release-6548

What's Changed

• safekeeper: reorder routes and their handlers. by @arssher in #8831
• Fix the pg_hintplan flakyness by @a-masterov in #8834
• proxy: remove dead code by @conradludgate in #8847
• proxy: Rename backend types and variants as prep for refactor by @cloneable in #8845
• fix(walproposer): Do not restart on safekeepers reordering by @ololobus in #8840
• Remove support for pageserver <-> compute protocol version 1 by @hlinnaka in #8774
• tests: Remove unnecessary entries from list of allowed errors by @hlinnaka in #8199
• .github: direct feature requests to the feedback form by @jcsp in #8849
• bench: flush before shutting down by @koivunej in #8844
• docs: rolling storage controller restarts RFC by @VladLazar in #8310
• pageserver: do vectored read on each dio-aligned section once by @yliang412 in #8763
• storcon: track pageserver availability zone by @VladLazar in #8852
• fix(storage-scrubber): make retry error into warnings by @skyzh in #8851
• bypass PageCache for InMemoryLayer + avoid Value::deser on L0 flush by @problame in #8537
• feat(compute_ctl): add periodic lease lsn requests for static computes by @prepor in #7994
• Undo walloging replorgin file on checkpoint by @knizhnik in #8794
• ingest: rate-limited warning if WAL commit timestamps lags for > wait_lsn_timeout by @problame in #8839
• proxy+pageserver: shared leaky bucket impl by @conradludgate in #8539
• pageserver: default to 4MiB stack size and add env var to control it by @problame in #8862
• Don't unarchive a timeline if its ancestor is archived by @arpad-m in #8853
• tenant background loops: periodic log message if long-running iteration by @problame in #8832
• proxy: improve test performance by @conradludgate in #8863
• fix(pageserver): add dry-run to force compact API by @skyzh in #8859
• fix(pageserver): reject non-i128 key on the write path by @skyzh in #8648
• Remove async_trait usages from safekeeper and neon_local by @arpad-m in #8864
• proxy: fix password hash cancellation by @conradludgate in #8868
• chore: speed up testing by @conradludgate in #8874
• storage controller: sleep between compute notify retries by @jcsp in #8869
• CI(promote-compatibility-data): fix job by @bayandin in #8871
• feat(storage-scrubber): check layer map valid by @skyzh in #8867
• pageserver: set default io_buffer_alignment to 512 bytes by @yliang412 in #8878
• fix(pageserver): fire layer eviction alert only when it's visible by @skyzh in #8882
• Add safekeeper scrubber test by @arpad-m in #8785
• Storage & Compute release 2024-09-02 by @vipvap in #8883

Full Changelog: release-6514...release-6548

release-6514

What's Changed

• feat(compute): Introduce new compute_subscriptions_count metric by @ololobus in #8796
• proxy: clippy lints: handle some low hanging fruit by @cloneable in #8829
• Timeline archival test by @arpad-m in #8824
• pageserver: startup: ensure local disk state is durable by @problame in #8835
• feat(pageserver): use split layer writer in gc-compaction by @skyzh in #8608
• fix(sql-exporter): Fix typo in gauge metric type by @ololobus in #8842
• Compute release 2024-08-28 by @lubennikovaav in #8858

Full Changelog: release-6506...release-6514

release-6506

What's Changed

• proxy: Fix some warnings by extended clippy checks by @cloneable in #8748
• l0_flush: remove support for mode page-cached by @problame in #8739
• Migrate physical GC and scan_metadata to remote_storage by @arpad-m in #8673
• Update aws -> infra for repo rename by @arpad-m in #8755
• scrubber: remove _generic postfix and two unused functions by @arpad-m in #8761
• safekeeper: check for non-consecutive writes in safekeeper.rs by @arssher in #8640
• task_mgr::spawn: require a TenantId by @problame in #8462
• Make Postgres 16 default version by @bayandin in #8745
• Revert "safekeeper: check for non-consecutive writes in safekeeper.rs" by @jcsp in #8771
• storcon: run db migrations after step down sequence by @VladLazar in #8756
• proxy: make TLS and WS handling optional by @conradludgate in #8752
• storcon: add peer jwt token by @VladLazar in #8764
• pageserver: add metric pageserver_secondary_heatmap_total_size by @jcsp in #8768
• impr(pageserver): abort on fatal I/O writer error by @skyzh in #8777
• Benchmarking: need to checkout actions to download Neon artifacts by @Bodobolero in #8770
• pageserver: add counter for wait time on background loop semaphore by @problame in #8769
• storcon: handle heartbeater shutdown gracefully by @koivunej in #8767
• test: test_timeline_ancestor_detach_errors rare allowed_error by @koivunej in #8782
• CI(build-and-test): run regression tests on arm by @bayandin in #8552
• remove workspace hack from libs by @conradludgate in #8780
• Add compute_logical_snapshot_files metric by @tristan957 in #8762
• safekeeper: add listing timelines by @arssher in #8789
• test: fix more flaky due to graceful shutdown by @koivunej in #8787
• Add compute_receive_lsn metric by @tristan957 in #8750
• test: avoid some too long shutdowns by flushing before shutdown by @koivunej in #8772
• fix(pageserver): unify initdb optimization for sparse keyspaces; fix force img generation by @skyzh in #8776
• pageserver: batch InMemoryLayer puts, remove need to sort items by LSN during ingest by @jcsp in #8591
• Sleep in test_scrubber_physical_gc by @arpad-m in #8798
• Revert "Use sycnhronous commit for logical replicaiton worker (#8645)" by @ololobus in #8792
• proxy: Enable stricter/pedantic clippy checks by @cloneable in #8775
• test_compatibility: flush in the end by @koivunej in #8804
• Fix race condition in LRU list update in get_cached_relsize by @knizhnik in #8807
• feat(pageserver): warn on aux v1 tenants + default to v2 by @skyzh in #8625
• Upgrade Postgres to 14.13, 15.8, and 16.4 by @tristan957 in #8704
• Document how to use "git merge" for PostgreSQL minor version upgrades. by @hlinnaka in #8692
• Require poetry >=1.8 by @MMeent in #8812
• Update serde_json to 1.0.125 by @arpad-m in #8813
• CI(autocomment): add arch to build type by @bayandin in #8809
• feat(pageserver): remove features=testing restriction for compact by @skyzh in #8815
• fix(pageserver): l0 check criteria by @skyzh in #8797
• test: copy dict to avoid error on retry by @koivunej in #8811
• storcon_cli: add support for drain and fill operations by @VladLazar in #8791
• update diesel by @conradludgate in #8816
• impr(pageserver): error if keys are unordered in merge iter by @skyzh in #8818
• proxy: flesh out JWT verification code by @conradludgate in #8805
• controller: use PageserverUtilization for scheduling by @jcsp in #8711
• storage controller: enable timeline CRUD operations to run concurrently with reconciliation & make them safer by @jcsp in #8783
• proxy: local auth backend by @conradludgate in #8806
• proxy: new local-proxy application by @conradludgate in #8736
• chore: improve markdown formatting by @mrl5 in #8825
• Storage & Compute release 2024-08-26 by @vipvap in #8827

New Contributors

• @cloneable made their first contribution in #8748
• @mrl5 made their first contribution in #8825

Full Changelog: release-6450...release-6506

release-6450

What's Changed

• Adopt list_streaming in tenant deletion by @arpad-m in #8504
• Add a test for clickhouse as a logical replication consumer by @a-masterov in #8408
• test: deflake test_duplicate_creation by @koivunej in #8536
• pageserver: fix return code from secondary_download_handler by @jcsp in #8508
• pageserver: fail if id is present in pageserver.toml by @problame in #8489
• l0_flush: use mode=direct by default => coverage in automated tests by @problame in #8534
• scrubber: enable cleaning up garbage tenants from known deletion bugs, add object age safety check by @jcsp in #8461
• Add metrics for input data considered and taken for compression by @arpad-m in #8522
• scrubber: add remote_storage based listing APIs and use them in find-large-objects by @arpad-m in #8541
• CI(benchmarking): make neonvm default provisioner by @bayandin in #8538
• [neon/acr] impr: push to ACR while building images by @antonyc in #8545
• feat(storcon): store scrubber metadata scan result by @yliang412 in #8480
• feat(scrubber): post scan_metadata results to storage controller by @yliang412 in #8502
• cleanup(compact_level0_phase1): some commentary and wrapping into block expressions by @problame in #8544
• cicd: change Azure storage details [1/2] by @fcdm in #8553
• compaction_level0_phase1: bypass PS PageCache for data blocks by @problame in #8543
• CI(regress-tests): run less regression tests by @bayandin in #8561
• feat(pageserver): support btm-gc-compaction for child branches by @skyzh in #8519
• refactor(page_service): Timeline gate guard holding + cancellation + shutdown by @problame in #8339
• Fix negative replication delay metric by @tristan957 in #8520
• cicd: change Azure storage details [2/2] by @fcdm in #8562
• controller: simplify reconciler generation increment logic by @jcsp in #8560
• storage_scrubber: migrate FindGarbage to remote_storage by @arpad-m in #8548
• CI(create-test-report): fix missing benchmark results in Allure report by @bayandin in #8540
• test(pageserver): add test_gc_feedback_with_snapshots by @skyzh in #8474
• pageserver: shutdown all walredo managers 8s into shutdown by @problame in #8572
• Reduce linux-raw-sys duplication by @arpad-m in #8577
• pageserver: add layer visibility calculation by @jcsp in #8511
• storage-scrubber: log version on start by @skyzh in #8571
• fix(pageserver): skip existing layers for btm-gc-compaction by @skyzh in #8498
• test_runner: don't create artifacts if Allure is not enabled by @bayandin in #8580
• pageserver: refine how we delete timelines after shard split by @jcsp in #8436
• storage_controller: start adding chaos hooks by @jcsp in #7946
• Wait for completion of the upload queue in flush_frozen_layer by @arpad-m in #8550
• Improve safekeepers eviction rate limiting by @petuhovskiy in #8456
• tests: add test_historic_storage_formats by @jcsp in #8423
• fix(pageserver): deadlock in gc-compaction by @skyzh in #8590
• feat(pageserver): support auto split layers based on size by @skyzh in #8574
• safekeeper: remove unused partial_backup_enabled option by @dotdister in #8547
• feat(pageserver): support split delta layers by @skyzh in #8599
• Update Postgres versions to use smgrexists() instead of access() to check if Oid is used by @knizhnik in #8597
• CI(trigger-e2e-tests): wait for promote-images job from the last commit by @bayandin in #8592
• CI(trigger-e2e-tests): fix deadlock with Build and Test workflow by @bayandin in #8606
• fix: allow awaiting logical size for root timelines by @koivunej in #8604
• CI(benchmarking): set pub/sub projects for LR tests by @bayandin in #8483
• feat(pageserver): support dry-run for gc-compaction, add statistics by @skyzh in #8557
• fix: make Timeline::set_disk_consistent_lsn use fetch_max by @koivunej in #8311
• feat: persistent gc blocking by @koivunej in #8600
• pageserver: remove legacy read path by @VladLazar in #8601
• chore: bump index part version by @koivunej in #8611
• storage_scrubber: migrate scan_safekeeper_metadata to remote_storage by @arpad-m in #8595
• Add package-mode=false to poetry. by @arssher in #8609
• Add a test using Debezium as a client for the logical replication by @a-masterov in #8568
• enable rum test by @a-masterov in #8380
• tests: improve stability of test_storage_controller_many_tenants by @jcsp in #8607
• pageserver: fixes for layer visibility metric by @jcsp in #8603
• pageserver: use layer visibility when composing heatmap by @jcsp in #8616
• pageserver: add bench_ingest by @jcsp in #7409
• scrubber: clean up scan_metadata before prod by @yliang412 in #8565
• storage broker: only print one line for version and build tag in init by @arpad-m in #8624
• Lower level for timeline cancellations during gc by @arpad-m in #8626
• Add missing colon to ArchivalConfigRequest specification by @arpad-m in #8627
• proxy: random changes by @conradludgate in #8602
• fix: EphemeralFiles can outlive their Timeline via enum LayerManager by @koivunej in #8229
• fix(pageserver): dump the key when it's invalid by @skyzh in #8633
• refactor(timeline_detach_ancestor): replace ordered reparented with a hashset by @koivunej in #8629
• cicd: push build-tools image to ACR as well by @fcdm in #8638
• fix: drain completed page_service connections by @koivunej in #8632
• feat(pageserver): add direct io pageserver config by @yliang412 in #8622
• pageserver: make bench_ingest build (but panic) on macOS by @jcsp in #8641
• Use sycnhronous commit for logical replicaiton worker by @knizhnik in #8645
• fix(test): do not fail test for filesystem race by @koivunej in #8643
• fix: stop leaking BackgroundPurges by @koivunej in #8650
• pageserver: remove paranoia double-calculation of retain_lsns by @jcsp in #8617
• tests: don't require kafka client for regular tests by @jcsp in #8662
• CI(build-tools): update Rust, Python, Mold by @bayandin in #8667
• Dockerfiles: fix LegacyKeyValueFormat & JSONArgsRecommended by @bayandin in #8664
• Run a subset of benchmarking job steps on GitHub action runners in Azure - closer to the system under test by @Bodobolero in #8651
• Revert "proxy: update tokio-postgres to allow arbitrary config params (#8076)" by @conradludgate in #8654
• fix(neon): disable create tablespace stmt by @skyzh in #8657
• CI(pin-build-tools-image): fix permissions for Azure login by @bayandin in #8671
• pageserver: don't treat NotInitialized::Stopped as unexpected by @jcsp in #8675
• storcon: skip draining shard if it's secondary is lagging too much by @VladLazar in #8644
• Dockerfiles: remove cachepot by @bayandin in #86...

release-6299

What's Changed

• Storage & Compute release 2024-08-12 by @vipvap in #8688

Full Changelog: release-6250...release-6299

release-proxy-6107

Merge pull request #8647 from neondatabase/rc/proxy/2024-08-08

Proxy release 2024-08-08

release-6250

What's Changed

• Storage release 2024-08-07 by @arpad-m in #8642

Full Changelog: release-6248...release-6250

release-6248

What's Changed

• Storage & Compute release 2024-08-05 by @vipvap in #8598

Full Changelog: release-6210...release-6248

release-6210

What's Changed

• [proxy/redis] impr: use redis_auth_type to switch between auth types by @antonyc in #8428
• fix(docs): race on monotonic rfc id by @skyzh in #8445
• tests: more generous memory allowance in test_compaction_l0_memory by @jcsp in #8446
• pageserver: remove absolute-order disk usage eviction by @jcsp in #8454
• tests: make test_change_pageserver more robust by @jcsp in #8442
• Start using remote_storage in S3 scrubber for PurgeGarbage by @arpad-m in #7932
• fix(scrubber): more robust metadata consistency checks by @yliang412 in #8344
• fix(docs): clearify the admin URL and token used in scrubber by @skyzh in #8441
• refactor(pageserver) remove task_mgr for most global tasks by @problame in #8449
• build: mark target/ and pg_install/ with CACHEDIR.TAG by @problame in #8448
• Mark body of archival_config endpoint as required by @arpad-m in #8458
• Change default version of Neon extensio to 1.4 by @knizhnik in #8405
• vm-image: Expose new LFC working set size metrics by @sharnoff in #8298
• fix(docs): remove incorrect flags for scrubber purge-garbage command by @yliang412 in #8463
• pageserver: use identity file as node id authority and remove init command and config-override flags by @VladLazar in #7766
• tests: make test_scrubber_physical_gc_ancestors more stable by @jcsp in #8453
• build(deps): bump openssl from 0.10.64 to 0.10.66 in /test_runner/pg_clients/rust/tokio-postgres by @dependabot in #8464
• pageserver: add LayerVisibilityHint by @jcsp in #8432
• Use smgrexists() instead of access() to enforce uniqueness of generated relfilenumber by @knizhnik in #7992
• Fix to use a tab instead of spaces by @shinyaaa in #8394
• tests: allow-list a controller heartbeat error by @jcsp in #8471
• temporarily disable cache saving in the registry as it is very slow by @cicdteam in #8475
• feat(pageserver): support retain_lsn in bottommost gc-compaction by @skyzh in #8328
• remote_storage: add list_streaming API call by @arpad-m in #8466
• followup(#8475): also disable 'cache-to' for neon-image-arch and neon-test-extensions by @problame in #8478
• Make async connect work on MacOS: it is necessary top call WaitLatchOrSocket before PQconnectPoll by @knizhnik in #8472
• pageserver: simplify LayerAccessStats by @jcsp in #8431
• Implement retry support for list_streaming by @arpad-m in #8481
• tests: simplify code around unstable test_basebackup_with_high_slru_count by @jcsp in #8477
• pageserver: maintain gc_info incrementally by @jcsp in #8427
• proxy: switch to leaky bucket by @conradludgate in #8470
• CI(build-and-test): move part of the pipeline to a reusable workflow by @bayandin in #8241
• pageserver: faster and simpler inmem layer vec read by @VladLazar in #8469
• pageserver: fix a compilation error by @jcsp in #8487
• devx: nicer diff hunk headers by @koivunej in #8482
• CI: Run ARM checks in the main pipeline by @bayandin in #8185
• followup(#8359): pre-initialize circuitbreaker metrics by @problame in #8491
• fix: remote timeline client shutdown trips circuit breaker by @problame in #8495
• Temporarily remove week-end test for res-aurora from pgbench-compare benchmarking runs by @Bodobolero in #8493
• storcon: make heartbeats restart aware by @VladLazar in #8222
• tests: always scrub on test exit when using S3Storage by @jcsp in #8437
• tests: adjust threshold in test_partial_evict_tenant by @jcsp in #8509
• storcon: shutdown with clean observed state by @VladLazar in #8494
• fix(pageserver): allow incomplete history in btm-gc-compaction by @skyzh in #8500
• Update pgrx to v 0.11.3 by @lubennikovaav in #8515
• Update links in markdown. by @MihaiBojin in #8501
• Handle NotInitialized::ShuttingDown error in shard split by @arpad-m in #8506
• Using own registry to cache layers in docker build workflows by @cicdteam in #8521
• remote_storage: expose last_modified in listings by @jcsp in #8497
• Update Rust to 1.80.0 by @arpad-m in #8518
• tests: checkpoint instead of compact in test_sharding_split_compaction by @jcsp in #8473
• .github: reduce test concurrency by @jcsp in #8444
• pageserver: make vectored get the default read path for the pageserver by @VladLazar in #8384
• CODEOWNERS: collapse safekeepers into storage by @jcsp in #8510
• storcon: introduce step down primitive by @VladLazar in #8512
• Fix sql-exporter-autoscaling for pg < 16 by @sharnoff in #8523
• refactor(layer load API): all errors are permanent by @problame in #8527
• raise wait_lsn timeout from 60s to 300s by @problame in #8529
• fix(pageserver): give L0 compaction priorities over image layer creation by @skyzh in #8443
• Persist archival information by @arpad-m in #8479
• CI(test_runner): Upload all test artifacts if preserve_database_files is enabled by @bayandin in #7990
• safekeeper: include application name in wal service root span by @problame in #8525
• test_runner: fix user_property usage in benchmarks by @bayandin in #8531
• Storage & Compute release 2024-07-29 by @vipvap in #8533

New Contributors

• @MihaiBojin made their first contribution in #8501

Full Changelog: release-6144...release-6210

release-6144

What's Changed

• Compute release 2024-06-26 by @kelvich in #8530

Full Changelog: release-6142...release-6144