You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Is your feature request related to a problem? Please describe.
Cortex provides a MaxLabelsSizeBytes limit for the max combined size in bytes of all labels and label values for a given metric name, but does not emit the metric this is compared to. Exposing such a metric would help users to know how close they are to hitting the limit in case they intend to add more labels to a metric name.
Describe the solution you'd like
Expose the labelSizeBytes as a native histogram metric that tenants can monitor.
The text was updated successfully, but these errors were encountered:
I like the proposal to add a metric for labelSizeBytes and allowing tenants to monitor it.
My only concern is how much it would be useful to track the exact metricName which the high labelSize is coming from.
Would you add a label for tenant and metricName? Having a metricName label even with native histogram can create a lot of data, no?
I think we could just add the tenant but not the metric name. If the tenant exceeds the label size limit the error returned will include the entire label set so the tenant should be able to deduce the metric name from there.
Is your feature request related to a problem? Please describe.
Cortex provides a
MaxLabelsSizeBytes
limit for the max combined size in bytes of all labels and label values for a given metric name, but does not emit the metric this is compared to. Exposing such a metric would help users to know how close they are to hitting the limit in case they intend to add more labels to a metric name.Describe the solution you'd like
Expose the
labelSizeBytes
as a native histogram metric that tenants can monitor.The text was updated successfully, but these errors were encountered: