Use `$__rate_interval` to make sure you get enough samples for `delta()` to
work. (delta(), rate(), increase(), function similarly).

On Mon, Jul 10, 2023 at 4:17 PM Helena Lundell <[email protected]> wrote:

> Thanks for the fast reply and sorry for my late response.
>
> We have a scraping on 15 seconds. My query interval is 30 min. So I should
> have data. And the Grafana dashboard below where I add difference(changed
> from delta) on legend values I see what I want.
> This difference from the Grafana panel is exactely what I am looking for
> and thought delta would give me.
>
> Is there an alternative to delta? Like first-last?
> On Saturday, July 1, 2023 at 3:07:22 PM UTC+2 Ben Kochie wrote:
>
>> Grafana defaults to a "minimum step" of 15 seconds. I bet your scrape
>> interval is longer than 15 seconds, so "$__interval" is matching no data.
>>
>> You can either adjust the datasource minimum step to your shortest scrape
>> interval, or set the "min step" in the query options to match the scrape
>> interval of your metric.
>>
>> On Sat, Jul 1, 2023 at 3:04 PM Helena Lundell <[email protected]> wrote:
>>
>>> I am trying to identify an alert for an increasing Kafka Lag. The
>>> metrics are in prometheus and I use Grafana 9.4 for visualisation.
>>>
>>> In Grafana I have a dasboard where I can see that the lag is increasing
>>> or decreasing. I have choosen to display the delta in the legend and it
>>> shows a number that does not really make sense, but it shows something.
>>>
>>> This is the base promql:
>>> avg by(consumergroup, topic, cluster)
>>> (kafka_consumergroup_lag{namespace="ns-kafka-int",
>>> consumergroup=~".*$container", cluster="$cluster"})
>>>
>>> When I add delta to this query I get no data:
>>> avg by(consumergroup, topic, cluster)
>>> (delta(kafka_consumergroup_lag{namespace="ns-kafka-int",
>>> consumergroup=~".*$container", cluster="$cluster"}[$__interval]))
>>>
>>> So adding delta gives me no result even though when I exclude delta I
>>> can see that the metric has been changing.
>>>
>>> The metric is a gaugue.
>>>
>>> Question:
>>> Am I doing something wrong?
>>> Can I trust the delta from the Grafana dashboard(someone might know)?
>>> Can I expect a positive or negative value to identify and increase?
>>>
>>> --
>>> 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/0f37aa72-dd90-4329-8e19-5587460b80c4n%40googlegroups.com
>>> <https://groups.google.com/d/msgid/prometheus-users/0f37aa72-dd90-4329-8e19-5587460b80c4n%40googlegroups.com?utm_medium=email&utm_source=footer>
>>> .
>>>
>> --
> 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/63cf733e-87bd-4530-9531-5a3a29556d8an%40googlegroups.com
> <https://groups.google.com/d/msgid/prometheus-users/63cf733e-87bd-4530-9531-5a3a29556d8an%40googlegroups.com?utm_medium=email&utm_source=footer>
> .
>

-- 
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/CABbyFmoW6S71ege9UVLiqvh-q8DxQ2VP_kjmSuVe0FqQ6ej_MA%40mail.gmail.com.

Reply via email to