info(...) function fails on time series churn #17134
Description
What did you do?
Running this test:
load 1m
data_metric{instance="a", job="work"} 10 20 30
data_metric{instance="b", job="work"} 11 21 31
load 1m
info_metric{instance="b", job="work", state="stopped"} 1 1
info_metric{instance="b", job="work", state="running"} _ _ 1
info_metric{instance="a", job="work", state="running"} 1 1 1
eval range from 0 to 2m step 1m info(data_metric, {__name__="info_metric"})
data_metric{instance="a", job="work", state="running"} 10 20 30
data_metric{instance="b", job="work", state="stopped"} 11 21
data_metric{instance="b", job="work", state="running"} _ _ 31
What did you expect to see?
Test passing.
What did you see instead? Under which circumstances?
Test fails with error message:
error evaluating query "info(data_metric, {__name__=\"info_metric\"})" (line 127): found duplicate series for info metric: existing {__name__="info_metric", instance="b", job="work", state="stopped"} @ 0, new {__name__="info_metric", instance="a", job="work", state="running"} @ 0
System information
No response
Prometheus version
3.5.0
Prometheus configuration file
Alertmanager version
Alertmanager configuration file
Logs