diff --git a/pgxn/neon/walproposer_pg.c b/pgxn/neon/walproposer_pg.c
index 7c23ff569f..c6188d7f30 100644
--- a/pgxn/neon/walproposer_pg.c
+++ b/pgxn/neon/walproposer_pg.c
@@ -1843,7 +1843,6 @@ CombineHotStanbyFeedbacks(HotStandbyFeedback *hs, WalProposer *wp)
 
 	for (int i = 0; i < wp->n_safekeepers; i++)
 	{
-		elog(LOG, "hs.ts=%ld hs.xmin=%d", wp->safekeeper[i].appendResponse.hs.ts, XidFromFullTransactionId(wp->safekeeper[i].appendResponse.hs.xmin));
  		if (wp->safekeeper[i].appendResponse.hs.ts != 0)
 		{
 			HotStandbyFeedback *skhs = &wp->safekeeper[i].appendResponse.hs;
@@ -1919,7 +1918,7 @@ walprop_pg_process_safekeeper_feedback(WalProposer *wp, XLogRecPtr commitLsn)
 	if (hsFeedback.ts != 0 && memcmp(&hsFeedback, &quorumFeedback.hs, sizeof hsFeedback) != 0)
 	{
 		quorumFeedback.hs = hsFeedback;
-		elog(LOG, "ProcessStandbyHSFeedback(xmin=%d, catalog_xmin=%d", XidFromFullTransactionId(hsFeedback.xmin), XidFromFullTransactionId(hsFeedback.catalog_xmin));
+		elog(DEBUG2, "ProcessStandbyHSFeedback(xmin=%d, catalog_xmin=%d", XidFromFullTransactionId(hsFeedback.xmin), XidFromFullTransactionId(hsFeedback.catalog_xmin));
 		ProcessStandbyHSFeedback(hsFeedback.ts,
 								 XidFromFullTransactionId(hsFeedback.xmin),
 								 EpochFromFullTransactionId(hsFeedback.xmin),
diff --git a/test_runner/fixtures/neon_fixtures.py b/test_runner/fixtures/neon_fixtures.py
index a107558abb..4361bb0eb7 100644
--- a/test_runner/fixtures/neon_fixtures.py
+++ b/test_runner/fixtures/neon_fixtures.py
@@ -4087,6 +4087,7 @@ def tenant_get_shards(
         # Assume an unsharded tenant
         return [(TenantShardId(tenant_id, 0, 0), override_pageserver or env.pageserver)]
 
+
 def wait_replica_caughtup(primary: Endpoint, secondary: Endpoint):
     primary_lsn = primary.safe_psql_scalar(
         "SELECT pg_current_wal_flush_lsn()::text", log_query=False
@@ -4101,6 +4102,7 @@ def wait_replica_caughtup(primary: Endpoint, secondary: Endpoint):
             return
         time.sleep(1)
 
+
 def wait_for_last_flush_lsn(
     env: NeonEnv,
     endpoint: Endpoint,
diff --git a/test_runner/regress/test_replication_lag.py b/test_runner/regress/test_replication_lag.py
index 01185e403e..c386d6e50f 100644
--- a/test_runner/regress/test_replication_lag.py
+++ b/test_runner/regress/test_replication_lag.py
@@ -1,3 +1,4 @@
+import time
 import threading
 
 from fixtures.log_helper import log
@@ -7,6 +8,7 @@ from fixtures.neon_fixtures import NeonEnv, PgBin, wait_replica_caughtup
 def test_replication_lag(neon_simple_env: NeonEnv, pg_bin: PgBin):
     env = neon_simple_env
     n_iterations = 10
+    max_retries = 10
 
     # Use aggressive GC and checkpoint settings
     tenant, _ = env.neon_cli.create_tenant(
@@ -34,6 +36,7 @@ def test_replication_lag(neon_simple_env: NeonEnv, pg_bin: PgBin):
 
         with env.endpoints.new_replica_start(origin=primary, endpoint_id="secondary") as secondary:
             wait_replica_caughtup(primary, secondary)
+            time.sleep(1) # Without this sleep replica sometime failed to find relation: could not open relation with OID 16404
             for _ in range(1, n_iterations):
                 primary_lsn = primary.safe_psql_scalar(
                     "SELECT pg_current_wal_flush_lsn()::text", log_query=False
@@ -41,7 +44,19 @@ def test_replication_lag(neon_simple_env: NeonEnv, pg_bin: PgBin):
                 secondary_lsn = secondary.safe_psql_scalar(
                     "SELECT pg_last_wal_replay_lsn()", log_query=False
                 )
-                balance = secondary.safe_psql_scalar("select sum(abalance) from pgbench_accounts")
+                retries = 0
+                while True:
+                    try:
+                        balance = secondary.safe_psql_scalar(
+                            "select sum(abalance) from pgbench_accounts"
+                        )
+                        break
+                    except Exception as error:
+                        print(f"Query failed: {error}")
+                        if retries < max_retries:
+                            retries += 1
+                        else:
+                            raise
                 log.info(
                     f"primary_lsn={primary_lsn}, secondary_lsn={secondary_lsn}, balance={balance}"
                 )