refactor(test): duplication with fullbackup, tar content hashing (#7828)

"taking a fullbackup" is an ugly multi-liner copypasted in multiple
places, most recently with timeline ancestor detach tests. move it under
`PgBin` which is not a great place, but better than yet another utility
function.

Additionally:
- cleanup `psql_env` repetition (PgBin already configures that)
- move the backup tar comparison as a yet another free utility function
- use backup tar comparison in `test_import.py` where a size check was
done previously
- cleanup extra timeline creation from test

Cc: #7715

test_runner/fixtures/neon_fixtures.py

@@ -2788,6 +2788,28 @@ def get_pg_controldata_checkpoint_lsn(self, pgdata: str) -> Lsn:
         log.info(f"last checkpoint at {checkpoint_lsn}")
         return Lsn(checkpoint_lsn)
 
+    def take_fullbackup(
+        self,
+        pageserver: NeonPageserver,
+        tenant: TenantId,
+        timeline: TimelineId,
+        lsn: Lsn,
+        output: Path,
+    ):
+        """
+        Request fullbackup from pageserver, store it at 'output'.
+        """
+        cmd = [
+            "psql",
+            "--no-psqlrc",
+            pageserver.connstr(),
+            "-c",
+            f"fullbackup {tenant} {timeline} {lsn}",
+            "-o",
+            str(output),
+        ]
+        self.run_capture(cmd)
+
 
 @pytest.fixture(scope="function")
 def pg_bin(test_output_dir: Path, pg_distrib_dir: Path, pg_version: PgVersion) -> PgBin:

test_runner/fixtures/utils.py

@@ -4,19 +4,24 @@
 import os
 import re
 import subprocess
+import tarfile
 import threading
 import time
+from hashlib import sha256
 from pathlib import Path
 from typing import (
+    IO,
     TYPE_CHECKING,
     Any,
     Callable,
     Dict,
     Iterable,
     List,
     Optional,
+    Set,
     Tuple,
     TypeVar,
+    Union,
 )
 from urllib.parse import urlencode
 
@@ -499,3 +504,48 @@ def __repr__(self) -> str:
 
     def __str__(self) -> str:
         return f"'aux-{self.value}'"
+
+
+def assert_pageserver_backups_equal(left: Path, right: Path, skip_files: Set[str]):
+    """
+    This is essentially:
+
+    lines=$(comm -3 \
+        <(mkdir left && cd left && tar xf "$left" && find . -type f -print0 | xargs sha256sum | sort -k2) \
+        <(mkdir right && cd right && tar xf "$right" && find . -type f -print0 | xargs sha256sum | sort -k2) \
+        | wc -l)
+    [ "$lines" = "0" ]
+
+    But in a more mac friendly fashion.
+    """
+    started_at = time.time()
+
+    def hash_extracted(reader: Union[IO[bytes], None]) -> bytes:
+        assert reader is not None
+        digest = sha256(usedforsecurity=False)
+        while True:
+            buf = reader.read(64 * 1024)
+            if not buf:
+                break
+            digest.update(buf)
+        return digest.digest()
+
+    def build_hash_list(p: Path) -> List[Tuple[str, bytes]]:
+        with tarfile.open(p) as f:
+            matching_files = (info for info in f if info.isreg() and info.name not in skip_files)
+            ret = list(
+                map(lambda info: (info.name, hash_extracted(f.extractfile(info))), matching_files)
+            )
+            ret.sort(key=lambda t: t[0])
+            return ret
+
+    left_list, right_list = map(build_hash_list, [left, right])
+
+    try:
+        assert 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")

test_runner/regress/test_fullbackup.py

@@ -1,7 +1,7 @@
 import os
 from pathlib import Path
 
-from fixtures.common_types import Lsn, TimelineId
+from fixtures.common_types import Lsn
 from fixtures.log_helper import log
 from fixtures.neon_fixtures import (
     NeonEnvBuilder,
@@ -19,17 +19,16 @@ def test_fullbackup(
     neon_env_builder: NeonEnvBuilder,
     pg_bin: PgBin,
     port_distributor: PortDistributor,
-    pg_distrib_dir: Path,
     test_output_dir: Path,
 ):
     env = neon_env_builder.init_start()
 
-    env.neon_cli.create_branch("test_fullbackup")
-    endpoint_main = env.endpoints.create_start("test_fullbackup")
+    # endpoint needs to be alive until the fullbackup so that we have
+    # prev_record_lsn for the vanilla_pg to start in read-write mode
+    # for some reason this does not happen if endpoint is shutdown.
+    endpoint_main = env.endpoints.create_start("main")
 
     with endpoint_main.cursor() as cur:
-        timeline = TimelineId(query_scalar(cur, "SHOW neon.timeline_id"))
-
         # data loading may take a while, so increase statement timeout
         cur.execute("SET statement_timeout='300s'")
         cur.execute(
@@ -41,17 +40,13 @@ def test_fullbackup(
         lsn = Lsn(query_scalar(cur, "SELECT pg_current_wal_insert_lsn()"))
         log.info(f"start_backup_lsn = {lsn}")
 
-    # Set LD_LIBRARY_PATH in the env properly, otherwise we may use the wrong libpq.
-    # PgBin sets it automatically, but here we need to pipe psql output to the tar command.
-    psql_env = {"LD_LIBRARY_PATH": str(pg_distrib_dir / "lib")}
-
     # Get and unpack fullbackup from pageserver
     restored_dir_path = env.repo_dir / "restored_datadir"
     os.mkdir(restored_dir_path, 0o750)
-    query = f"fullbackup {env.initial_tenant} {timeline} {lsn}"
     tar_output_file = test_output_dir / "fullbackup.tar"
-    cmd = ["psql", "--no-psqlrc", env.pageserver.connstr(), "-c", query, "-o", str(tar_output_file)]
-    pg_bin.run_capture(cmd, env=psql_env)
+    pg_bin.take_fullbackup(
+        env.pageserver, env.initial_tenant, env.initial_timeline, lsn, tar_output_file
+    )
     subprocess_capture(
         env.repo_dir, ["tar", "-xf", str(tar_output_file), "-C", str(restored_dir_path)]
     )
@@ -61,7 +56,7 @@ def test_fullbackup(
     # use resetwal to overwrite it
     pg_resetwal_path = os.path.join(pg_bin.pg_bin_path, "pg_resetwal")
     cmd = [pg_resetwal_path, "-D", str(restored_dir_path)]
-    pg_bin.run_capture(cmd, env=psql_env)
+    pg_bin.run_capture(cmd)
 
     # Restore from the backup and find the data we inserted
     port = port_distributor.get_port()

test_runner/regress/test_import.py

@@ -21,7 +21,7 @@
     wait_for_upload,
 )
 from fixtures.remote_storage import RemoteStorageKind
-from fixtures.utils import subprocess_capture
+from fixtures.utils import assert_pageserver_backups_equal, subprocess_capture
 
 
 def test_import_from_vanilla(test_output_dir, pg_bin, vanilla_pg, neon_env_builder):
@@ -248,15 +248,9 @@ def _import(
     path to the backup archive file"""
     log.info(f"start_backup_lsn = {lsn}")
 
-    # Set LD_LIBRARY_PATH in the env properly, otherwise we may use the wrong libpq.
-    # PgBin sets it automatically, but here we need to pipe psql output to the tar command.
-    psql_env = {"LD_LIBRARY_PATH": str(pg_distrib_dir / "lib")}
-
     # Get a fullbackup from pageserver
-    query = f"fullbackup { env.initial_tenant} {timeline} {lsn}"
     tar_output_file = test_output_dir / "fullbackup.tar"
-    cmd = ["psql", "--no-psqlrc", env.pageserver.connstr(), "-c", query, "-o", str(tar_output_file)]
-    pg_bin.run_capture(cmd, env=psql_env)
+    pg_bin.take_fullbackup(env.pageserver, env.initial_tenant, timeline, lsn, tar_output_file)
 
     # Stop the first pageserver instance, erase all its data
     env.endpoints.stop_all()
@@ -305,22 +299,11 @@ def _import(
     assert endpoint.safe_psql("select count(*) from tbl") == [(expected_num_rows,)]
 
     # Take another fullbackup
-    query = f"fullbackup { tenant} {timeline} {lsn}"
     new_tar_output_file = test_output_dir / "fullbackup-new.tar"
-    cmd = [
-        "psql",
-        "--no-psqlrc",
-        env.pageserver.connstr(),
-        "-c",
-        query,
-        "-o",
-        str(new_tar_output_file),
-    ]
-    pg_bin.run_capture(cmd, env=psql_env)
+    pg_bin.take_fullbackup(env.pageserver, tenant, timeline, lsn, new_tar_output_file)
 
     # Check it's the same as the first fullbackup
-    # TODO pageserver should be checking checksum
-    assert os.path.getsize(tar_output_file) == os.path.getsize(new_tar_output_file)
+    assert_pageserver_backups_equal(tar_output_file, new_tar_output_file, set())
 
     # Check that gc works
     pageserver_http = env.pageserver.http_client()

test_runner/regress/test_next_xid.py

@@ -5,7 +5,7 @@
 
 from fixtures.common_types import Lsn, TenantId, TimelineId
 from fixtures.log_helper import log
-from fixtures.neon_fixtures import NeonEnvBuilder, wait_for_wal_insert_lsn
+from fixtures.neon_fixtures import NeonEnvBuilder, PgBin, wait_for_wal_insert_lsn
 from fixtures.pageserver.utils import (
     wait_for_last_record_lsn,
 )
@@ -71,22 +71,17 @@ def test_next_xid(neon_env_builder: NeonEnvBuilder):
 def test_import_at_2bil(
     neon_env_builder: NeonEnvBuilder,
     test_output_dir: Path,
-    pg_distrib_dir: Path,
-    pg_bin,
+    pg_bin: PgBin,
     vanilla_pg,
 ):
     neon_env_builder.enable_pageserver_remote_storage(RemoteStorageKind.LOCAL_FS)
     env = neon_env_builder.init_start()
     ps_http = env.pageserver.http_client()
 
-    # Set LD_LIBRARY_PATH in the env properly, otherwise we may use the wrong libpq.
-    # PgBin sets it automatically, but here we need to pipe psql output to the tar command.
-    psql_env = {"LD_LIBRARY_PATH": str(pg_distrib_dir / "lib")}
-
     # Reset the vanilla Postgres instance to somewhat before 2 billion transactions.
     pg_resetwal_path = os.path.join(pg_bin.pg_bin_path, "pg_resetwal")
     cmd = [pg_resetwal_path, "--next-transaction-id=2129920000", "-D", str(vanilla_pg.pgdatadir)]
-    pg_bin.run_capture(cmd, env=psql_env)
+    pg_bin.run_capture(cmd)
 
     vanilla_pg.start()
     vanilla_pg.safe_psql("create user cloud_admin with password 'postgres' superuser")