From 4b750a6940fc132f56e1ee8c77c056bb241f826a Mon Sep 17 00:00:00 2001
From: Bilal Akhtar <bilal@cockroachlabs.com>
Date: Wed, 17 Jan 2024 17:17:19 -0500
Subject: [PATCH] db: fix sstable bound inversion in ingestSynthesizeShared

Previously, for shared file ingestions, we could have seqnum
substitution result in invalid bounds (eg. smallest > largest).
This change detects this case as an artifact of seqnum substitution
and addresses it.

Fixes #3225.
---
 ingest.go              |  8 ++++++
 testdata/ingest_shared | 57 ++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 65 insertions(+)

diff --git a/ingest.go b/ingest.go
index 957dc5454f..98ef20a47e 100644
--- a/ingest.go
+++ b/ingest.go
@@ -147,6 +147,14 @@ func ingestSynthesizeShared(
 		if sm.LargestPointKey.IsExclusiveSentinel() {
 			largestPointKey = base.MakeRangeDeleteSentinelKey(sm.LargestPointKey.UserKey)
 		}
+		if opts.Comparer.Equal(smallestPointKey.UserKey, largestPointKey.UserKey) &&
+			smallestPointKey.Trailer < largestPointKey.Trailer {
+			// We get kinds from the sender, however we substitute our own sequence
+			// numbers. This can result in cases where an sstable [b#5,SET-b#4,DELSIZED]
+			// becomes [b#0,SET-b#0,DELSIZED] when we synthesize it here, but the
+			// kinds need to be reversed now because DelSized > Set.
+			smallestPointKey, largestPointKey = largestPointKey, smallestPointKey
+		}
 		meta.ExtendPointKeyBounds(opts.Comparer.Compare, smallestPointKey, largestPointKey)
 	}
 	if err := meta.Validate(opts.Comparer.Compare, opts.Comparer.FormatKey); err != nil {
diff --git a/testdata/ingest_shared b/testdata/ingest_shared
index 2c60c29aca..b7f659eef0 100644
--- a/testdata/ingest_shared
+++ b/testdata/ingest_shared
@@ -1240,3 +1240,60 @@ c: (., [c-cc) @7=foo UPDATED)
 dd: (., [dd-e) @7=foo UPDATED)
 .
 .
+
+# Regression test for https://github.com/cockroachdb/pebble/issues/3225
+# Handle cases where an sstable starts and ends at the same user key but different
+# internal key, and the ordering of internal key kinds changes due to seqnum
+# substitution.
+
+reset
+----
+
+switch 1
+----
+ok
+
+batch
+set a foo
+----
+
+file-only-snapshot s11
+ a z
+----
+ok
+
+batch
+del a
+----
+
+flush
+----
+
+compact a-z
+----
+ok
+
+lsm
+----
+6:
+  000005:[a#11,DEL-a#10,SET]
+
+replicate 1 2 a z
+----
+replicated 1 shared SSTs
+
+switch 2
+----
+ok
+
+lsm
+----
+6:
+  000005:[a#10,SET-a#10,DEL]
+
+iter
+first
+next
+----
+.
+.