Allow to change compute safekeeper list without restart.

- Add --safekeepers option to neon_local reconfigure
- Add it to python Endpoint reconfigure
- Implement config reload in walproposer by restarting the whole bgw when
  safekeeper list changes.

ref https://github.com/neondatabase/neon/issues/6341

test_runner/fixtures/neon_fixtures.py

@@ -1933,6 +1933,7 @@ class NeonCli(AbstractNeonCli):
         endpoint_id: str,
         tenant_id: Optional[TenantId] = None,
         pageserver_id: Optional[int] = None,
+        safekeepers: Optional[List[int]] = None,
         check_return_code=True,
     ) -> "subprocess.CompletedProcess[str]":
         args = ["endpoint", "reconfigure", endpoint_id]
@@ -1940,6 +1941,8 @@ class NeonCli(AbstractNeonCli):
             args.extend(["--tenant-id", str(tenant_id)])
         if pageserver_id is not None:
             args.extend(["--pageserver-id", str(pageserver_id)])
+        if safekeepers is not None:
+            args.extend(["--safekeepers", (",".join(map(str, safekeepers)))])
         return self.raw_cli(args, check_return_code=check_return_code)
 
     def endpoint_stop(
@@ -3484,6 +3487,7 @@ class Endpoint(PgProtocol, LogUtils):
         self.pg_port = pg_port
         self.http_port = http_port
         self.check_stop_result = check_stop_result
+        # passed to endpoint create and endpoint reconfigure
         self.active_safekeepers: List[int] = list(map(lambda sk: sk.id, env.safekeepers))
         # path to conf is <repo_dir>/endpoints/<endpoint_id>/pgdata/postgresql.conf
 
@@ -3552,6 +3556,7 @@ class Endpoint(PgProtocol, LogUtils):
         self,
         remote_ext_config: Optional[str] = None,
         pageserver_id: Optional[int] = None,
+        safekeepers: Optional[List[int]] = None,
         allow_multiple: bool = False,
     ) -> "Endpoint":
         """
@@ -3561,6 +3566,11 @@ class Endpoint(PgProtocol, LogUtils):
 
         assert self.endpoint_id is not None
 
+        # If `safekeepers` is not None, they are remember them as active and use
+        # in the following commands.
+        if safekeepers is not None:
+            self.active_safekeepers = safekeepers
+
         log.info(f"Starting postgres endpoint {self.endpoint_id}")
 
         self.env.neon_cli.endpoint_start(
@@ -3624,9 +3634,17 @@ class Endpoint(PgProtocol, LogUtils):
     def is_running(self):
         return self._running._value > 0
 
-    def reconfigure(self, pageserver_id: Optional[int] = None):
+    def reconfigure(
+        self, pageserver_id: Optional[int] = None, safekeepers: Optional[List[int]] = None
+    ):
         assert self.endpoint_id is not None
-        self.env.neon_cli.endpoint_reconfigure(self.endpoint_id, self.tenant_id, pageserver_id)
+        # If `safekeepers` is not None, they are remember them as active and use
+        # in the following commands.
+        if safekeepers is not None:
+            self.active_safekeepers = safekeepers
+        self.env.neon_cli.endpoint_reconfigure(
+            self.endpoint_id, self.tenant_id, pageserver_id, self.active_safekeepers
+        )
 
     def respec(self, **kwargs):
         """Update the endpoint.json file used by control_plane."""

test_runner/regress/test_wal_acceptor.py

@@ -1725,7 +1725,10 @@ def test_delete_force(neon_env_builder: NeonEnvBuilder, auth_enabled: bool):
 
 
 # Basic pull_timeline test.
-def test_pull_timeline(neon_env_builder: NeonEnvBuilder):
+# When live_sk_change is False, compute is restarted to change set of
+# safekeepers; otherwise it is live reload.
+@pytest.mark.parametrize("live_sk_change", [False, True])
+def test_pull_timeline(neon_env_builder: NeonEnvBuilder, live_sk_change: bool):
     neon_env_builder.auth_enabled = True
 
     def execute_payload(endpoint: Endpoint):
@@ -1758,8 +1761,7 @@ def test_pull_timeline(neon_env_builder: NeonEnvBuilder):
     log.info("Use only first 3 safekeepers")
     env.safekeepers[3].stop()
     endpoint = env.endpoints.create("main")
-    endpoint.active_safekeepers = [1, 2, 3]
-    endpoint.start()
+    endpoint.start(safekeepers=[1, 2, 3])
 
     execute_payload(endpoint)
     show_statuses(env.safekeepers, tenant_id, timeline_id)
@@ -1771,29 +1773,22 @@ def test_pull_timeline(neon_env_builder: NeonEnvBuilder):
     log.info("Initialize new safekeeper 4, pull data from 1 & 3")
     env.safekeepers[3].start()
 
-    res = (
-        env.safekeepers[3]
-        .http_client(auth_token=env.auth_keys.generate_safekeeper_token())
-        .pull_timeline(
-            {
-                "tenant_id": str(tenant_id),
-                "timeline_id": str(timeline_id),
-                "http_hosts": [
-                    f"http://localhost:{env.safekeepers[0].port.http}",
-                    f"http://localhost:{env.safekeepers[2].port.http}",
-                ],
-            }
-        )
+    res = env.safekeepers[3].pull_timeline(
+        [env.safekeepers[0], env.safekeepers[2]], tenant_id, timeline_id
     )
     log.info("Finished pulling timeline")
     log.info(res)
 
     show_statuses(env.safekeepers, tenant_id, timeline_id)
 
-    log.info("Restarting compute with new config to verify that it works")
-    endpoint.stop_and_destroy().create("main")
-    endpoint.active_safekeepers = [1, 3, 4]
-    endpoint.start()
+    action = "reconfiguing" if live_sk_change else "restarting"
+    log.info(f"{action} compute with new config to verify that it works")
+    new_sks = [1, 3, 4]
+    if not live_sk_change:
+        endpoint.stop_and_destroy().create("main")
+        endpoint.start(safekeepers=new_sks)
+    else:
+        endpoint.reconfigure(safekeepers=new_sks)
 
     execute_payload(endpoint)
     show_statuses(env.safekeepers, tenant_id, timeline_id)