Maybe what you're trying to do is: sum by (consumergroup, topic) (rate(kafka_consumergroup_current_offset[5m]) * 60) == 0 unless sum by (topic) (rate(kafka_consumergroup_current_offset[5m]) * 60) < 1
That is: alert on any combination of (consumergroup,topic) where the 5-minute rate of consumption is zero, unless the rate for that topic across all consumers is less than 1 per minute. As far as I can tell, kafka_consumergroup_current_offset is a counter, and therefore you should use either rate() or increase(). The only difference is that rate(foo[5m]) gives the increase per second, while increase(foo[5m]) gives the increase per 5 minutes. Hence: rate(kafka_consumergroup_current_offset[5m]) * 60 increase(kafka_consumergroup_current_offset[5m]) / 5 should both be the same, giving the per-minute increase. On Friday 19 April 2024 at 18:30:21 UTC+1 Brian Candler wrote: > Sorry, first link was wrong. > https://groups.google.com/g/prometheus-users/c/IeW_3nyGkR0/m/unto0oGQAQAJ > https://groups.google.com/g/prometheus-users/c/83pEAX44L3M/m/E20UmVJyBQAJ > > On Friday 19 April 2024 at 18:28:29 UTC+1 Brian Candler wrote: > >> Can you give examples of the metrics in question, and what conditions >> you're trying to check for? >> >> Looking at your specific PromQL query: Firstly, in my experience, it's >> very unusual in Prometheus queries to use ==bool or >bool, and in this >> specific case definitely seems to be wrong. >> >> Secondly, you won't be able to join the LH and RH sides of your >> expression with "and" unless either they have exactly the same label sets, >> or you modify your condition using "and on (...)" or "and ignoring (...)". >> >> "and" is a vector intersection operator, where the result vector includes >> a value if the labels match, and the value is taken from the LHS, and that >> means it doesn't combine the values like you might be used to in other >> programming languages. For example, >> >> vector(0) and vector(1) => value is 0 >> vector(1) and vector(0) => value is 1 >> vector(42) and vector(99) => value is 42 >> >> This is as described in the documentation >> <https://prometheus.io/docs/prometheus/latest/querying/operators/#logical-set-binary-operators> >> : >> >> vector1 and vector2 results in a vector consisting of the elements of >> vector1 for which there are elements in vector2 with exactly matching >> label sets. Other elements are dropped. The metric name and values are >> carried over from the left-hand side vector. >> >> PromQL alerts on the presence of values, and in PromQL you need to think >> in terms of "what (labelled) values are present or absent in this vector", >> using the "and/unless" operators to suppress elements in the result vector, >> and the "or" operator to add additional elements to the result vector. >> >> Maybe these explanations help: >> https://groups.google.com/g/prometheus-users/c/IeW_3nyGkR0/m/NH2_CRPaAQAJ >> https://groups.google.com/g/prometheus-users/c/83pEAX44L3M/m/E20UmVJyBQAJ >> >> On Friday 19 April 2024 at 16:31:23 UTC+1 Robson Jose wrote: >> >>> Good afternoon, I would like to know if it is possible to do this query, >>> the value that should return is applications with a value of 0 in the first >>> query and greater than one in the 2nd >>> >>> ( >>> sum by (consumergroup, topic) >>> (delta(kafka_consumergroup_current_offset{}[5m])/5) ==bool 0 >>> ) >>> and ( >>> sum by (topic) (delta(kafka_consumergroup_current_offset{}[5m])/5) >>> >bool 1 >>> ) >>> >> -- 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/9796ea54-47c9-47dc-8f87-460de1468a66n%40googlegroups.com.
