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
On the stage of merging statuses from each storage node, the same label could have different values on each node.
But vmselect does not receive the actual values, so it can't compare them and find intersections or sum up unique values. Instead, it picks a node with the highest number of unique values and assumes it is the closest one to reality.
It would be good to add some explanations and expectations to the numbers shown on Cardinality Explorer.
The text was updated successfully, but these errors were encountered:
Starting from v1.94.0, cardinality explorer tab shows the information about possible inaccuracy in the provided statistics for cluster version of VictoriaMetrics. Closing the feature request as done.
On the stage of merging statuses from each storage node, the same label could have different values on each node.
But
vmselect
does not receive the actual values, so it can't compare them and find intersections or sum up unique values. Instead, it picks a node with the highest number of unique values and assumes it is the closest one to reality.It would be good to add some explanations and expectations to the numbers shown on Cardinality Explorer.
The text was updated successfully, but these errors were encountered: