diff --git a/docs/CHANGELOG.md b/docs/CHANGELOG.md
index 43084d925..7be173f82 100644
--- a/docs/CHANGELOG.md
+++ b/docs/CHANGELOG.md
@@ -32,9 +32,9 @@ See also [LTS releases](https://docs.victoriametrics.com/lts-releases/).
 
 * FEATURE: [vmbackupmanager](https://docs.victoriametrics.com/vmbackupmanager.html): allow specifying custom backup interval via `-backupInterval` command-line flag. See [this feature request](https://github.com/VictoriaMetrics/VictoriaMetrics/issues/5966).
 
+* BUGFIX: fix incorrect merge of partitions which could lead to losing records for IndexDB. See [this](https://github.com/VictoriaMetrics/VictoriaMetrics/issues/5959) and [this](https://github.com/VictoriaMetrics/VictoriaMetrics/issues/6069) issues.
 * BUGFIX: [vmbackupmanager](https://docs.victoriametrics.com/vmbackupmanager.html): fix panic when performing a restore from GCS. Previously, filters for GCS objects were not properly set which caused objects to be returned instead of prefixes.
 
-
 ## [v1.100.0](https://github.com/VictoriaMetrics/VictoriaMetrics/releases/tag/v1.100.0)
 
 Released at 2024-04-04
diff --git a/lib/mergeset/table.go b/lib/mergeset/table.go
index f22f2406f..b9202f4e5 100644
--- a/lib/mergeset/table.go
+++ b/lib/mergeset/table.go
@@ -198,7 +198,7 @@ func (riss *rawItemsShards) addIbsToFlush(tb *Table, ibsToFlush []*inmemoryBlock
 	}
 	riss.ibsToFlush = append(riss.ibsToFlush, ibsToFlush...)
 	if len(riss.ibsToFlush) >= maxBlocksPerShard*cgroup.AvailableCPUs() {
-		ibsToMerge = ibsToFlush
+		ibsToMerge = riss.ibsToFlush
 		riss.ibsToFlush = nil
 	}
 	riss.ibsToFlushLock.Unlock()