The use case is to alert on the disk usage deviation from the mean, within
a cluster of hosts. The alert should indicate which host deviates from the
median disk usage.
So this could be written as:
metric format: `disk_usage_pct{cluster=foo, instance=foo01}`
`max(disk_usage_pct) by (cluster)
On 17/12/2020 22:43, Raghu Udiyar wrote:
I supposed it wouldn't be an aggregation - its more like a filter.
What I want is to return the series with the max value, with all
labels intact. --
It does sound a bit like you are wanting topk. Does that do what you are
looking for?
--
You
I supposed it wouldn't be an aggregation - its more like a filter. What I
want is to return the series with the max value, with all labels intact.
--
You received this message because you are subscribed to the Google Groups
"Prometheus Users" group.
To unsubscribe from this group and stop
On 17/12/2020 17:57, Raghu Udiyar wrote:
Hello
When using aggregations such as `max(metric1) by (lable1)`, all the
other labels are stripped. Is it possible to preserve the labels, akin
to *selecting* the `max` series.
This is more of a filter, not an aggregation.
I'm not quite sure I
Hello
When using aggregations such as `max(metric1) by (lable1)`, all the other
labels are stripped. Is it possible to preserve the labels, akin to
*selecting* the `max` series.
This is more of a filter, not an aggregation.
Thanks
--
You received this message because you are subscribed to
5 matches
Mail list logo
