From d74aff6d2a04c0bebff85d5c8e43d4eb42dae939 Mon Sep 17 00:00:00 2001
From: Abhijat Malviya <abhijat@dragonflydb.io>
Date: Sat, 10 May 2025 11:08:46 +0530
Subject: [PATCH] fix(hset_family): Fix val being overwritten by TTL

When hset is loaded from rdb, if a ttl is specified by the user, the
code recreates a string_view on top of the same memory location,
tset_blob_, that was used for val earlier. This causes the val string
view to point to the same value now as TTL has. For example if the val
string view was 'x' earlier (points to tset_blob_, size 1), and ttl is
7777777, val now becomes '7'.

To fix this val is now given it's own string. TTL is kept as string
view as the pointer is not reused anywhere in the following loop.

Signed-off-by: Abhijat Malviya <abhijat@dragonflydb.io>
---
 src/server/rdb_load.cc              |  3 +-
 tests/dragonfly/replication_test.py | 43 +++++++++++++++++++----------
 2 files changed, 31 insertions(+), 15 deletions(-)

diff --git a/src/server/rdb_load.cc b/src/server/rdb_load.cc
index 01b2d8653..a3444e92a 100644
--- a/src/server/rdb_load.cc
+++ b/src/server/rdb_load.cc
@@ -482,11 +482,12 @@ void RdbLoaderBase::OpaqueObjLoader::CreateHMap(const LoadTrace* ltrace) {
       }
     });
     std::string key;
+    std::string val;
     for (size_t i = 0; i < ltrace->arr.size(); i += increment) {
       // ToSV may reference an internal buffer, therefore we can use only before the
       // next call to ToSV. To workaround, copy the key locally.
       key = ToSV(ltrace->arr[i].rdb_var);
-      string_view val = ToSV(ltrace->arr[i + 1].rdb_var);
+      val = ToSV(ltrace->arr[i + 1].rdb_var);
 
       if (ec_)
         return;
diff --git a/tests/dragonfly/replication_test.py b/tests/dragonfly/replication_test.py
index 3e1bba444..6a31de040 100644
--- a/tests/dragonfly/replication_test.py
+++ b/tests/dragonfly/replication_test.py
@@ -1,23 +1,18 @@
-import random
-
-from itertools import chain, repeat
-import re
-import pytest
-import asyncio
-import async_timeout
 import platform
-import pymemcache
-import logging
+import shutil
 import tarfile
 import urllib.request
-import shutil
-from redis import asyncio as aioredis
-from .utility import *
-from .instance import DflyInstanceFactory, DflyInstance
-from .seeder import Seeder as SeederV2
+from itertools import chain, repeat
+
+import async_timeout
+import pymemcache
+
 from . import dfly_args
+from .instance import DflyInstanceFactory, DflyInstance
 from .proxy import Proxy
 from .seeder import DebugPopulateSeeder
+from .seeder import Seeder as SeederV2
+from .utility import *
 
 ADMIN_PORT = 1211
 
@@ -3115,3 +3110,23 @@ async def test_partial_replication_on_same_source_master(df_factory, use_takeove
         lines = replica2.find_in_logs(f"Started full with localhost:{replica1.port}")
         assert len(lines) == 0
         assert len(replica1.find_in_logs("No partial sync due to diff")) > 0
+
+
+async def test_replicate_hset_with_expiry(df_factory: DflyInstanceFactory):
+    master = df_factory.create(proactor_threads=2)
+    replica = df_factory.create(proactor_threads=2)
+
+    master.start()
+    replica.start()
+
+    cm = master.client()
+    await cm.execute_command("HSETEX key 86400 name 1234")
+
+    cr = replica.client()
+    await cr.execute_command(f"REPLICAOF localhost {master.port}")
+    await wait_available_async(cr)
+
+    result = await cr.hgetall("key")
+
+    assert "name" in result
+    assert result["name"] == "1234"