test_import_from_pageserver_small: try to make less flaky (#7843)

With #7828 and proper fullbackup testing the test became flaky
([evidence]).

- produce better assertion messages in `assert_pageserver_backups_equal`
- use read only endpoint to confirm the row count

[evidence]:
https://neon-github-public-dev.s3.amazonaws.com/reports/pr-7839/9192447962/index.html#suites/89cfa994d71769e01e3fc4f475a1f3fa/49009214d0f8b8ce

test_runner/fixtures/utils.py

@@ -541,11 +541,22 @@ def build_hash_list(p: Path) -> List[Tuple[str, bytes]]:
 
     left_list, right_list = map(build_hash_list, [left, right])
 
-    try:
-        assert len(left_list) == len(right_list)
+    assert len(left_list) == len(
+        right_list
+    ), f"unexpected number of files on tar files, {len(left_list)} != {len(right_list)}"
 
-        for left_tuple, right_tuple in zip(left_list, right_list):
-            assert left_tuple == right_tuple
-    finally:
-        elapsed = time.time() - started_at
-        log.info(f"assert_pageserver_backups_equal completed in {elapsed}s")
+    mismatching = set()
+
+    for left_tuple, right_tuple in zip(left_list, right_list):
+        left_path, left_hash = left_tuple
+        right_path, right_hash = right_tuple
+        assert (
+            left_path == right_path
+        ), f"file count matched, expected these to be same paths: {left_path}, {right_path}"
+        if left_hash != right_hash:
+            mismatching.add(left_path)
+
+    assert len(mismatching) == 0, f"files with hash mismatch: {mismatching}"
+
+    elapsed = time.time() - started_at
+    log.info(f"assert_pageserver_backups_equal completed in {elapsed}s")

test_runner/regress/test_import.py

@@ -163,7 +163,7 @@ def test_import_from_pageserver_small(
 
     num_rows = 3000
     lsn = _generate_data(num_rows, endpoint)
-    _import(num_rows, lsn, env, pg_bin, timeline, env.pg_distrib_dir, test_output_dir)
+    _import(num_rows, lsn, env, pg_bin, timeline, test_output_dir)
 
 
 @pytest.mark.timeout(1800)
@@ -193,9 +193,7 @@ def test_import_from_pageserver_multisegment(
     log.info(f"timeline logical size = {logical_size / (1024 ** 2)}MB")
     assert logical_size > 1024**3  # = 1GB
 
-    tar_output_file = _import(
-        num_rows, lsn, env, pg_bin, timeline, env.pg_distrib_dir, test_output_dir
-    )
+    tar_output_file = _import(num_rows, lsn, env, pg_bin, timeline, test_output_dir)
 
     # Check if the backup data contains multiple segment files
     cnt_seg_files = 0
@@ -235,7 +233,6 @@ def _import(
     env: NeonEnv,
     pg_bin: PgBin,
     timeline: TimelineId,
-    pg_distrib_dir: Path,
     test_output_dir: Path,
 ) -> Path:
     """Test importing backup data to the pageserver.
@@ -295,7 +292,7 @@ def _import(
     wait_for_upload(client, tenant, timeline, lsn)
 
     # Check it worked
-    endpoint = env.endpoints.create_start(endpoint_id, tenant_id=tenant)
+    endpoint = env.endpoints.create_start(endpoint_id, tenant_id=tenant, lsn=lsn)
     assert endpoint.safe_psql("select count(*) from tbl") == [(expected_num_rows,)]
 
     # Take another fullbackup