To debug it, simply run the two halves separately: (a) pve_up == 0 (b) pve_guest_info
Match up the the timeseries from (a) and (b) with identical "id" value. I think you'll find at least one case where a particular value of "id" links to more than one (a) and more than one (b). group_left requires that multiple instances of (a) map to exactly one instance of (b). So another query to try is: count by (id) (pve_guest_info) > 1 and see if you can find multiple instances of pve_guest_info for the same "id". If you do, then the solution depends on the meaning of the metric. Is "id" only unique when in combination with some other label? Then join on both labels at once: (pve_up == 0) * on(id, foo) group_left(name) pve_guest_info where "foo" is this other label. (Clearly it has to exist on both pve_up and pve_guest_info metrics though) -- 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/0b5b8a61-c9cd-4704-b75f-2af79db06792n%40googlegroups.com.
