Why do statistics, issue counts, and release summaries disagree? #78
|
Which counts can be compared directly, and how should I compare errors between two releases without mixing time scopes? |
Replies: 1 comment
|
Statistics default to the last 24 hours. An unbounded search can include older events, and an issue's For a useful comparison, choose explicit identical UTC
Breakdowns return their top 20 values with
|
Statistics default to the last 24 hours. An unbounded search can include older events, and an issue's
event_countis live and all-time. Those values do not have to agree.For a useful comparison, choose explicit identical UTC
fromandtobounds and align environment, level, text, tag, and issue filters. Then run statistics for each release using the same bounds and bucket size. Record those inputs with the result. In MCP, statistics do not expose a status filter; use REST if the comparison requires one.total_countcounts project events in the chosen occurrence-time range before other filters.matching_count,unique_issues, buckets, and breakdowns apply the filters. A release filter can t…