tests: allow-lists for occasional failures (#6074)

test_creating_tenant_conf_after...
- Test detaches a tenant and then re-attaches immediatel: this causes a
race between pending remote LSN update and the generation bump in the
attachment.

test_gc_cutoff:
- Test rapidly restarts a pageserver before one generation has had the
chance to process deletions from the previous generation

test_runner/regress/test_gc_cutoff.py

@@ -35,6 +35,11 @@ def test_gc_cutoff(neon_env_builder: NeonEnvBuilder, pg_bin: PgBin):
 
     pageserver_http.configure_failpoints(("after-timeline-gc-removed-layers", "exit"))
 
+    # Because this test does a rapid series of restarts of the same node, it's possible that
+    # we are restarted again before we can clean up deletion lists form the previous generation,
+    # resulting in a subsequent startup logging a warning.
+    env.pageserver.allowed_errors.append(".*Dropping stale deletions for tenant.*")
+
     for _ in range(5):
         with pytest.raises(subprocess.SubprocessError):
             pg_bin.run_capture(["pgbench", "-P1", "-N", "-c5", "-T500", "-Mprepared", connstr])

test_runner/regress/test_tenant_conf.py

@@ -314,6 +314,10 @@ def test_creating_tenant_conf_after_attach(neon_env_builder: NeonEnvBuilder):
 
     assert not config_path.exists(), "detach did not remove config file"
 
+    # The re-attach's increment of the generation number may invalidate deletion queue
+    # updates in flight from the previous attachment.
+    env.pageserver.allowed_errors.append(".*Dropped remote consistent LSN updates.*")
+
     env.pageserver.tenant_attach(tenant_id)
     wait_until(
         number_of_iterations=5,