax_requests_total:sum_rate30s
expr: sum(rate(django_http_ajax_requests_total[30s])) BY (job)
- record: job:django_http_responses_before_middlewares_total:sum_rate30s
expr: sum(rate(django_http_responses_before_middlewares_total[30s])) BY (job)
- record: job:django_http_requests_unknown_latency_including_middlewares_total:sum_rate30s
expr: sum(rate(django_http_requests_unknown_latency_including_middlewares_total[30s]))
BY (job)
- record: job:django_http_requests_body_total_bytes:sum_rate30s
expr: sum(rate(django_http_requests_body_total_bytes[30s])) BY (job)
- record: job:django_http_responses_streaming_total:sum_rate30s
expr: sum(rate(django_http_responses_streaming_total[30s])) BY (job)
- record: job:django_http_responses_body_total_bytes:sum_rate30s
expr: sum(rate(django_http_responses_body_total_bytes[30s])) BY (job)
- record: job:django_http_requests_total:sum_rate30s
expr: sum(rate(django_http_requests_total_by_method[30s])) BY (job)
- record: job:django_http_requests_total_by_method:sum_rate30s
expr: sum(rate(django_http_requests_total_by_method[30s])) BY (job, method)
- record: job:django_http_requests_total_by_transport:sum_rate30s
expr: sum(rate(django_http_requests_total_by_transport[30s])) BY (job, transport)
- record: job:django_http_requests_total_by_view:sum_rate30s
expr: sum(rate(django_http_requests_total_by_view_transport_method[30s])) BY (job,
view)
- record: job:django_http_requests_total_by_view_transport_method:sum_rate30s
expr: sum(rate(django_http_requests_total_by_view_transport_method[30s])) BY (job,
view, transport, method)
- record: job:django_http_responses_total_by_templatename:sum_rate30s
expr: sum(rate(django_http_responses_total_by_templatename[30s])) BY (job, templatename)
- record: job:django_http_responses_total_by_status:sum_rate30s
expr: sum(rate(django_http_responses_total_by_status[30s])) BY (job, status)
- record: job:django_http_responses_total_by_charset:sum_rate30s
expr: sum(rate(django_http_responses_total_by_charset[30s])) BY (job, charset)
- record: job:django_http_exceptions_total_by_type:sum_rate30s
expr: sum(rate(django_http_exceptions_total_by_type[30s])) BY (job, type)
- record: job:django_http_exceptions_total_by_view:sum_rate30s
expr: sum(rate(django_http_exceptions_total_by_view[30s])) BY (job, view)
- record: job:django_http_requests_latency_including_middlewares_seconds:quantile_rate30s
expr: histogram_quantile(0.5, sum(rate(django_http_requests_latency_including_middlewares_seconds_bucket[30s]))
BY (job, le))
labels:
quantile: "50"
- record: job:django_http_requests_latency_including_middlewares_seconds:quantile_rate30s
expr: histogram_quantile(0.95, sum(rate(django_http_requests_latency_including_middlewares_seconds_bucket[30s]))
BY (job, le))
labels:
quantile: "95"
- record: job:django_http_requests_latency_including_middlewares_seconds:quantile_rate30s
expr: histogram_quantile(0.99, sum(rate(django_http_requests_latency_including_middlewares_seconds_bucket[30s]))
BY (job, le))
labels:
quantile: "99"
- record: job:django_http_requests_latency_including_middlewares_seconds:quantile_rate30s
expr: histogram_quantile(0.999, sum(rate(django_http_requests_latency_including_middlewares_seconds_bucket[30s]))
BY (job, le))
labels:
quantile: "99.9"
- record: job:django_http_requests_latency_seconds:quantile_rate30s
expr: histogram_quantile(0.5, sum(rate(django_http_requests_latency_seconds_bucket[30s]))
BY (job, le))
labels:
quantile: "50"
- record: job:django_http_requests_latency_seconds:quantile_rate30s
expr: histogram_quantile(0.95, sum(rate(django_http_requests_latency_seconds_bucket[30s]))
BY (job, le))
labels:
quantile: "95"
- record: job:django_http_requests_latency_seconds:quantile_rate30s
expr: histogram_quantile(0.99, sum(rate(django_http_requests_latency_seconds_bucket[30s]))
BY (job, le))
labels:
quantile: "99"
- record: job:django_http_requests_latency_seconds:quantile_rate30s
expr: histogram_quantile(0.999, sum(rate(django_http_requests_latency_seconds_bucket[30s]))
BY (job, le))
labels:
quantile: "99.9"
- record: job:django_model_inserts_total:sum_rate1m
expr: sum(rate(django_model_inserts_total[1m])) BY (job, model)
- record: job:django_model_updates_total:sum_rate1m
expr: sum(rate(django_model_updates_total[1m])) BY (job, model)
- record: job:django_model_deletes_total:sum_rate1m
expr: sum(rate(django_model_deletes_total[1m])) BY (job, model)
- record: job:django_db_new_connections_total:sum_rate30s
expr: sum(rate(django_db_new_connections_total[30s])) BY (alias, vendor)
- record: job:django_db_new_connection_errors_total:sum_rate30s
expr: sum(rate(django_db_new_connection_errors_total[30s])) BY (alias, vendor)
- record: job:django_db_execute_total:sum_rate30s
expr: sum(rate(django_db_execute_total[30s])) BY (alias, vendor)
- record: job:django_db_execute_many_total:sum_rate30s
expr: sum(rate(django_db_execute_many_total[30s])) BY (alias, vendor)
- record: job:django_db_errors_total:sum_rate30s
expr: sum(rate(django_db_errors_total[30s])) BY (alias, vendor, type)
- record: job:django_migrations_applied_total:max
expr: max(django_migrations_applied_total) BY (job, connection)
- record: job:django_migrations_unapplied_total:max
expr: max(django_migrations_unapplied_total) BY (job, connection)
{% endraw %}

17
roles/prometheus/templates/ilo.rules.yml.j2
View file

@ -11,48 +11,53 @@ groups:
rules:
- alert: IloBiosSelfTestDegraded
expr: cpqHeHWBiosCondition >= 3
expr: cpqHeHWBiosCondition{cpqHeHWBiosCondition!~"ok|other"} == 1
for: 3m
labels:
severity: critical
annotations:
summary: >-
Une erreur a été détectée lors du POST du serveur
({{ raw('$labels.cpqHeHWBiosCondition') }})
- alert: IloTemperatureSensorDegraded
expr: cpqHeTemperatureCondition >= 3
expr: cpqHeTemperatureCondition{cpqHeTemperatureCondition!~"ok|other"} == 1
jeltz marked this conversation as resolved
jeltz commented 2022-01-01 17:22:48 +01:00
Review
Copy link

À quoi correspond other ? C'est jamais une erreur ?

À quoi correspond `other` ? C'est jamais une erreur ?
erdnaxe commented 2022-01-01 17:55:13 +01:00
Review
Copy link

other = le module est absent par exemple, on a BEAUCOUP de other

other = le module est absent par exemple, on a BEAUCOUP de other
jeltz commented 2022-01-01 17:56:42 +01:00
Review
Copy link

Ok

Ok
for: 3m
labels:
severity: critical
annotations:
summary: >-
Le capteur de température est dégradé
({{ raw('$labels.cpqHeTemperatureCondition') }})
- alert: IloFanDegraded
expr: cpqHeFltTolFanCondition >= 3
expr: cpqHeFltTolFanCondition{cpqHeFltTolFanCondition!~"ok|other"} == 1
for: 3m
labels:
severity: critical
annotations:
summary: >-
Le ventilateur est dégradé
({{ raw('$labels.cpqHeFltTolFanCondition') }})
- alert: IloPowerSupplyDegraded
expr: cpqHeFltTolPowerSupplyStatus >= 2
expr: cpqHeFltTolPowerSupplyStatus{cpqHeFltTolPowerSupplyStatus!="noError"} == 1
for: 3m
labels:
severity: critical
annotations:
summary: >-
L'alimentation est dégradée
({{ raw('$labels.cpqHeFltTolPowerSupplyStatus') }})
- alert: IloOverrideSwitchState
expr: cpqSm2CntlriLOSecurityOverrideSwitchState == 2
expr: cpqSm2CntlriLOSecurityOverrideSwitchState{cpqSm2CntlriLOSecurityOverrideSwitchState="set"} == 1
for: 3m
labels:
severity: critical
annotations:
summary: >-
Le switch de réinitialisation n'est pas à l'état d'origine
Le switch de réinitialisation n'est pas à l'état d'origine,
l'authentification est bypassée
...

72
roles/prometheus/templates/snmp.yml.j2
View file

@ -497,7 +497,7 @@ ilo:
help: This is the current Power Meter reading in Watts - 1.3.6.1.4.1.232.6.2.15.3
- name: cpqHeHWBiosCondition
oid: 1.3.6.1.4.1.232.6.2.16.1
type: gauge
type: EnumAsStateSet
help: This value indicates an error has been detected during Pre-OS Test (POST)
or during initial hardware initialization - 1.3.6.1.4.1.232.6.2.16.1
enum_values:
@ -505,25 +505,6 @@ ilo:
2: ok
3: degraded
4: failed
- name: cpqHeTemperatureChassis
oid: 1.3.6.1.4.1.232.6.2.6.8.1.1
type: gauge
help: The System Chassis number. - 1.3.6.1.4.1.232.6.2.6.8.1.1
indexes:
- labelname: cpqHeTemperatureChassis
type: gauge
- labelname: cpqHeTemperatureIndex
type: gauge
- name: cpqHeTemperatureIndex
oid: 1.3.6.1.4.1.232.6.2.6.8.1.2
type: gauge
help: A number that uniquely specifies this temperature sensor description. -
1.3.6.1.4.1.232.6.2.6.8.1.2
indexes:
- labelname: cpqHeTemperatureChassis
type: gauge
- labelname: cpqHeTemperatureIndex
type: gauge
- name: cpqHeTemperatureLocale
oid: 1.3.6.1.4.1.232.6.2.6.8.1.3
type: EnumAsInfo
@ -569,7 +550,7 @@ ilo:
type: gauge
- name: cpqHeTemperatureCondition
oid: 1.3.6.1.4.1.232.6.2.6.8.1.6
type: gauge
type: EnumAsStateSet
help: The Temperature sensor condition - 1.3.6.1.4.1.232.6.2.6.8.1.6
indexes:
- labelname: cpqHeTemperatureChassis
@ -581,25 +562,6 @@ ilo:
2: ok
3: degraded
4: failed
- name: cpqHeFltTolFanChassis
oid: 1.3.6.1.4.1.232.6.2.6.7.1.1
type: gauge
help: The System Chassis number. - 1.3.6.1.4.1.232.6.2.6.7.1.1
indexes:
- labelname: cpqHeFltTolFanChassis
type: gauge
- labelname: cpqHeFltTolFanIndex
type: gauge
- name: cpqHeFltTolFanIndex
oid: 1.3.6.1.4.1.232.6.2.6.7.1.2
type: gauge
help: A number that uniquely specifies this fan description. -
1.3.6.1.4.1.232.6.2.6.7.1.2
indexes:
- labelname: cpqHeFltTolFanChassis
type: gauge
- labelname: cpqHeFltTolFanIndex
type: gauge
- name: cpqHeFltTolFanLocale
oid: 1.3.6.1.4.1.232.6.2.6.7.1.3
type: EnumAsInfo
@ -626,7 +588,7 @@ ilo:
13: bridgeCard
- name: cpqHeFltTolFanCondition
oid: 1.3.6.1.4.1.232.6.2.6.7.1.9
type: gauge
type: EnumAsStateSet
help: The fan condition - 1.3.6.1.4.1.232.6.2.6.7.1.9
indexes:
- labelname: cpqHeFltTolFanChassis
@ -638,27 +600,9 @@ ilo:
2: ok
3: degraded
4: failed
- name: cpqHeFltTolPowerSupplyChassis
oid: 1.3.6.1.4.1.232.6.2.9.3.1.1
type: gauge
help: The system chassis number. - 1.3.6.1.4.1.232.6.2.9.3.1.1
indexes:
- labelname: cpqHeFltTolPowerSupplyChassis
type: gauge
- labelname: cpqHeFltTolPowerSupplyBay
type: gauge
- name: cpqHeFltTolPowerSupplyBay
oid: 1.3.6.1.4.1.232.6.2.9.3.1.2
type: gauge
help: The bay number to index within this chassis. - 1.3.6.1.4.1.232.6.2.9.3.1.2
indexes:
- labelname: cpqHeFltTolPowerSupplyChassis
type: gauge
- labelname: cpqHeFltTolPowerSupplyBay
type: gauge
- name: cpqHeFltTolPowerSupplyStatus
oid: 1.3.6.1.4.1.232.6.2.9.3.1.5
type: gauge
type: EnumAsStateSet
help: The status of the power supply. - 1.3.6.1.4.1.232.6.2.9.3.1.5
indexes:
- labelname: cpqHeFltTolPowerSupplyChassis
@ -685,7 +629,7 @@ ilo:
17: noPowerInput
- name: cpqSm2CntlrInterfaceStatus
oid: 1.3.6.1.4.1.232.9.2.2.17
type: gauge
type: EnumAsStateSet
help: Remote Insight/ Integrated Lights-Out Interface Status - 1.3.6.1.4.1.232.9.2.2.17
enum_values:
1: other
@ -693,7 +637,7 @@ ilo:
3: notResponding
- name: cpqSm2CntlriLOSecurityOverrideSwitchState
oid: 1.3.6.1.4.1.232.9.2.2.27
type: gauge
type: EnumAsStateSet
help: Integrated Lights-Out Security Override Switch State - 1.3.6.1.4.1.232.9.2.2.27
enum_values:
1: notSupported
@ -701,7 +645,7 @@ ilo:
3: notSet
- name: cpqSm2CntlrLicenseActive
oid: 1.3.6.1.4.1.232.9.2.2.30
type: gauge
type: EnumAsStateSet
help: Remote Insight License State - 1.3.6.1.4.1.232.9.2.2.30
enum_values:
1: none
@ -714,7 +658,7 @@ ilo:
8: iloAdvancedPremiumSecurity
- name: cpqSm2CntlrServerPowerState
oid: 1.3.6.1.4.1.232.9.2.2.32
type: gauge
type: EnumAsStateSet
help: The current power state for the server - 1.3.6.1.4.1.232.9.2.2.32
enum_values:
1: unknown