PostgreSQL 11 Open Items
From PostgreSQL wiki
Jump to navigationJump to searchOpen Issues
NOTE: Please add new open items to the bottom of the list.
Decisions to Recheck Mid-Beta
- reconsider jit=on default shortly before release
- Resolved: we'll disable it for v11
- Fixed in: 0fdadfb0192935e0f8bec7fa5ec9dbb470d6d587
Older Bugs
Live issues
- IsInParallelMode() check in set_config_option is wrong (was: WARNING in parallel index creation)
- pg_dump misses public role on schema public
- Default values in partition tables don't work as expected and allow NOT NULL violation
- Breakage in plpgsql composite-variable handling
- test_pg_dump missing cleanup actions
- 9.4 and reproducible "ERROR: could not read block 0 in file..."
Fixed issues
- Allocation done in critical section when initializing WAL
- Fixed in: a86bf6057edf7bc344f1ab6ba7824b561ed9068a
- Parallel query can fail on machines with strict alignment rules
- Fixed in: ca5ca25d08710d7f5aaf0e390aae284bc594a6da
- pg_constraint.conincluding is useless.
- Fixed in: c076f3d74af6e4836206142446bdcc2f93e149d6
- Parallel CREATE INDEX doesn't have relmapper.c state propogated when reindexing mapped catalog relations during VACUUM FULL.
- Fixed in: 4974d7f87e62a58e80c6524e49677cb25cc10e12
- Failed index build can corrupt relcache contents
- Fixed in: b3f919da07540d7c968e8495482336304badcef7
- Fix slot's xmin advancement and subxact's lost snapshots in decoding
- Fixed in: f49a80c481f74fa81
- SHOW ALL does not honor pg_read_all_settings membership
- Fixed in: 0c8910a0cab7c1e439bf5f5850122c36359e6799
- postgres_fdw: Oddity in pushing down inherited UPDATE/DELETE joins to remote servers
- Fixed in: 7fc7dac1a711d0dbd01d2daf6dc97d27d6c6409c
- Continued WAL record can prevent standby from startup
- Fixed in: 0668719801838aa6a8bda330ff9b3d20097ea844
- FinishPreparedTransaction missing HOLD_INTERRUPTS section
- Fixed in: 8f9be261f43772ccee2eae94d971bac6557cbe6a
- PANIC during crash recovery of a recently promoted standby
- Fixed in: 3c64dcb1e35dccbfce531182fa9b9f48bec414ad
- Replication status in logical replication
- Fixed in: 0414ac6a1eb2e457c8019c5a558bd72b37dede89
- LLVM jit and window functions on a temporary table
- LLVM jit and matview
- Fixed in: 3acc4acd9bcbefbfaf789762950726c6208daf1b
- "could not read block" buildfarm failures
- Bug identified, two prototype patches exist, discussion ongoing
- Fixed in: 3acc4acd9bcbefbfaf789762950726c6208daf1b
- Intermittent pg_ctl failures on Windows
- Fixed in: 0ba06e0bfb8cfd24ff17aca92aa72245ddd6c4d7
- Hash index misses call to PageSetChecksumInplace() for empty pages
- Fixed in: 16e7bcfac57fc21dd2124b39132c58971e41448a
- EXPLAIN ANALYZE has wrong timings for JIT when parallelism is in use
- Fixed in: 33001fd7a7072d483272115a9376478fdc007fb9
- Sequence owner not updated when owning table is foreign
- Fixed in: 0320ddaf3c08ea17d616549d0e7f45239592fc76
- Hot standby server with archive_mode=on keeps initial WAL segment
- Fixed in: 78ea8b5daab9237fd42d7a8a836c1c451765499f
- out-of-order XID insertion in KnownAssignedXids
- Fixed in: 1df21ddb19c6e764fc9378c900515a5d642ad820
- Restore CurrentUserId only if 'prevUser' is valid when abort transaction
- Fixed in: b4721f39505b56dd7b556aef5428a0850230ca59
- Fixes for missing schema qualifications
- Fixed in: ee2b37ae044f34851baba69e9ba737077326414e
Non-bugs
- SCRAM with channel binding downgrade attack
- Other thread
- patch exists
- Channel binding has been removed from the list of major features. A more generic solution involving as well downgrades like scram -> md5 need to be designed and thought about.
Resolved Issues
resolved before RC1
- Transaction timestamps and procedures yield inconsistent result
- Segfault when creating partition with a primary key and sql_drop trigger exists
- Fixed in 1a852f7c1e
- Multiple primary keys attaching partition
- Fixed in ff347f8aff
- Assertion failure with ALTER TABLE ATTACH PARTITION with log_min_messages >= DEBUG1
- Fixed in c905b67b79
- pg_upgrade failed with ERROR: null relpartbound for relation 18159 error.
- Fixed in 1c7f585b5c
resolved before 11beta4
- "ERROR: cache lookup failed for constraint" when ALTERING column referenced by fkey of a partitioned table
- Fixed in 6009bad913
- Error during run-time pruning
- Fixed in 1eb9221585
resolved before 11beta3
- Explain buffers wrong counter with parallel plans
- reported as a possible defect in commit 01edb5c7fc3bcf6aea15f2b3be36189b52ad9d1a
- Fixed in dac7fe13bbf10c249a519d6a41c683c42c7e3312
- Advance replication slot patch broke detection of nonexistant output plugins
- Broken in 9c7d06d60680c7f00d931233873dee81fdb311c6
- Fixed in 7326a7d63801673d3569def65cb7955d6486f254
- Server crashed with dense_rank on partition table
- David Rowley reports that the first bad commit is reported 4b9094eb6e14dfdbed61278ea8e51cc846e43579 (Adapt to LLVM 7+ Orc API changes.)
- Amit Langote believes based on the core dump that the problem may be in bf6c614a2f2c58312b3be34a47e7fb7362e07bcb) (Do execGrouping.c via expression eval machinery, take two.)
- Fixed in e60cfcefe635b90bab60e885be689700759a01c6
resolved before 11beta2
- Fast default feature fails in pg_upgrade
- Fixed in: 2448adf29c543befbac59f1ecfbb3ef4b0d808ce with fixup 123efbccea
- pg_replication_slot_advance may cause assertion failures with incorrect LSN values
- Fix committed in f731cfa94c00164814625d5753d376a4a7c43fff
- pg_config.h.win32 missing a set of flags from pg_config.h.in added in v11 development
- Fixed in: bde64eb6107622e8438dd61b93afd4d1adf178b3
- Server crash during parallel append path generation
- Fixed in: 403318b71f7058ecbfb65bcc7de1eec96cd35d3f
- Performance regression with PostgreSQL 11 and partitioning
- Fixed in: 7d872c91a3f9d49b56117557cdbb0c3d4c620687
resolved before 11beta1
- Indexes on partitioned tables and foreign partitions
- appears to be a bug in 8b08f7d4 (Local partitioned indexes)
- Fixed in: 4eaa53727542c39cca71b80e8ff3e1f742d64452
- Memory leaks with _SPI_stack handling in TopMemoryContext
- Regression caused by commit 8561e48.
- Fixed in: 30c66e77be1d890c3cca766259c0bec80bcac1b5
- Corrupted btree index on HEAD because of covering indexes
- Fixed in: 6db4b49986be3fe59a1f6ba6fabf9852864efc3e
- Decide if we want a GUC to disable partition pruning
- Fixed in: 055fb8d33da6ff9003e3da4b9944bdcd2e2b2a49
- expand_tuple segfaults
- coverage report shows it's completely untested, too
- Bug fix: 7c44c46deb495a2f3861f402d7f2109263e3d50a
- Add coverage: b39fd897e0398a6bdc6552daa7cacdf9c0e46d7e
- Failed assertion on pfree() via perform_pruning_combine_step
- Original commit: 9fdb675fc5d2de825414e05939727de8b120ae81
- Bug fix: 7ba6ee815dc90d4fab7226d343bf72aa28c9aa5c
- Parallel index workers don't have activity set
- Original commit: 9da0cc35284bdbe8d442d732963303ff0e0a40bc
- Bug fix: 7de4a1bcc56f494acbd0d6e70781df877dc8ecb5
- check_ssl_key_file_permissions should be in be-secure-common.c
- Original commit: 8a3d9425290ff5f6434990349886afae9e1c6008
- Patch exists
- Bug fix: 2764d5dcfa84d240c901c20ec6e194f72d82b78a
- Missing break statement after transformCallStmt in transformStmt
- Original commit: 76b6aa41f41db66004b1c430f17a546d4102fbe7
- Bug fix: 13c7c65ec900a30bcddcb27f5fd138dcdbc2ca2e
- Parallel index creation does not properly clean up after error
- Original commit: 29d58fd3adae9057c3fd502393b2f131bc96eaf9
- Bug fix: 47cb9ca49a611fa518e1a0fe46526507c96a5612
- wrong memory context used in FmgrInfo's contained in PartitionKey
- Bug fix (HEAD): a4d56f583e7cff052c2699e62d867ae1c8fda4f3
- Bug fix (PG 10): 5f11c6ec61a579d60347a5d13af7e42b17fadc56
Important Dates
Current schedule:
- feature freeze: 8th of April 2018
- beta1: 24th of May 2018
- beta2: 28th of June 2018
- beta3: 9th of August 2018
- beta4: 20th of September 2018
- rc1: 11th of October 2018
- ga: 18th of October 2018