Yes, using a recording rule will work fine too.
> Prometheus seems unable to do a min_over_time() from a sum()
That's because the input to min_over_time is a range vector - which covers
multiple points over a period of time, not a single instant of time.
As I said before, you can turn an instant vector into a range vector using
a subquery. Something like:
min_over_time((sum by (instance) (apt_upgrades_pending))[48h:5m])
A subquery is just like specifying a time range on a metric, but you have
to specify the sampling interval (here "5m"). The instant query is repeated
at this interval to generate a range of data points. You don't need to do
this on a plain metric because it just uses all the individual samples
which were scraped.
On Sunday, 21 November 2021 at 22:24:44 UTC [email protected] wrote:
> > sum by (instance) (apt_upgrades_pending) > 0
>
> I think that your suggestion will work. I'll try it out for the next days.
> Many thanks.
>
> About the other way, for completeness: Instead of using a subquery, I
> wonder whether this would work too:
>
> - record: apt_upgrades_pending:sum_grouped_by_instance
> expr: sum by ( instance ) ( apt_upgrades_pending )
>
> - alert: AptNotUpdatingPackages
> expr: min_over_time( apt_upgrades_pending:sum_grouped_by_instance[7d]
> ) > 0
> for: 0m
> labels:
> severity: warning
>
> Prometheus seems unable to do a min_over_time() from a sum(). Is that kind
> of intermediate step with "record" a generic way to overcome this
> limitation? Or have I missed something?
>
--
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/ab1ed651-2a60-4ccd-ac79-9c237b0a87dcn%40googlegroups.com.
