On Friday, 30 October 2020 13:18:11 UTC, [email protected] wrote:
>
> Using different jobs to scrape different groups of same exporters.
Any particular reason for separating them over multiple jobs?
In any case, what problem does it give you? Your alerting rules can be
written without reference to the job label, e.g.
expr: count without (cpu, job) ( ... )
although it may be clearer to say what label(s) you *do* want to group by,
e.g.
expr: count by (instance) ( ... )
> At the same time I would like to make alerting rules be clean by
> introducing recording rules. My question is following approach fine?
I don't like having two recording rules writing to the same metric, and I'm
not sure if it's allowed. But again, I don't see why you want to separate
those rules by job anyway, when a single rule can cover both jobs. Why not:
- name: node-exporter.rules
rules:
- expr: |
count without (cpu) (
count without (mode) (
node_cpu_seconds_total
)
)
record: instance:node_num_cpu:sum
or:
- name: node-exporter.rules
rules:
- expr: |
count without (cpu) (
count without (mode) (
node_cpu_seconds_total{job=~"node_1|node_2"}
)
)
record: instance:node_num_cpu:sum
?
> I understand that in expr you can use logical operators, but than if
> recording metric is used for calculation some other recorded metrics it's
> going to break or get to complicated. Is there a clean way?
>
I'm not sure what you're asking, as you didn't give any example of how you
wanted to use a logical operator.
Logical operators come in two flavours:
foo > 0 is a filter. It allows the timeseries through only if its value
is > 0, and filters it out otherwise.
foo > bool 0 is a value. It always passes some value through, but that
value is either 0 or 1 depending on the condition.
Normally filters are used in alerting rules. Alerting rules trigger if one
or more timeseries passes through the filters.
--
You received this message because you are subscribed to the Google Groups
"Prometheus Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To view this discussion on the web visit
https://groups.google.com/d/msgid/prometheus-users/683dd70e-215a-42ff-bfa7-ded9901b62b1o%40googlegroups.com.
