Fix divide by 0 bug in exemplar sharding calculations (#3936)

* Fix divide by 0 bug in exemplar sharding calculations * changelog
grafana · Aug 5, 2024 · 7dc8134 · 7dc8134
1 parent d3a9caf
commit 7dc8134
Show file tree

Hide file tree

Showing 2 changed files with 10 additions and 2 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -53,6 +53,7 @@
 * [BUGFIX] Fix frontend parsing error on cached responses [#3759](https://github.com/grafana/tempo/pull/3759) (@mdisibio)
 * [BUGFIX] Fix autocomplete of a query using scoped instrinsics [#3865](https://github.com/grafana/tempo/pull/3865) (@mdisibio)
 * [BUGFIX] Fix metrics query histograms and quantiles on traceDuration [#3879](https://github.com/grafana/tempo/pull/3879) (@mdisibio)
+* [BUGFIX] Fix divide by 0 bug in query frontend exemplar calculations [#3936](https://github.com/grafana/tempo/pull/3936) (@mdisibio)
 * [BUGFIX] max_global_traces_per_user: take into account ingestion.tenant_shard_size when converting to local limit [#3618](https://github.com/grafana/tempo/pull/3618) (@kvrhdn)
 * [BUGFIX] Fix http connection reuse on GCP and AWS by reading io.EOF through the http body. [#3760](https://github.com/grafana/tempo/pull/3760) (@bmteller)
 * [BUGFIX] Improved handling of complete blocks in localblocks processor after enabling flusing [#3805](https://github.com/grafana/tempo/pull/3805) (@mapno)

diff --git a/modules/frontend/metrics_query_range_sharder.go b/modules/frontend/metrics_query_range_sharder.go
@@ -368,7 +368,7 @@ func (s *queryRangeSharder) buildBackendRequests(ctx context.Context, tenantID s
 
 	queryHash := hashForQueryRangeRequest(&searchReq)
 
-	exemplars := s.exemplarsPerShard(uint32(len(metas)))
+	exemplarsPerBlock := s.exemplarsPerShard(uint32(len(metas)))
 	for _, m := range metas {
 		if m.EndTime.Before(m.StartTime) {
 			// Ignore blocks with bad timings from debugging
@@ -380,6 +380,13 @@ func (s *queryRangeSharder) buildBackendRequests(ctx context.Context, tenantID s
 			continue
 		}
 
+		exemplars := exemplarsPerBlock
+		if exemplars > 0 {
+			// Scale the number of exemplars per block to match the size
+			// of each sub request on this block. For very small blocks or other edge cases, return at least 1.
+			exemplars = max(uint32(float64(exemplars)*float64(m.TotalRecords)/float64(pages)), 1)
+		}
+
 		for startPage := 0; startPage < int(m.TotalRecords); startPage += pages {
 			subR := parent.Clone(ctx)
 
@@ -415,7 +422,7 @@ func (s *queryRangeSharder) buildBackendRequests(ctx context.Context, tenantID s
 				Size_:            m.Size,
 				FooterSize:       m.FooterSize,
 				DedicatedColumns: dc,
-				Exemplars:        max(exemplars/(m.TotalRecords/uint32(pages)), 1),
+				Exemplars:        exemplars,
 			}
 
 			subR = api.BuildQueryRangeRequest(subR, queryRangeReq)