Since these services are replicated (two per service),
we should check for the max, not min, as one of the replicas
could be down, but still the API is Up.
Change-Id: I6dad9b74d65415f5f4f91e00c732792010487619
Now it is able to install proemtheus on host and
use it to scrape data from other prometheuses.
Prod-related: PROD-17350
Change-Id: I11d65920c0e9b685dead22a5551521fa15d13f25
New version changes:
* different alerts format
* rewritten storage (some config flags removed)
Closes-Bug: PROD-16609
Change-Id: I805fa322e4744e98177d6c3e29589ebc6fb917a2
This option is usefull when all alarms should be send to two or more receivers simultaneously - without need to use continue parameter
Change-Id: I6b8d833776549614d0381378a2e92cb5bc05af77
Otherwise the alert fires as soon as Prometheus can't scrape a target.
It is too aggressive in case of transient connectivity issues or
endpoint restart.
Change-Id: Ib3de5b141db7a7f2397bf332844a9c44d38f2d3c
On compute nodes we get:
cmp01.virtual-mcp11-contrail.local:
----------
ID: libvirt_exporter_service
Function: service.running
Name: libvirt-exporter
Result: False
Comment: The following requisites were not found:
watch:
file: libvirt_exporter_service_config_file
Change-Id: I31069ec37702f8466da27b76c0becc6802cea272
This change allows to enable the monitoring only for the K8S
infrastructure (default) or for both the infrastructure and the
workloads running on top of K8S.
Change-Id: Idc68576f2d64840c0454fadc204174027bbaa454
Tested succesfully with the following test cases:
source_labels: instance
source_labels: instance,__name__,test
source_labels: [instance,__name__,test]
source_labels:
- instance
- __name__
- test
Change-Id: I9750bb0a4204335dd202cc4020445e850824c497
Depends-On: I2f85fe2f9cc67d2d03d5e8dad5ac0b74e82b0fcc
This change will enable possibility to add custom
labels to alerts.
This can be used ex. to provite routing key for
pushkin (notification service)
If given key already exists in alert, it will NOT
be overriden.
prometheus:
server:
config:
alert:
labels_add:
route: 'mail,skype'
Change-Id: I0b21d8a9695e8a15d0d6b80e6ddd0476682c80c2
For some alerts we use reclass to get variables
(ex. threshold). To get those reclass variables
we use `prometheus_server` dict, which on nodes
without prometheus class is None.
We need to provide at least empty dict in that
case.
Change-Id: Ic1f2c1776c4852701dcf9f8a76f933926b2d9aec