John Spray 7eaa7a496b pageserver: cancellation handling in writes to postgres client socket (#5503) ...

## Problem

Writes to the postgres client socket from the page server were not
wrapped in cancellation handling, so a stuck client connection could
prevent tenant shutdowwn.

## Summary of changes

All the places we call flush() to write to the socket, we should be
respecting the cancellation token for the task.

In this PR, I explicitly pass around a CancellationToken rather than
doing inline `task_mgr::shutdown_token` calls, to avoid coupling it to
the global task_mgr state and make it easier to refactor later.

I have some follow-on commits that add a Shutdown variant to QueryError
and use it more extensively, but that's pure refactor so will keep
separate from this bug fix PR.

Closes: https://github.com/neondatabase/neon/issues/5341

2023-10-09 15:54:17 +01:00

..

data/extension_test/5670669815

Feat/postgres 16 (#4761)

2023-09-12 15:11:32 +02:00

test_ancestor_branch.py

make flush frozen layer an atomic operation (#4720)

2023-07-20 13:39:19 -04:00

test_attach_tenant_config.py

test: Remote storage refactorings (#5243)

2023-09-08 13:54:23 +03:00

test_auth.py

python: more linting (#4734)

2023-07-18 12:56:40 +03:00

test_backpressure.py

python: more linting (#4734)

2023-07-18 12:56:40 +03:00

test_basebackup_error.py

Rename "Postgres nodes" in control_plane to endpoints.

2023-04-13 14:34:29 +03:00

test_branch_and_gc.py

timeline creation: reflect failures due to ancestor LSN issues in status code (#4600)

2023-07-03 15:21:10 +03:00

test_branch_behind.py

timeline creation: reflect failures due to ancestor LSN issues in status code (#4600)

2023-07-03 15:21:10 +03:00

test_branching.py

fix: deny branching, starting compute from not yet uploaded timelines (#5484)

2023-10-09 17:03:38 +03:00

test_broken_timeline.py

tests: Default to LOCAL_FS for pageserver remote storage (#5402)

2023-09-28 12:25:20 +03:00

test_build_info_metric.py

python: more linting (#4734)

2023-07-18 12:56:40 +03:00

test_clog_truncate.py

Rename "Postgres nodes" in control_plane to endpoints.

2023-04-13 14:34:29 +03:00

test_close_fds.py

tests: enable multiple pageservers in neon_local and neon_fixture (#5231)

2023-09-08 16:19:57 +01:00

test_compatibility.py

Enable compatibility tests on Postgres 16 (#5314)

2023-09-18 12:58:34 +01:00

test_config.py

Rename "Postgres nodes" in control_plane to endpoints.

2023-04-13 14:34:29 +03:00

test_crafted_wal_end.py

Rename "Postgres nodes" in control_plane to endpoints.

2023-04-13 14:34:29 +03:00

test_createdropdb.py

Add a test case for "CREATE DATABASE STRATEGY=file_copy". (#5301)

2023-09-15 16:50:57 +03:00

test_createuser.py

Rename "Postgres nodes" in control_plane to endpoints.

2023-04-13 14:34:29 +03:00

test_ddl_forwarding.py

Fix invalid database resulting from failed DROP DB (#5423)

2023-09-29 19:39:28 +01:00

test_disk_usage_eviction.py

disk_usage_based_eviction: switch warmup to use full table scans (#5384)

2023-09-27 10:00:21 +01:00

test_download_extensions.py

Feat/postgres 16 (#4761)

2023-09-12 15:11:32 +02:00

test_duplicate_layers.py

pageserver: Tune 503 Resource unavailable (#5489)

2023-10-06 09:59:14 +01:00

test_fsm_truncate.py

Rename "Postgres nodes" in control_plane to endpoints.

2023-04-13 14:34:29 +03:00

test_fullbackup.py

tests: split neon_fixtures.py (#4871)

2023-08-03 17:20:24 +03:00

test_gc_aggressive.py

refactor: globals in tests (#5298)

2023-09-13 22:05:30 +03:00

test_gc_cutoff.py

refactor: failpoint support (#5033)

2023-08-19 01:01:44 +03:00

test_hot_standby.py

test_runner: remove excessive timeouts (#4659)

2023-08-09 16:27:53 +01:00

test_import.py

Move tenant & timeline dir method to NeonPageserver and use them everywhere (#5262)

2023-09-15 11:17:18 +01:00

test_large_schema.py

tests: enable multiple pageservers in neon_local and neon_fixture (#5231)

2023-09-08 16:19:57 +01:00

test_layer_eviction.py

Move tenant & timeline dir method to NeonPageserver and use them everywhere (#5262)

2023-09-15 11:17:18 +01:00

test_layer_writers_fail.py

Move tenant & timeline dir method to NeonPageserver and use them everywhere (#5262)

2023-09-15 11:17:18 +01:00

test_lfc_resize.py

Do not handle lfc_change_limit in processes not haing PGPROC structure (#5332)

2023-09-19 21:55:36 +03:00

test_logging.py

add libmetric metric for each logged log message (#4055)

2023-04-25 14:10:18 +02:00

test_lsn_mapping.py

Rename "Postgres nodes" in control_plane to endpoints.

2023-04-13 14:34:29 +03:00

test_multixact.py

Mx offset bug (#4775)

2023-07-21 21:20:53 +03:00

test_neon_cli.py

tests: enable multiple pageservers in neon_local and neon_fixture (#5231)

2023-09-08 16:19:57 +01:00

test_neon_local_cli.py

tests: split neon_fixtures.py (#4871)

2023-08-03 17:20:24 +03:00

test_next_xid.py

Rename "Postgres nodes" in control_plane to endpoints.

2023-04-13 14:34:29 +03:00

test_normal_work.py

Rename "Postgres nodes" in control_plane to endpoints.

2023-04-13 14:34:29 +03:00

test_old_request_lsn.py

python: more linting (#4734)

2023-07-18 12:56:40 +03:00

test_ondemand_download.py

test: test_download_remote_layers_api again (#5322)

2023-09-16 18:27:19 +03:00

test_pageserver_api.py

tests: enable multiple pageservers in neon_local and neon_fixture (#5231)

2023-09-08 16:19:57 +01:00

test_pageserver_catchup.py

Rename "Postgres nodes" in control_plane to endpoints.

2023-04-13 14:34:29 +03:00

test_pageserver_generations.py

pageserver: implement emergency mode for operating without control plane (#5469)

2023-10-06 17:25:21 +01:00

test_pageserver_metric_collection.py

tests: Default to LOCAL_FS for pageserver remote storage (#5402)

2023-09-28 12:25:20 +03:00

test_pageserver_restart.py

test: duplicate L1 layer (#5412)

2023-10-04 16:52:32 +01:00

test_pageserver_restarts_under_workload.py

pageserver: cancellation handling in writes to postgres client socket (#5503)

2023-10-09 15:54:17 +01:00

test_parallel_copy.py

python: more linting (#4734)

2023-07-18 12:56:40 +03:00

test_pg_regress.py

test: make pg_regress less flaky, hopefully (#4903)

2023-08-10 15:24:43 +03:00

test_pitr_gc.py

Rename "Postgres nodes" in control_plane to endpoints.

2023-04-13 14:34:29 +03:00

test_proxy_metric_collection.py

refactor(test_consumption_metrics): split for pageserver and proxy (#5324)

2023-09-16 18:05:35 +03:00

test_proxy.py

proxy http: reproduce issue with transactions in pool (#5293)

2023-09-12 17:13:25 -04:00

test_read_trace.py

tests: enable multiple pageservers in neon_local and neon_fixture (#5231)

2023-09-08 16:19:57 +01:00

test_read_validation.py

python: more linting (#4734)

2023-07-18 12:56:40 +03:00

test_readonly_node.py

Rename "Postgres nodes" in control_plane to endpoints.

2023-04-13 14:34:29 +03:00

test_recovery.py

fix: demote warnings, fix flakyness (#4837)

2023-07-31 07:43:12 +00:00

test_remote_storage.py

Move tenant & timeline dir method to NeonPageserver and use them everywhere (#5262)

2023-09-15 11:17:18 +01:00

test_setup.py

python: more linting (#4734)

2023-07-18 12:56:40 +03:00

test_sni_router.py

tests: split neon_fixtures.py (#4871)

2023-08-03 17:20:24 +03:00

test_subxacts.py

test: make pg_regress less flaky, hopefully (#4903)

2023-08-10 15:24:43 +03:00

test_tenant_conf.py

Move tenant & timeline dir method to NeonPageserver and use them everywhere (#5262)

2023-09-15 11:17:18 +01:00

test_tenant_delete.py

test_runner: allow race in test_tenant_delete_is_resumed_on_attach (#5478)

2023-10-06 09:49:31 +01:00

test_tenant_detach.py

pageserver: cancellation handling in writes to postgres client socket (#5503)

2023-10-09 15:54:17 +01:00

test_tenant_relocation.py

Move tenant & timeline dir method to NeonPageserver and use them everywhere (#5262)

2023-09-15 11:17:18 +01:00

test_tenant_size.py

tests: use approximate equality in test_get_tenant_size_with_multiple_branches (#5411)

2023-09-29 09:15:43 +01:00

test_tenant_tasks.py

propagate lock guard to background deletion task (#4495)

2023-06-15 17:30:12 +03:00

test_tenants_with_remote_storage.py

Move tenant & timeline dir method to NeonPageserver and use them everywhere (#5262)

2023-09-15 11:17:18 +01:00

test_tenants.py

tests: Default to LOCAL_FS for pageserver remote storage (#5402)

2023-09-28 12:25:20 +03:00

test_threshold_based_eviction.py

fix(consumption_metrics): exp backoff retry (#5317)

2023-09-16 01:11:01 +03:00

test_timeline_delete.py

pageserver: prefer 503 to 500 for transient unavailability (#5439)

2023-10-03 17:00:55 +01:00

test_timeline_size.py

tests: Default to LOCAL_FS for pageserver remote storage (#5402)

2023-09-28 12:25:20 +03:00

test_truncate.py

python: more linting (#4734)

2023-07-18 12:56:40 +03:00

test_twophase.py

Rename "Postgres nodes" in control_plane to endpoints.

2023-04-13 14:34:29 +03:00

test_unlogged.py

Rename "Postgres nodes" in control_plane to endpoints.

2023-04-13 14:34:29 +03:00

test_vm_bits.py

Handle update of VM in XLOG_HEAP_LOCK/XLOG_HEAP2_LOCK_UPDATED WAL records (#4896)

2023-09-15 17:47:29 +03:00

test_wal_acceptor_async.py

tests: enable higher concurrency and adjust tests with outlier runtime (#4904)

2023-08-08 09:16:21 +01:00

test_wal_acceptor.py

Wait for sk tli init in test_timeline_status (#5467)

2023-10-04 22:53:34 +01:00

test_wal_receiver.py

Raise pageserver walreceiver timeouts.

2023-06-19 15:59:38 +04:00

test_wal_restore.py

tests: enable multiple pageservers in neon_local and neon_fixture (#5231)

2023-09-08 16:19:57 +01:00

test_walredo_not_left_behind_on_detach.py

Move tenant & timeline dir method to NeonPageserver and use them everywhere (#5262)

2023-09-15 11:17:18 +01:00