For ANY-SUBLINK, we adopted a two-stage pull-up approach to handle
different types of scenarios. In the first stage, the sublink is pulled up
as a subquery. Because of this, when writing this code, we did not have
the ability to perform lateral joins, and therefore, we were unable to
pull up Var with varlevelsup=1. Now that we have the ability to use
lateral joins, we can eliminate this limitation.
Author: Andy Fan <zhihui.fan1213@gmail.com>
Author: Tom Lane <tgl@sss.pgh.pa.us>
Reviewed-by: Tom Lane <tgl@sss.pgh.pa.us>
Reviewed-by: Richard Guo <guofenglinux@gmail.com>
Reviewed-by: Alena Rybakina <lena.ribackina@yandex.ru>
Reviewed-by: Andrey Lepikhov <a.lepikhov@postgrespro.ru>
pathkeys_useful_for_ordering() contained some needless checks to return
0 when either root->query_pathkeys or pathkeys lists were empty. This is
already handled by pathkeys_count_contained_in(), so let's have it do the
work instead of having redundant checks.
Similarly, in pathkeys_useful_for_grouping(), checking pathkeys is an
empty list just before looping over it isn't required. Technically,
neither is the list empty check for group_pathkeys, but I felt a bit
more work would have to be done to get the equivalent behavior if we'd
left it up to the foreach loop to call list_member_ptr().
This was noticed by Andy while he was reviewing a patch to improve the
UNION planner. Since that patch adds another function similar to
pathkeys_useful_for_ordering() and since I wasn't planning to copy these
redundant checks over to the new function, let's adjust the existing
code so that both functions will be consistent.
Author: Andy Fan
Discussion: https://postgr.es/m/87o7cti48f.fsf@163.com
This is extracted from a larger patch to improve the UNION planner.
While working on that, I found myself having to check what the 'rows'
parameter is for. It's not obvious that passing a negative number is the
way to have the rows estimate calculated and to find that out you need
to read code in create_append_path() and in cost_append().
Discussion: https://postgr.es/m/CAApHDvpb_63XQodmxKUF8vb9M7CxyUyT4sWvEgqeQU-GB7QFoQ@mail.gmail.com
Thanks to commit 3b00fdba9f, this check in the SIGTERM handler for
the startup process is now obsolete and can be removed. Instead of
leaving around the dead function write_stderr_signal_safe(), I've
opted to just remove it for now.
This partially reverts commit 97550c0711.
Reviewed-by: Andres Freund, Noah Misch
Discussion: https://postgr.es/m/20231121212008.GA3742740%40nathanxps13
Presently, we rely on each individual signal handler to save the
initial value of errno and then restore it before returning if
needed. This is easily forgotten and, if missed, often goes
undetected for a long time.
In commit 3b00fdba9f, we introduced a wrapper signal handler
function that checks whether MyProcPid matches getpid(). This
commit moves the aforementioned errno restoration code from the
individual signal handlers to the new wrapper handler so that we no
longer need to worry about missing it.
Reviewed-by: Andres Freund, Noah Misch
Discussion: https://postgr.es/m/20231121212008.GA3742740%40nathanxps13
For most purposes, multiranges act like dependent objects of the
associated range type: you can't create them separately or drop them
separately. This is like the way that autogenerated array types
behave. However, a couple of points were overlooked: array types
automatically track the ownership of their base type, and array types
do not have their own permissions but use those of the base type,
while multiranges didn't emulate those behaviors. This is fairly
broken, mainly because pg_dump doesn't think it needs to worry about
multiranges as separate objects, and thus it fails to dump/restore
ownership or permissions of multiranges.
There's no apparent value in letting a multirange diverge from
its parent's ownership or permissions, so let's make them act like
arrays in these respects. However, we continue to let multiranges
be renamed or moved to a different schema independently of their
parent, since that doesn't break anything.
Discussion: https://postgr.es/m/1580383.1705343264@sss.pgh.pa.us
This commit introduces a new SQL function pg_sync_replication_slots()
which is used to synchronize the logical replication slots from the
primary server to the physical standby so that logical replication can be
resumed after a failover or planned switchover.
A new 'synced' flag is introduced in pg_replication_slots view, indicating
whether the slot has been synchronized from the primary server. On a
standby, synced slots cannot be dropped or consumed, and any attempt to
perform logical decoding on them will result in an error.
The logical replication slots on the primary can be synchronized to the
hot standby by using the 'failover' parameter of
pg-create-logical-replication-slot(), or by using the 'failover' option of
CREATE SUBSCRIPTION during slot creation, and then calling
pg_sync_replication_slots() on standby. For the synchronization to work,
it is mandatory to have a physical replication slot between the primary
and the standby aka 'primary_slot_name' should be configured on the
standby, and 'hot_standby_feedback' must be enabled on the standby. It is
also necessary to specify a valid 'dbname' in the 'primary_conninfo'.
If a logical slot is invalidated on the primary, then that slot on the
standby is also invalidated.
If a logical slot on the primary is valid but is invalidated on the
standby, then that slot is dropped but will be recreated on the standby in
the next pg_sync_replication_slots() call provided the slot still exists
on the primary server. It is okay to recreate such slots as long as these
are not consumable on standby (which is the case currently). This
situation may occur due to the following reasons:
- The 'max_slot_wal_keep_size' on the standby is insufficient to retain
WAL records from the restart_lsn of the slot.
- 'primary_slot_name' is temporarily reset to null and the physical slot
is removed.
The slot synchronization status on the standby can be monitored using the
'synced' column of pg_replication_slots view.
A functionality to automatically synchronize slots by a background worker
and allow logical walsenders to wait for the physical will be done in
subsequent commits.
Author: Hou Zhijie, Shveta Malik, Ajin Cherian based on an earlier version by Peter Eisentraut
Reviewed-by: Masahiko Sawada, Bertrand Drouvot, Peter Smith, Dilip Kumar, Nisha Moond, Kuroda Hayato, Amit Kapila
Discussion: https://postgr.es/m/514f6f2f-6833-4539-39f1-96cd1e011f23@enterprisedb.com
This reverts commit 95fb5b4902, for reasons similar to what led to
1aa8324b81. In this case, the callback was called once per row, which
is less worse than the previous callback introduced for COPY TO called
once per argument for each row, still the patch set discussed to plug in
custom routines to the COPY paths would be able to know which subroutine
to use depending on its CopyFromState, so this led to a suboptimal
approach at the end.
For now, this part is reverted to consider better which approach to use.
Discussion: https://postgr.es/m/20240206014125.qofww7ew3dx3v3uk@awork3.anarazel.de
The comment was inexact because query_id_enabled will not be switched to
"true" even if compute_query_id is "on", unless a module requests for
it.
While on it, this adds a comment to mention that IsQueryIdEnabled()
should be used to check if query ID computation is enabled or not.
Author: Yugo Nagata
Reviewed-by: Julien Rouhaud
Discussion: https://postgr.es/m/20240209153823.e29a68cadb14225f1362a2cf@sraoss.co.jp
Previously, an interval microseconds field close to INT64_MAX or
INT64_MIN could overflow, producing a result with not even the
correct sign, while being rounded to match a precision specification.
This seems worth fixing, but not worth back-patching, in part
because the ereturn() notation doesn't exist very far back.
Report and patch by Joseph Koshakow (some cosmetic mods by me)
Discussion: https://postgr.es/m/CAAvxfHfpuLgqJYzkUcher466Z1LpmE+5Sm+zc8L6zKCOQ+6TDQ@mail.gmail.com
For a long time, our outfuncs.c code has supposed that the string
contents of a BitString node could just be printed literally with
no concern for quoting/escaping. Now, that's okay if the string
literal contains only valid binary or hex digits ... but our lexer
doesn't check that, preferring to let bitin() be the sole authority
on what's valid. So we could have raw parse trees that contain
incorrect BitString literals, and that can result in failures when
WRITE_READ_PARSE_PLAN_TREES debugging is enabled.
Fix by using outToken() to print the string field, and debackslash()
to read it. This results in a change in the emitted representation
only in cases that would have failed before, and don't represent valid
SQL in the first place. Between that and the fact that we don't store
raw parse trees in the catalogs, I judge this safe to apply without a
catversion bump.
Per bug #18340 from Alexander Lakhin. Back-patch to v16; before that,
we lacked readfuncs support for BitString nodes, so that the problem
was only cosmetic.
Discussion: https://postgr.es/m/18340-4aa1ae6ed4121912@postgresql.org
The macOS Finder application creates .DS_Store files in directories
when opened, which creates problems for serverside utilities which
expect all files to be PostgreSQL specific files. Skip these files
when encountered in pg_checksums, pg_rewind and pg_basebackup.
This was extracted from a larger patchset for skipping hidden files
and system files, where the concencus was to just skip these. Since
this is equally likely to happen in every version, backpatch to all
supported versions.
Reported-by: Mark Guertin <markguertin@gmail.com>
Reviewed-by: Michael Paquier <michael@paquier.xyz>
Reviewed-by: Tobias Bussmann <t.bussmann@gmx.net>
Discussion: https://postgr.es/m/E258CE50-AB0E-455D-8AAD-BB4FE8F882FB@gmail.com
Backpatch-through: v12
If available, read directly from WAL buffers, avoiding the need to go
through the filesystem. Only for physical replication for now, but can
be expanded to other callers.
In preparation for replicating unflushed WAL data.
Author: Bharath Rupireddy
Discussion: https://postgr.es/m/CALj2ACXKKK%3DwbiG5_t6dGao5GoecMwRkhr7GjVBM_jg54%2BNa%3DQ%40mail.gmail.com
Reviewed-by: Andres Freund, Alvaro Herrera, Nathan Bossart, Dilip Kumar, Nitin Jadhav, Melih Mutlu, Kyotaro Horiguchi
Doing this instead of regular updates serves two purposes. First, that avoids
possible waiting on the row-level lock. Second, that avoids dealing with
TOAST.
It's known that changes made by heap_inplace_update() may be lost due to
concurrent normal updates. However, we are OK with that. The subsequent
connections will still have a chance to set "dathasloginevt" to false.
Reported-by: Alexander Lakhin
Discussion: https://postgr.es/m/e2a0248e-5f32-af0c-9832-a90d303c2c61%40gmail.com
Timezones are not immutable and so neither is any function that relies on
them. In commit 66ea94e8, we introduced a few methods which do casting
from one time to another and thus may involve the current timezone. To
preserve the immutability of jsonpath functions currently marked
immutable, disallow these methods from being called from non-TZ aware
functions.
Jeevan Chalke, per a report from Jian He.
Since its introduction, pg_get_expr() has intended to silently
return NULL if called with an invalid relation OID, as can happen
when scanning the catalogs concurrently with relation drops.
However, there is a race condition: we check validity of the OID
at the start, but it could get dropped just afterward, leading to
failures. This is the cause of some intermittent instability we're
seeing in a proposed new test case, and presumably it's a hazard in
the field as well.
We can fix this by AccessShareLock-ing the target relation for the
duration of pg_get_expr(). Since we don't require any permissions
on the target relation, this is semantically a bit undesirable. But
it turns out that the set_relation_column_names() subroutine already
takes a transient AccessShareLock on that relation, and has done since
commit 2ffa740be in 2012. Given the lack of complaints about that, it
seems like there should be no harm in holding the lock a bit longer.
Back-patch to all supported branches.
Discussion: https://postgr.es/m/31ddcc01-a71b-4e8c-9948-01d1c47293ca@eisentraut.org
group_keys_reorder_by_pathkeys() function searched for matching pathkeys
within root->group_pathkeys. That could lead to picking an aggregate pathkey
and using its pathkey->pk_eclass->ec_sortref as an argument of
get_sortgroupref_clause_noerr(). Given that ec_sortref of an aggregate pathkey
references aggregate targetlist not query targetlist, this leads to incorrect
query optimization.
Fix this by looking for matching pathkeys only within the first
num_groupby_pathkeys pathkeys.
Reported-by: David G. Johnston
Discussion: https://postgr.es/m/CAKFQuwY3Ek%3DcLThgd8FdaSc5JRDVt0FaV00gMcWra%2BTAR4gGUw%40mail.gmail.com
Author: Andrei Lepikhov, Alexander Korotkov
Fix for 344d62fb9a9: That commit introduced unlogged sequences and
made it so that identity/serial sequences automatically get the
persistence level of their owning table. But this works only for
CREATE TABLE and not for ALTER TABLE / ADD COLUMN. The latter would
always create the sequence as logged (default), independent of the
persistence setting of the table. This is fixed here.
Note: It is allowed to change the persistence of identity sequences
directly using ALTER SEQUENCE. So mistakes in existing databases can
be fixed manually.
Reviewed-by: Ashutosh Bapat <ashutosh.bapat.oss@gmail.com>
Discussion: https://www.postgresql.org/message-id/flat/c4b6e2ed-bcdf-4ea7-965f-e49761094827%40eisentraut.org
This commit fixes an oversight introduced in c61a2f5841, where COPY TO
would attempt to do encoding conversions even if the encodings of the
client and the server matched for multi-byte encodings. All conversions
go through pg_any_to_server() that makes the conversion a no-op when the
encodings of the client and the server match, even for multi-byte
encodings.
The logic was fine, but setting CopyToStateData->need_transcoding would
cause strlen() to be called for nothing for each attribute of all the
rows copied, and that was showing high in some profiles (more attributes
make that easier to reach). This change improves the runtime of some
worst-case COPY TO queries by 15%~ (number present at least here).
This is a performance improvement, so no backpatch is done out of
caution as this is not a regression.
Reported-by: Andres Freund
Analyzed-by: Andres Freund
Author: Michael Paquier
Reviewed-by: Heikki Linnakangas
Discussion: https://postgr.es/m/20240206020504.edijzczkgd25ek6z@awork3.anarazel.de
The TransactionIdInRecentPast() should return false for all the transactions
older than TransamVariables->oldestClogXid. However, the function contains
a bug in comparison FullTransactionId to TransactionID allowing full
transactions between nextXid - 2^32 and oldestClogXid - 2^31.
This commit fixes TransactionIdInRecentPast() by turning the oldestClogXid into
FullTransactionId first, then performing the comparison.
Backpatch to all supported versions.
Reported-by: Egor Chindyaskin
Bug: 18212
Discussion: https://postgr.es/m/18212-547307f8adf57262%40postgresql.org
Author: Karina Litskevich
Reviewed-by: Kyotaro Horiguchi
Backpatch-through: 12
A comment in grouping_planner() claimed that the PlannerInfo
upper_targets array was not used in core code. However, the code that
generated the paths for the UPPERREL_PARTIAL_DISTINCT rel made that
comment untrue.
Here we adjust the create_distinct_paths() function signature to pass
down the PathTarget the same as is done for create_grouping_paths(),
thus making the aforementioned comment true again.
In passing adjust the order of the upper_targets[] assignments. These
seem to be following the reverse enum order apart from
UPPERREL_PARTIAL_DISTINCT.
Also, update the header comment for generate_gather_paths() to mention
the function is also used to create gather paths for partial distinct
paths.
Author: Richard Guo, David Rowley
Discussion: https://postgr.es/m/CAMbWs48u9VoVOouJsys1qOaC9WVGVmBa+wT1dx8KvxF5GPzezA@mail.gmail.com
Commit 861f86beea used REGBUF_NO_CHANGE at one of the places in the hash
index to register the clean buffers but forgot to avoid setting LSN in
that case.
Reported-by: Michael Paquier
Author: Kuroda Hayato
Reviewed-by: Amit Kapila, Michael Paquier
Discussion: https://postgr.es/m/ZbyVVG_7eW3YD5-A@paquier.xyz
Following are a few clean-ups related to failover option support in slots:
1. Improve the documentation in create_subscription.sgml.
2. Remove the spurious blank line in subscriptioncmds.c.
3. Remove the NOTICE for alter_replication_slot in subscriptioncmds.c as
we would sometimes print it even when nothing has changed. One can find
the change by enabling log_replication_commands on the publisher.
4. Optimize ReplicationSlotAlter() function to prevent disk flushing when
the slot's data remains unchanged.
Author: Hou Zhijie
Reviewed-by: Amit Kapila
Discussion: https://postgr.es/m/514f6f2f-6833-4539-39f1-96cd1e011f23@enterprisedb.com
Discussion: https://postgr.es/m/OS0PR01MB57164904651FB588A518E98894472@OS0PR01MB5716.jpnprd01.prod.outlook.com
This reverts commit 2889fd23be, following a discussion with Andres
Freund as this callback, being called once per attribute when sending a
relation's row, can involve a lot of indirect function calls (more
attributes to deal with means more impact). The effects of a dispatch
at this level would become more visible when improving the per-row code
execution of COPY TO, impacting future potential performance
improvements.
Discussion: https://postgr.es/m/20240206014125.qofww7ew3dx3v3uk@awork3.anarazel.de
The new concurrency model proposed for slru.c to improve performance
does not include any single lock that would coordinate processes
doing concurrent reads/writes on SlruShared->latest_page_number.
We can instead use atomic reads and writes for that variable.
Author: Dilip Kumar <dilipbalaut@gmail.com>
Reviewed-by: Andrey M. Borodin <x4mmm@yandex-team.ru>
Discussion: https://postgr.es/m/CAFiTN-vzDvNz=ExGXz6gdyjtzGixKSqs0mKHMmaQ8sOSEFZ33A@mail.gmail.com
The standalone functions fasthash{32,64} use length for two purposes:
how many bytes to hash, and how to perturb the internal seed.
Developers using the incremental interface may not know the length
ahead of time (e.g. for C strings). In this case, it's advised to
pass length to the finalizer, but initialization still needed some
length up front, in the form of a placeholder macro.
Separate the concerns by having the standalone functions perturb the
internal seed themselves from their own length parameter, allowing
to remove "len" from fasthash_init(), as well as the placeholder macro.
Discussion: https://postgr.es/m/CANWCAZbTUk2LOyhsFo33gjLyLAHZ7ucXCi5K9u%3D%2BPtnTShDKtw%40mail.gmail.com
When assertions are disabled, the built SQL statement is invalid and
you get a "syntax error". So this isn't a serious problem, but let's
avoid the assertion failure.
Backpatch to all supported versions.
Reviewed-by: Noah Misch
The internal commands in REFRESH MATERIALIZED VIEW CONCURRENTLY are
correctly executed in SECURITY_RESTRICTED_OPERATION mode, except for
creating the temporary "diff" table, because you cannot create
temporary tables in SRO mode. But creating the temporary "diff" table
is a pretty complex CTAS command that selects from another temporary
table created earlier in the command. If you can cajole that CTAS
command to execute code defined by the table owner, the table owner
can run code with the privileges of the user running the REFRESH
command.
The proof-of-concept reported to the security team relied on CREATE
RULE to convert the internally-built temp table to a view. That's not
possible since commit b23cd185fd, and I was not able to find a
different way to turn the SELECT on the temp table into code
execution, so as far as I know this is only exploitable in v15 and
below. That's a fiddly assumption though, so apply this patch to
master and all stable versions.
Thanks to Pedro Gallegos for the report.
Security: CVE-2023-5869
Reviewed-by: Noah Misch
This patch provides support for regular (non-replication) connections in
libpqrcv_connect(). This can be used to execute SQL statements on the
primary server without starting a walsender.
A new API libpqrcv_get_dbname_from_conninfo() is also added to extract the
database name from the given connection-info.
Note that this patch doesn't change any existing functionality but later
patches implementing the slot synchronization will use this functionality
to connect to the primary server to fetch required slot information.
Author: Shveta Malik, Hou Zhijie, Ajin Cherian
Reviewed-by: Peter Smith, Bertrand Drouvot, Dilip Kumar, Masahiko Sawada, Nisha Moond, Kuroda Hayato, Amit Kapila
Discussion: https://postgr.es/m/514f6f2f-6833-4539-39f1-96cd1e011f23@enterprisedb.com
These routines are used by the text and CSV formats to send an output
representation of a string, applying quotes if required. This is
similar to 95fb5b4902, reducing the number of "if" branches that need
to be checked on a per-row basis when sending representation of fields
in text or CSV mode.
While on it, this simplifies the signature of CopyAttributeOutCSV() as
it is possible to know that an attribute is alone on a line thanks to
CopyToState. Headers should not use quotes, even if forced at query
level.
Extracted from a larger patch by the same author.
Author: Sutou Kouhei
Discussion: https://postgr.es/m/20231204.153548.2126325458835528809.kou@clear-code.com
CopyReadAttributes{CSV,Text}() are used to parse lines for text and CSV
format. This reduces the number of "if" branches that need to be
checked when parsing fields in CSV and text mode when dealing with a
COPY FROM, something that can become more noticeable with more
attributes and more lines to process.
Extracted from a larger patch by the same author.
Author: Sutou Kouhei
Discussion: https://postgr.es/m/20231204.153548.2126325458835528809.kou@clear-code.com
A ResourceOwnerEnlarge() call was missing. That led to an error:
ERROR: ResourceOwnerRemember called but array was full
and an assertion failure, if you tried to extend a temp relation again
after a failure. Alexander's test case used running out of disk space
to trigger the original failure.
This bug was introduced in the large ResourceOwner rewrite commit
b8bff07daa. Before that, the UnpinLocalBuffer() call guaranteed that
the subsequent PinLocalBuffer() will succeed, but after the rewrite,
releasing an old resource doesn't guarantee that there is space for a
new one.
Add a comment explaining why the UnpinBuffer + PinBuffer calls in
BufferAlloc(), with no ResourceOwnerEnlarge() in between, are safe.
Reported-by: Alexander Lakhin
Discussion: https://www.postgresql.org/message-id/dc574fea-c83e-a600-08cd-10881762e4fa@gmail.com
Here we adjust the partial path generation for parallel DISTINCT queries
to add Sort nodes on top of any unsorted partial distinct paths.
This increases the likelihood of the planner pushing a Sort below a Gather
Merge which enables the final phase of the parallel distinct to be
implemented using a Unique node in more cases.
Sorting the partial distinct paths is particularly useful when the
DISTINCT query has an ORDER BY and LIMIT clause as this can allow cheaper
plans by having the workers Hash Aggregate then Sort before feeding the
results into the Gather Merge. The non-parallel portion of the plan then
becomes very cheap as it leaves only Unique and Limit to do in the leader
process.
Author: Richard Guo
Reviewed-by: David Rowley
Discussion: https://postgr.es/m/CAMbWs48u9VoVOouJsys1qOaC9WVGVmBa+wT1dx8KvxF5GPzezA@mail.gmail.com
An OS crash could leave PG_VERSION empty or missing. The same symptom
appeared in a backup by block device snapshot, taken after the next
checkpoint and before the OS flushes the PG_VERSION blocks. Device
snapshots are not a documented backup method, however. Back-patch to
v15, where commit 9c08aea6a3 introduced
STRATEGY=WAL_LOG and made it the default.
Discussion: https://postgr.es/m/20240130195003.0a.nmisch@google.com
Restoring a base backup taken in the middle of CreateDirAndVersionFile()
or write_relmap_file() would lose the function's effects. The symptom
was absence of the database directory, PG_VERSION file, or
pg_filenode.map. If missing the directory, recovery would fail. Either
missing file would not fail recovery but would render the new database
unusable. Fix CreateDirAndVersionFile() with the transam/README "action
first and then write a WAL entry" strategy. That has a side benefit of
moving filesystem mutations out of a critical section, reducing the ways
to PANIC. Fix the write_relmap_file() call with a lock acquisition, so
it interacts with checkpoints like non-CREATE DATABASE calls do.
Back-patch to v15, where commit 9c08aea6a3
introduced STRATEGY=WAL_LOG and made it the default.
Discussion: https://postgr.es/m/20240130195003.0a.nmisch@google.com
This impacts the statistics retrieved in transactions for the following
views when updating the value of stats_fetch_consistency, leading to
behaviors contrary to what is documented since 605994651b as an update
of this parameter should discard all statistics snapshot data:
- pg_stat_archiver
- pg_stat_bgwriter
- pg_stat_checkpointer
- pg_stat_io
- pg_stat_slru
- pg_stat_wal
For example, updating stats_fetch_consistency from "snapshot" to "cache"
in a transaction did not re-fetch any fresh data, using data cached from
the time when "snapshot" was in use.
Author: Shinya Kato
Discussion: https://postgr.es/m/d77fc5190d4dbe1738d77231488e768b@oss.nttdata.com
Backpatch-through: 15
When building a MergeAppendPath which has child paths that are not
sorted correctly for the MergeAppend's sort order, we apply the cost of
sorting those paths to the MergeAppendPath costs.
Here we fix a bug where the number of tuples specified that needed to be
sorted was effectively pg_class.reltuples rather than the number of
expected row in the subpath. This effectively penalizes MergeAppend
plans any time any filter is present on the MergeAppend subpath as the
sort cost added is to sort all tuples in the table rather than just the
ones expected the path to return.
This did not affect UNION ALL type queries as the RelOptInfo tuples is
set from the subquery's path rows. It does affect MergeAppends uses for
inheritance and partitioned tables.
This is a long-standing bug introduced when MergeAppend was first added
in 11cad29c9. No backpatch as this could result in plan changes.
Author: Alexander Kuzmenkov
Reviewed-by: Ashutosh Bapat, Aleksander Alekseev, David Rowley
Discussion: https://postgr.es/m/CALzhyqyhoXQDR-Usd_0HeWk%3DuqNLzoVeT8KhRoo%3DpV_KzgO3QQ%40mail.gmail.com
Tom Lane