Previously if a promotion was triggered while recovery was paused,
the paused state continued. Also recovery could be paused by executing
pg_wal_replay_pause() even while a promotion was ongoing. That is,
recovery pause had higher priority over a standby promotion.
But this behavior was not desirable because most users basically wanted
the recovery to complete as soon as possible and the server to become
the master when they requested a promotion.
This commit changes recovery so that it prefers a promotion over
recovery pause. That is, if a promotion is triggered while recovery
is paused, the paused state ends and a promotion continues. Also
this commit makes recovery pause functions like pg_wal_replay_pause()
throw an error if they are executed while a promotion is ongoing.
Internally, this commit adds new internal function PromoteIsTriggered()
that returns true if a promotion is triggered. Since the name of
this function and the existing function IsPromoteTriggered() are
confusingly similar, the commit changes the name of IsPromoteTriggered()
to IsPromoteSignaled, as more appropriate name.
Author: Fujii Masao
Reviewed-by: Atsushi Torikoshi, Sergei Kornilov
Discussion: https://postgr.es/m/00c194b2-dbbb-2e8a-5b39-13f14048ef0a@oss.nttdata.com
restore_command has only been used until now by the backend, but there
is a pending patch for pg_rewind to make use of that in the frontend.
Author: Alexey Kondratov
Reviewed-by: Andrey Borodin, Andres Freund, Alvaro Herrera, Alexander
Korotkov, Michael Paquier
Discussion: https://postgr.es/m/a3acff50-5a0d-9a2c-b3b2-ee36168955c1@postgrespro.ru
This commit introduces new wait events BackupWaitWalArchive and
RecoveryPause. The former is reported while waiting for the WAL files
required for the backup to be successfully archived. The latter is
reported while waiting for recovery in pause state to be resumed.
Author: Fujii Masao
Reviewed-by: Michael Paquier, Atsushi Torikoshi, Robert Haas
Discussion: https://postgr.es/m/f0651f8c-9c96-9f29-0ff9-80414a15308a@oss.nttdata.com
Until now, only selected bulk operations (e.g. COPY) did this. If a
given relfilenode received both a WAL-skipping COPY and a WAL-logged
operation (e.g. INSERT), recovery could lose tuples from the COPY. See
src/backend/access/transam/README section "Skipping WAL for New
RelFileNode" for the new coding rules. Maintainers of table access
methods should examine that section.
To maintain data durability, just before commit, we choose between an
fsync of the relfilenode and copying its contents to WAL. A new GUC,
wal_skip_threshold, guides that choice. If this change slows a workload
that creates small, permanent relfilenodes under wal_level=minimal, try
adjusting wal_skip_threshold. Users setting a timeout on COMMIT may
need to adjust that timeout, and log_min_duration_statement analysis
will reflect time consumption moving to COMMIT from commands like COPY.
Internally, this requires a reliable determination of whether
RollbackAndReleaseCurrentSubTransaction() would unlink a relation's
current relfilenode. Introduce rd_firstRelfilenodeSubid. Amend the
specification of rd_createSubid such that the field is zero when a new
rel has an old rd_node. Make relcache.c retain entries for certain
dropped relations until end of transaction.
Back-patch to 9.5 (all supported versions). This introduces a new WAL
record type, XLOG_GIST_ASSIGN_LSN, without bumping XLOG_PAGE_MAGIC. As
always, update standby systems before master systems. This changes
sizeof(RelationData) and sizeof(IndexStmt), breaking binary
compatibility for affected extensions. (The most recent commit to
affect the same class of extensions was
089e4d405d0f3b94c74a2c6a54357a84a681754b.)
Kyotaro Horiguchi, reviewed (in earlier, similar versions) by Robert
Haas. Heikki Linnakangas and Michael Paquier implemented earlier
designs that materially clarified the problem. Reviewed, in earlier
designs, by Andrew Dunstan, Andres Freund, Alvaro Herrera, Tom Lane,
Fujii Masao, and Simon Riggs. Reported by Martijn van Oosterhout.
Discussion: https://postgr.es/m/20150702220524.GA9392@svana.org
The function assumed forkNum=MAIN_FORKNUM and page_std=true, ignoring
the actual arguments. Existing callers passed exactly those values, so
there's no live bug. Back-patch to v12, where the function first
appeared, because another fix needs this.
Discussion: https://postgr.es/m/20191118045434.GA1173436@rfd.leadboat.com
swap_relation_files() calls toast_get_valid_index() to find and lock
this index, just before swapping with the rebuilt TOAST index. The
latter function releases the lock before returning. Potential for
mischief is low; a concurrent session can issue ALTER INDEX ... SET
(fillfactor = ...), which is not alarming. Nonetheless, changing
pg_class.relfilenode without a lock is unconventional. Back-patch to
9.5 (all supported versions), because another fix needs this.
Discussion: https://postgr.es/m/20191226001521.GA1772687@rfd.leadboat.com
Remove comments that are a throw back to a time when nbtree cared about
write-ordering dependencies. The comments are similar to those removed
by commit 9ee7414e, among others.
This commit renames RecoveryWalAll and RecoveryWalStream wait events to
RecoveryWalStream and RecoveryRetrieveRetryInterval, respectively,
in order to make the names and what they are more consistent. For example,
previously RecoveryWalAll was reported as a wait event while the recovery
was waiting for WAL from a stream, and which was confusing because the name
was very different from the situation where the wait actually could happen.
The names of macro variables for those wait events also are renamed
accordingly.
This commit also changes the category of RecoveryRetrieveRetryInterval to
Timeout from Activity because the wait event is reported while waiting based
on wal_retrieve_retry_interval.
Author: Fujii Masao
Reviewed-by: Kyotaro Horiguchi, Atsushi Torikoshi
Discussion: https://postgr.es/m/124997ee-096a-5d09-d8da-2c7a57d0816e@oss.nttdata.com
Only internal page splits need to call _bt_pgaddtup() instead of
PageAddItem(), and only for data items, one of which will end up at the
first offset (or first offset after the high key offset) on the new
right page. This data item alone will need to be truncated in
_bt_pgaddtup().
Since there is no reason why retail inserts ever need to truncate the
incoming item, use a raw PageAddItem() call there instead. Even
_bt_split() uses raw PageAddItem() calls for left page and right page
high keys. Clearly the _bt_pgaddtup() shim function wasn't really
encapsulating anything. _bt_pgaddtup() should now be thought of as a
_bt_split() helper function.
Note that the assertions from commit d1e241c2 verify that retail inserts
never insert an item at an internal page's negative infinity offset.
This invariant could only ever be violated as a result of a basic logic
error in nbtinsert.c.
Commit 2b272734, which added the fastpath rightmost leaf page cache
insert optimization, added code to _bt_doinsert() to handle using and
invalidating the backend local block cache. It doesn't seem like a good
place to handle these low level details, though. _bt_doinsert() is
supposed to be a high level function -- it is the main entry point to
nbtinsert.c.
Restructure the code by placing handling of the rightmost block cache at
the start of a new _bt_search() shim function, _bt_search_insert(). The
new function is called from _bt_doinsert(), which uses it as a
_bt_search() variant that conveniently accepts its BTInsertState state
as an argument. _bt_doinsert() no longer needs to directly consider the
fastpath optimization.
Discussion: https://postgr.es/m/CAH2-Wzk59cxKJRd=rfbyub6-V4yWRjsOYRkUNHBLT1P1GdtCQQ@mail.gmail.com
Copying block and offset numbers to local variables in _bt_insertonpg()
made the code less readable. Remove the variables. There is already
code that conditionally calls BufferGetBlockNumber() in the same block,
so consistently do it that way instead.
Calling BufferGetBlockNumber() is very cheap, but we might as well avoid
it when it isn't truly necessary. It isn't truly necessary for
_bt_insertonpg() to call BufferGetBlockNumber() in almost all cases.
Spotted while working on a patch that refactors the fastpath rightmost
leaf page cache optimization, which was added by commit 2b272734.
We don't need to manually clean up allocations in a SRF's
multi_call_memory_ctx, because the SRF_RETURN_DONE infrastructure
takes care of that (and also ensures that it will happen even if the
function never gets a final call, which simple manual cleanup cannot
do).
Hence, the code removed by this patch is a waste of code and cycles.
Worse, it gives the impression that cleaning up manually is a thing,
which can lead to more serious errors such as those fixed in
commits 085b6b667 and b4570d33a. So we should get rid of it.
These are not quite actual bugs though, so I couldn't muster the
enthusiasm to back-patch. Fix in HEAD only.
Justin Pryzby
Discussion: https://postgr.es/m/20200308173103.GC1357@telsasoft.com
Refactor nbtinsert.c so that the final itemsz of each new non-pivot
tuple (the MAXALIGN()'d size) is determined once. Most of the functions
used by leaf page inserts used the insertstate.itemsz value already.
This commit makes everything use insertstate.itemsz as standard
practice. The goal is to decouple tuple size from "effective" tuple
size. Making this distinction isn't truly necessary right now, but that
might change in the future.
Also explain why we consistently apply MAXALIGN() to get an effective
index tuple size. This was rather unclear, in part because it isn't
actually strictly necessary right now.
Introduce a GUC and a tablespace option to control I/O prefetching, much
like effective_io_concurrency, but for work that is done on behalf of
many client sessions.
Use the new setting in heapam.c instead of the hard-coded formula
effective_io_concurrency + 10 introduced by commit 558a9165e0. Go with
a default value of 10 for now, because it's a round number pretty close
to the value used for that existing case.
Discussion: https://postgr.es/m/CA%2BhUKGJUw08dPs_3EUcdO6M90GnjofPYrWp4YSLaBkgYwS-AqA%40mail.gmail.com
The init_ps_display() arguments were mostly lies by now, so to match
typical usage, just use one argument and let the caller assemble it
from multiple sources if necessary. The only user of the additional
arguments is BackendInitialize(), which was already doing string
assembly on the caller side anyway.
Remove the second argument of set_ps_display() ("force") and just
handle that in init_ps_display() internally.
BackendInitialize() also used to set the initial status as
"authentication", but that was very far from where authentication
actually happened. So now it's set to "initializing" and then
"authentication" just before the actual call to
ClientAuthentication().
Reviewed-by: Julien Rouhaud <rjuju123@gmail.com>
Reviewed-by: Kuntal Ghosh <kuntalghosh.2007@gmail.com>
Reviewed-by: Alvaro Herrera <alvherre@2ndquadrant.com>
Discussion: https://www.postgresql.org/message-id/flat/c65e5196-4f04-4ead-9353-6088c19615a3@2ndquadrant.com
Previously, hard links were not used on Windows and Cygwin, but they
support them just fine in currently supported OS versions, so we can
use them there as well.
Since all supported platforms now support hard links, we can remove
the alternative code paths.
Rename durable_link_or_rename() to durable_rename_excl() to make the
purpose more clear without referencing the implementation details.
Discussion: https://www.postgresql.org/message-id/flat/72fff73f-dc9c-4ef4-83e8-d2e60c98df48%402ndquadrant.com
Commit 074251db added an assertion that verified the fastpath/rightmost
page insert optimization's assumption about free space: There should
always be enough free space on the page to insert the new item without
splitting the page. Otherwise, we end up using the "concurrent root
page split" phony/fake stack path in _bt_insert_parent(). This does not
lead to incorrect behavior, but it is likely to be far slower than
simply using the regular _bt_search() path. The assertion catches
serious performance bugs that would probably take a long time to detect
any other way.
It seems much more natural to make this assertion just before the point
that we generate a fake/phony descent stack. Move the assert there.
This also makes _bt_insertonpg() a bit more readable.
Only a very basic logic bug in a _bt_insertonpg() caller could lead to a
violation of this invariant. Besides, any newitemoff used for an
internal page is sanitized using other "can't happen" errors in
_bt_getstackbuf() or its callers, before _bt_insertonpg() even gets
called.
Also, move the error/assertion from the insert-without-split path of
_bt_insertonpg() to the top of the same function. There is no reason
why this invariant only applies to insertions that happen to not result
in a page split; cover every insertion. The assertion naturally belongs
next to the existing generic assertions that document relatively
high-level invariants for the item being inserted.
The need for this was removed by
8b9e9644dc.
A number of files now need to include utils/acl.h or
parser/parse_node.h explicitly where they previously got it indirectly
somehow.
Since parser/parse_node.h already includes nodes/parsenodes.h, the
latter is then removed where the former was added. Also, remove
nodes/pg_list.h from objectaddress.h, since that's included via
nodes/parsenodes.h.
Reviewed-by: Tom Lane <tgl@sss.pgh.pa.us>
Reviewed-by: Alvaro Herrera <alvherre@2ndquadrant.com>
Discussion: https://www.postgresql.org/message-id/flat/7601e258-26b2-8481-36d0-dc9dca6f28f1%402ndquadrant.com
This commit replaces 0 used as an initial value of XLogSource variable,
with XLOG_FROM_ANY. Also this commit changes those variable so that
XLogSource instead of int is used as the type for them. These changes
are for code readability and debugger-friendliness.
Author: Kyotaro Horiguchi
Reviewed-by: Fujii Masao
Discussion: https://postgr.es/m/20200227.124830.2197604521555566121.horikyota.ntt@gmail.com
At the end of recovery, standby mode is turned off to re-fetch the last
valid record from archive or pg_wal. Previously, if recovery target was
reached and standby mode was turned off while the current WAL source
was stream, recovery could try to retrieve WAL file containing the last
valid record unexpectedly from stream even though not in standby mode.
This caused an assertion failure. That is, the assertion test confirms that
WAL file should not be retrieved from stream if standby mode is not true.
This commit moves back the current WAL source to archive if it's stream
even though not in standby mode, to avoid that assertion failure.
This issue doesn't cause the server to crash when built with assertion
disabled. In this case, the attempt to retrieve WAL file from stream not
in standby mode just fails. And then recovery tries to retrieve WAL file
from archive or pg_wal.
Back-patch to all supported branches.
Author: Kyotaro Horiguchi
Reviewed-by: Fujii Masao
Discussion: https://postgr.es/m/20200227.124830.2197604521555566121.horikyota.ntt@gmail.com
I noticed while testing some other stuff that the CHECK_ENCODING_ROUNDTRIP
logic in ginCompressPostingList could account for over 50% of the runtime
of an INSERT with a GIN index. While that's not relevant to production
performance, it's still kind of annoying in a debug build. Replacing
the loop around short memcmp's with one long memcmp works just as well
and is significantly faster, at least on my machine.
Our usual practice for "poor man's enum" catalog columns is to define
macros for the possible values and use those, not literal constants,
in C code. But for some reason lost in the mists of time, this was
never done for typalign/attalign or typstorage/attstorage. It's never
too late to make it better though, so let's do that.
The reason I got interested in this right now is the need to duplicate
some uses of the TYPSTORAGE constants in an upcoming ALTER TYPE patch.
But in general, this sort of change aids greppability and readability,
so it's a good idea even without any specific motivation.
I may have missed a few places that could be converted, and it's even
more likely that pending patches will re-introduce some hard-coded
references. But that's not fatal --- there's no expectation that
we'd actually change any of these values. We can clean up stragglers
over time.
Discussion: https://postgr.es/m/16457.1583189537@sss.pgh.pa.us
_bt_split() is passed NULL as its insertion scankey for internal page
splits. Two recently added Assert() statements failed to consider this,
leading to a crash with pg_upgrade'd BREE_VERSION < 4 indexes. Remove
the assertions.
The assertions in question were added by commit 0d861bbb, which added
nbtree deduplication. It would be possible to fix the assertions
directly instead, but they weren't adding much anyway.
This commit adds pg_stat_progress_basebackup view that reports
the progress while an application like pg_basebackup is taking
a base backup. This uses the progress reporting infrastructure
added by c16dc1aca5, adding support for streaming base backup.
Bump catversion.
Author: Fujii Masao
Reviewed-by: Kyotaro Horiguchi, Amit Langote, Sergei Kornilov
Discussion: https://postgr.es/m/9ed8b801-8215-1f3d-62d7-65bff53f6e94@oss.nttdata.com
This also involves renaming src/include/utils/hashutils.h, which
becomes src/include/common/hashfn.h. Perhaps an argument can be
made for keeping the hashutils.h name, but it seemed more
consistent to make it match the name of the file, and also more
descriptive of what is actually going on here.
Patch by me, reviewed by Suraj Kharage and Mark Dilger. Off-list
advice on how not to break the Windows build from Davinder Singh
and Amit Kapila.
Discussion: http://postgr.es/m/CA+TgmoaRiG4TXND8QuM6JXFRkM_1wL2ZNhzaUKsuec9-4yrkgw@mail.gmail.com
Deduplication reduces the storage overhead of duplicates in indexes that
use the standard nbtree index access method. The deduplication process
is applied lazily, after the point where opportunistic deletion of
LP_DEAD-marked index tuples occurs. Deduplication is only applied at
the point where a leaf page split would otherwise be required. New
posting list tuples are formed by merging together existing duplicate
tuples. The physical representation of the items on an nbtree leaf page
is made more space efficient by deduplication, but the logical contents
of the page are not changed. Even unique indexes make use of
deduplication as a way of controlling bloat from duplicates whose TIDs
point to different versions of the same logical table row.
The lazy approach taken by nbtree has significant advantages over a GIN
style eager approach. Most individual inserts of index tuples have
exactly the same overhead as before. The extra overhead of
deduplication is amortized across insertions, just like the overhead of
page splits. The key space of indexes works in the same way as it has
since commit dd299df8 (the commit that made heap TID a tiebreaker
column).
Testing has shown that nbtree deduplication can generally make indexes
with about 10 or 15 tuples for each distinct key value about 2.5X - 4X
smaller, even with single column integer indexes (e.g., an index on a
referencing column that accompanies a foreign key). The final size of
single column nbtree indexes comes close to the final size of a similar
contrib/btree_gin index, at least in cases where GIN's posting list
compression isn't very effective. This can significantly improve
transaction throughput, and significantly reduce the cost of vacuuming
indexes.
A new index storage parameter (deduplicate_items) controls the use of
deduplication. The default setting is 'on', so all new B-Tree indexes
automatically use deduplication where possible. This decision will be
reviewed at the end of the Postgres 13 beta period.
There is a regression of approximately 2% of transaction throughput with
synthetic workloads that consist of append-only inserts into a table
with several non-unique indexes, where all indexes have few or no
repeated values. The underlying issue is that cycles are wasted on
unsuccessful attempts at deduplicating items in non-unique indexes.
There doesn't seem to be a way around it short of disabling
deduplication entirely. Note that deduplication of items in unique
indexes is fairly well targeted in general, which avoids the problem
there (we can use a special heuristic to trigger deduplication passes in
unique indexes, since we're specifically targeting "version bloat").
Bump XLOG_PAGE_MAGIC because xl_btree_vacuum changed.
No bump in BTREE_VERSION, since the representation of posting list
tuples works in a way that's backwards compatible with version 4 indexes
(i.e. indexes built on PostgreSQL 12). However, users must still
REINDEX a pg_upgrade'd index to use deduplication, regardless of the
Postgres version they've upgraded from. This is the only way to set the
new nbtree metapage flag indicating that deduplication is generally
safe.
Author: Anastasia Lubennikova, Peter Geoghegan
Reviewed-By: Peter Geoghegan, Heikki Linnakangas
Discussion:
https://postgr.es/m/55E4051B.7020209@postgrespro.ruhttps://postgr.es/m/4ab6e2db-bcee-f4cf-0916-3a06e6ccbb55@postgrespro.ru
Invent the concept of a B-Tree equalimage ("equality implies image
equality") support function, registered as support function 4. This
indicates whether it is safe (or not safe) to apply optimizations that
assume that any two datums considered equal by an operator class's order
method must be interchangeable without any loss of semantic information.
This is static information about an operator class and a collation.
Register an equalimage routine for almost all of the existing B-Tree
opclasses. We only need two trivial routines for all of the opclasses
that are included with the core distribution. There is one routine for
opclasses that index non-collatable types (which returns 'true'
unconditionally), plus another routine for collatable types (which
returns 'true' when the collation is a deterministic collation).
This patch is infrastructure for an upcoming patch that adds B-Tree
deduplication.
Author: Peter Geoghegan, Anastasia Lubennikova
Discussion: https://postgr.es/m/CAH2-Wzn3Ee49Gmxb7V1VJ3-AC8fWn-Fr8pfWQebHe8rYRxt5OQ@mail.gmail.com
Multi-insert for heap is not yet used actively for catalogs, but the
code to support this case is in place for logical decoding. The
existing code forgot to issue a XLOG_HEAP2_NEW_CID record for the first
tuple inserted, leading to failures when attempting to use multiple
inserts for catalogs at decoding time. This commit fixes the problem by
WAL-logging the needed CID.
This is not an active bug, so no back-patch is done.
Author: Daniel Gustafsson
Discussion: https://postgr.es/m/E0D4CC67-A1CF-4DF4-991D-B3AC2EB5FAE9@yesql.se
The comments in fd.c have long claimed that all file allocations should
go through that module, but in reality that's not always practical.
fd.c doesn't supply APIs for invoking some FD-producing syscalls like
pipe() or epoll_create(); and the APIs it does supply for non-virtual
FDs are mostly insistent on releasing those FDs at transaction end;
and in some cases the actual open() call is in code that can't be made
to use fd.c, such as libpq.
This has led to a situation where, in a modern server, there are likely
to be seven or so long-lived FDs per backend process that are not known
to fd.c. Since NUM_RESERVED_FDS is only 10, that meant we had *very*
few spare FDs if max_files_per_process is >= the system ulimit and
fd.c had opened all the files it thought it safely could. The
contrib/postgres_fdw regression test, in particular, could easily be
made to fall over by running it under a restrictive ulimit.
To improve matters, invent functions Acquire/Reserve/ReleaseExternalFD
that allow outside callers to tell fd.c that they have or want to allocate
a FD that's not directly managed by fd.c. Add calls to track all the
fixed FDs in a standard backend session, so that we are honestly
guaranteeing that NUM_RESERVED_FDS FDs remain unused below the EMFILE
limit in a backend's idle state. The coding rules for these functions say
that there's no need to call them in code that just allocates one FD over
a fairly short interval; we can dip into NUM_RESERVED_FDS for such cases.
That means that there aren't all that many places where we need to worry.
But postgres_fdw and dblink must use this facility to account for
long-lived FDs consumed by libpq connections. There may be other places
where it's worth doing such accounting, too, but this seems like enough
to solve the immediate problem.
Internally to fd.c, "external" FDs are limited to max_safe_fds/3 FDs.
(Callers can choose to ignore this limit, but of course it's unwise
to do so except for fixed file allocations.) I also reduced the limit
on "allocated" files to max_safe_fds/3 FDs (it had been max_safe_fds/2).
Conceivably a smarter rule could be used here --- but in practice,
on reasonable systems, max_safe_fds should be large enough that this
isn't much of an issue, so KISS for now. To avoid possible regression
in the number of external or allocated files that can be opened,
increase FD_MINFREE and the lower limit on max_files_per_process a
little bit; we now insist that the effective "ulimit -n" be at least 64.
This seems like pretty clearly a bug fix, but in view of the lack of
field complaints, I'll refrain from risking a back-patch.
Discussion: https://postgr.es/m/E1izCmM-0005pV-Co@gemulon.postgresql.org
VACUUM may truncate heap in several batches. The activity report
is logged for each batch, and contains the number of pages in the table
before and after the truncation, and also the elapsed time during
the truncation. Previously the elapsed time reported in each batch was
the total elapsed time since starting the truncation until finishing
each batch. For example, if the truncation was processed dividing into
three batches, the second batch reported the accumulated time elapsed
during both first and second batches. This is strange and confusing
because the number of pages in the table reported together is not
total. Instead, each batch should report the time elapsed during
only that batch.
The cause of this issue was that the resource usage snapshot was
initialized only at the beginning of the truncation and was never
reset later. This commit fixes the issue by changing VACUUM so that
the resource usage snapshot is reset at each batch.
Back-patch to all supported branches.
Reported-by: Tatsuhito Kasahara
Author: Tatsuhito Kasahara
Reviewed-by: Masahiko Sawada, Fujii Masao
Discussion: https://postgr.es/m/CAP0=ZVJsf=NvQuy+QXQZ7B=ZVLoDV_JzsVC1FRsF1G18i3zMGg@mail.gmail.com
btbuild() has nothing to say about how NULL values compare in nbtree.
Besides, there are _bt_compare() header comments that describe how NULL
values are handled.
Commit b10714080 moved the GinPageSetDeleteXid() call to a spot where
the "page" variable was pointing to the wrong page, causing the XID
to be inserted on a page that's not being deleted, thus allowing later
GinPageIsRecyclable tests to recycle the deleted page too soon.
It might be a good idea to stop using the single "page" variable for
multiple purposes in this function. But for the moment I just moved
the GinPageSetDeleteXid() call down beside the GinPageSetDeleted()
call, which seems like a more logical place for it anyway.
Back-patch to v11, as the faulty patch was. (Fortunately, the bug
hasn't made it into any release yet.)
Discussion: https://postgr.es/m/21620.1581098806@sss.pgh.pa.us
Tuple conversion incorrectly concluded that no conversion was needed
as long as all the attributes lined up. But if the source tuple has a
missing attribute (from addition of a column with default), then the
destination tupdesc might not reflect the same default. The typical
symptom was that the affected columns would be unexpectedly NULL.
Repair by always forcing conversion if the source has missing
attributes, which will be filled in by the deform operation. (In
theory we could optimize for when the destination has the same
default, but that seemed overkill.)
Backpatch to 11 where missing attributes were added.
Per bug #16242.
Vik Fearing (discovery, code, testing) and me (analysis, testcase).
Discussion: https://postgr.es/m/16242-d1c9fca28445966b@postgresql.org
Using 32 bit counters means they can now realistically wrap around when
vacuuming extremely large tables. Because they're signed integers,
stats printed by vacuum look very odd when they do.
We'd love to backpatch this, but refrain because the variables are
exported and could cause third-party code to break.
Reviewed-by: Julien Rouhaud, Tom Lane, Michael Paquier
Discussion: https://postgr.es/m/20200131205926.GA16367@alvherre.pgsql
Using a lookup table of digit pairs reduces the number of divisions
needed, and calculating the length upfront saves some work; these
ideas are taken from the code previously committed for floats.
David Fetter, reviewed by Kyotaro Horiguchi, Tels, and me.
Discussion: https://postgr.es/m/20190924052620.GP31596%40fetter.org
If no DSM slots are available, a ParallelContext can still be
created, but its seg pointer is NULL. Teach parallel btree build
to cope with that by falling back to a regular non-parallel build,
to avoid crashing with a segmentation fault.
Back-patch to 11, where parallel CREATE INDEX landed.
Reported-by: Nicola Contu
Reviewed-by: Peter Geoghegan
Discussion: https://postgr.es/m/CA%2BhUKGJgJEBnkuODBVomyK3MWFvDBbMVj%3Dgdt6DnRPU-5sQ6UQ%40mail.gmail.com
We used to strategically place newlines after some function call left
parentheses to make pgindent move the argument list a few chars to the
left, so that the whole line would fit under 80 chars. However,
pgindent no longer does that, so the newlines just made the code
vertically longer for no reason. Remove those newlines, and reflow some
of those lines for some extra naturality.
Reviewed-by: Michael Paquier, Tom Lane
Discussion: https://postgr.es/m/20200129200401.GA6303@alvherre.pgsql
Before, if a recovery target is configured, but the archive ended
before the target was reached, recovery would end and the server would
promote without further notice. That was deemed to be pretty wrong.
With this change, if the recovery target is not reached, it is a fatal
error.
Based-on-patch-by: Leif Gunnar Erlandsen <leif@lako.no>
Reviewed-by: Kyotaro Horiguchi <horikyota.ntt@gmail.com>
Discussion: https://www.postgresql.org/message-id/flat/993736dd3f1713ec1f63fc3b653839f5@lako.no
The following changes make the predicate locking functions more
generic and suitable for use by future access methods:
- PredicateLockTuple() is renamed to PredicateLockTID(). It takes
ItemPointer and inserting transaction ID instead of HeapTuple.
- CheckForSerializableConflictIn() takes blocknum instead of buffer.
- CheckForSerializableConflictOut() no longer takes HeapTuple or buffer.
Author: Ashwin Agrawal
Reviewed-by: Andres Freund, Kuntal Ghosh, Thomas Munro
Discussion: https://postgr.es/m/CALfoeiv0k3hkEb3Oqk%3DziWqtyk2Jys1UOK5hwRBNeANT_yX%2Bng%40mail.gmail.com
The signature of XLogReadRecord() required the caller to pass the starting
WAL position as argument, or InvalidXLogRecPtr to continue reading at the
end of previous record. That's slightly awkward to the callers, as most
of them don't want to randomly jump around in the WAL stream, but start
reading at one position and then read everything from that point onwards.
Remove the 'RecPtr' argument and add a new function XLogBeginRead() to
specify the starting position instead. That's more convenient for the
callers. Also, xlogreader holds state that is reset when you change the
starting position, so having a separate function for doing that feels like
a more natural fit.
This changes XLogFindNextRecord() function so that it doesn't reset the
xlogreader's state to what it was before the call anymore. Instead, it
positions the xlogreader to the found record, like XLogBeginRead().
Reviewed-by: Kyotaro Horiguchi, Alvaro Herrera
Discussion: https://www.postgresql.org/message-id/5382a7a3-debe-be31-c860-cb810c08f366%40iki.fi
Detection of WAL records having references to invalid pages
during recovery causes PostgreSQL to raise a PANIC-level error,
aborting the recovery. Setting ignore_invalid_pages to on causes
the system to ignore those WAL records (but still report a warning),
and continue recovery. This behavior may cause crashes, data loss,
propagate or hide corruption, or other serious problems.
However, it may allow you to get past the PANIC-level error,
to finish the recovery, and to cause the server to start up.
Author: Fujii Masao
Reviewed-by: Michael Paquier
Discussion: https://www.postgresql.org/message-id/CAHGQGwHCK6f77yeZD4MHOnN+PaTf6XiJfEB+Ce7SksSHjeAWtg@mail.gmail.com
In commit 40d964ec99, we changed the way memory is allocated for dead
tuples but forgot to update the place where we compute the maximum
number of dead tuples. This could lead to invalid memory requests.
Reported-by: Andres Freund
Diagnosed-by: Andres Freund
Author: Masahiko Sawada
Reviewed-by: Amit Kapila and Dilip Kumar
Discussion: https://postgr.es/m/20200121060020.e3cr7s7fj5rw4lok@alap3.anarazel.de
The BRIN add_value() and union() functions need to make a longer-lived
copy of the argument, if they want to store it in the BrinValues struct
also passed as argument. The functions for the "inclusion operator
classes" used with box, range and inet types didn't take into account
that the union helper function might return its argument as is, without
making a copy. Check for that case, and make a copy if necessary. That
case arises at least with the range_union() function, when one of the
arguments is an 'empty' range:
CREATE TABLE brintest (n numrange);
CREATE INDEX brinidx ON brintest USING brin (n);
INSERT INTO brintest VALUES ('empty');
INSERT INTO brintest VALUES (numrange(0, 2^1000::numeric));
INSERT INTO brintest VALUES ('(-1, 0)');
SELECT brin_desummarize_range('brinidx', 0);
SELECT brin_summarize_range('brinidx', 0);
Backpatch down to 9.5, where BRIN was introduced.
Discussion: https://www.postgresql.org/message-id/e6e1d6eb-0a67-36aa-e779-bcca59167c14%40iki.fi
Reviewed-by: Emre Hasegeli, Tom Lane, Alvaro Herrera
This feature allows the vacuum to leverage multiple CPUs in order to
process indexes. This enables us to perform index vacuuming and index
cleanup with background workers. This adds a PARALLEL option to VACUUM
command where the user can specify the number of workers that can be used
to perform the command which is limited by the number of indexes on a
table. Specifying zero as a number of workers will disable parallelism.
This option can't be used with the FULL option.
Each index is processed by at most one vacuum process. Therefore parallel
vacuum can be used when the table has at least two indexes.
The parallel degree is either specified by the user or determined based on
the number of indexes that the table has, and further limited by
max_parallel_maintenance_workers. The index can participate in parallel
vacuum iff it's size is greater than min_parallel_index_scan_size.
Author: Masahiko Sawada and Amit Kapila
Reviewed-by: Dilip Kumar, Amit Kapila, Robert Haas, Tomas Vondra,
Mahendra Singh and Sergei Kornilov
Tested-by: Mahendra Singh and Prabhat Sahu
Discussion:
https://postgr.es/m/CAD21AoDTPMgzSkV4E3SFo1CH_x50bf5PqZFQf4jmqjk-C03BWg@mail.gmail.comhttps://postgr.es/m/CAA4eK1J-VoR9gzS5E75pcD-OH0mEyCdp8RihcwKrcuw7J-Q0+w@mail.gmail.com
The strategy of GIN index scan is driven by opclass-specific extract_query
method. This method that needed search mode is GIN_SEARCH_MODE_ALL. This
mode means that matching tuple may contain none of extracted entries. Simple
example is '!term' tsquery, which doesn't need any term to exist in matching
tsvector.
In order to handle such scan key GIN calculates virtual entry, which contains
all TIDs of all entries of attribute. In fact this is full scan of index
attribute. And typically this is very slow, but allows to handle some queries
correctly in GIN. However, current algorithm calculate such virtual entry for
each GIN_SEARCH_MODE_ALL scan key even if they are multiple for the same
attribute. This is clearly not optimal.
This commit improves the situation by introduction of "exclude only" scan keys.
Such scan keys are not capable to return set of matching TIDs. Instead, they
are capable only to filter TIDs produced by normal scan keys. Therefore,
each attribute should contain at least one normal scan key, while rest of them
may be "exclude only" if search mode is GIN_SEARCH_MODE_ALL.
The same optimization might be applied to the whole scan, not per-attribute.
But that leads to NULL values elimination problem. There is trade-off between
multiple possible ways to do this. We probably want to do this later using
some cost-based decision algorithm.
Discussion: https://postgr.es/m/CAOBaU_YGP5-BEt5Cc0%3DzMve92vocPzD%2BXiZgiZs1kjY0cj%3DXBg%40mail.gmail.com
Author: Nikita Glukhov, Alexander Korotkov, Tom Lane, Julien Rouhaud
Reviewed-by: Julien Rouhaud, Tomas Vondra, Tom Lane
Introduce new fields amusemaintenanceworkmem and amparallelvacuumoptions
in IndexAmRoutine for parallel vacuum. The amusemaintenanceworkmem tells
whether a particular IndexAM uses maintenance_work_mem or not. This will
help in controlling the memory used by individual workers as otherwise,
each worker can consume memory equal to maintenance_work_mem. The
amparallelvacuumoptions tell whether a particular IndexAM participates in
a parallel vacuum and if so in which phase (bulkdelete, vacuumcleanup) of
vacuum.
Author: Masahiko Sawada and Amit Kapila
Reviewed-by: Dilip Kumar, Amit Kapila, Tomas Vondra and Robert Haas
Discussion:
https://postgr.es/m/CAD21AoDTPMgzSkV4E3SFo1CH_x50bf5PqZFQf4jmqjk-C03BWg@mail.gmail.comhttps://postgr.es/m/CAA4eK1LmcD5aPogzwim5Nn58Ki+74a6Edghx4Wd8hAskvHaq5A@mail.gmail.com
Earlier, we use to postpone deleting empty pages till the second stage of
vacuum to amortize the cost of scanning internal pages. However, that can
sometimes (say vacuum is canceled or errored between first and second
stage) delay the pages to be recycled.
Another thing is that to facilitate deleting empty pages in the second
stage, we need to share the information about internal and empty pages
between different stages of vacuum. It will be quite tricky to share this
information via DSM which is required for the upcoming parallel vacuum
patch.
Also, it will bring the logic to reclaim deleted pages closer to nbtree
where we delete empty pages in each pass.
Overall, the advantages of deleting empty pages in each pass outweigh the
advantages of postponing the same.
Author: Dilip Kumar, with changes by Amit Kapila
Reviewed-by: Sawada Masahiko and Amit Kapila
Discussion: https://postgr.es/m/CAA4eK1LGr+MN0xHZpJ2dfS8QNQ1a_aROKowZB+MPNep8FVtwAA@mail.gmail.com
Make the value null only at pg_stat_activity-output time, as suggested
by Tom Lane, instead of messing with the internal state. This should
appease buildfarm members with force_parallel_mode=regress, which are
running parallel queries on logical replication walsenders.
The fact that walsenders can run parallel queries should perhaps be
studied more carefully, but for the moment let's get rid of the red
blots in buildfarm.
Backpatch to pg10, like the previous commit.
Discussion: https://postgr.es/m/30804.1578438763@sss.pgh.pa.us
Returning a non-NULL time is pointless, sinc a walsender is not a
process that would be running normal transactions anyway, but the code
was unintentionally exposing the process start time intermittently,
which was not only bogus but it also confused monitoring systems looking
for idle transactions. Fix by avoiding all updates in walsenders.
Backpatch to 11, where walsenders started appearing in pg_stat_activity.
Reported-by: Tomas Vondra
Discussion: https://postgr.es/m/20191209234409.exe7osmyalwkt5j4@development
Instead of always calling heap_fetch_toast_slice during detoasting,
invoke a table AM callback which, when the toast table is a heap
table, will be heap_fetch_toast_slice.
This makes it possible for a table AM other than heap to be used
as a TOAST table. It also completes the series of commits intended
to improve the interaction of tableam with TOAST that began with
commit 8b94dab06617ef80a0901ab103ebd8754427ef5a; detoast.c is
now, hopefully, fully AM-independent.
Patch by me, reviewed by Andres Freund and Peter Eisentraut.
Discussion: http://postgr.es/m/CA+TgmoZv-=2iWM4jcw5ZhJeL18HF96+W1yJeYrnGMYdkFFnEpQ@mail.gmail.com
Previously, the toast table had to be implemented by the same AM that
was used for the main table, which was bad, because the detoasting
code won't work with anything but heap. This commit doesn't fix the
latter problem, although there's another patch coming which does,
but it does let you pick something that works (i.e. heap, right now).
Patch by me, reviewed by Andres Freund.
Discussion: http://postgr.es/m/CA+TgmoZv-=2iWM4jcw5ZhJeL18HF96+W1yJeYrnGMYdkFFnEpQ@mail.gmail.com
The fastpath insert optimization's incomplete split flag Assert() is
redundant. We'll reach the more general Assert() within
_bt_findinsertloc() in all cases. (Besides, Assert()'ing that the
rightmost page doesn't have the flag set never made much sense.)
Commit 558a9165e0 taught _bt_delitems_delete() to produce its own XID
horizon on the primary. Standbys no longer needed to generate their own
latestRemovedXid, since they could just use the explicitly logged value
from the primary instead. The deleted offset numbers array from the
xl_btree_delete WAL record was no longer used by the REDO routine for
anything other than deleting the items.
This enables a minor optimization: We now treat the array as buffer
state, not generic WAL data, following _bt_delitems_vacuum()'s example.
This should be a minor win, since it allows us to avoid including the
deleted items array in cases where XLogInsert() stores the whole buffer
anyway. The primary goal here is to make the code more maintainable,
though. Removing inessential differences between the two functions
highlights the fundamental differences that remain.
Also change xl_btree_delete to use uint32 for the size of the array of
item offsets being deleted. This brings xl_btree_delete closer to
xl_btree_vacuum. Furthermore, it seems like a good idea to use an
explicit-width integer type (the field was previously an "int").
Bump XLOG_PAGE_MAGIC because xl_btree_delete changed.
Discussion: https://postgr.es/m/CAH2-Wzkz4TjmezzfAbaV1zYrh=fr0bCpzuJTvBe5iUQ3aUPsCQ@mail.gmail.com
Adjust a comment that describes how alignment of the new left page high
key works in btree_xlog_split(), the nbtree page split REDO routine.
The wording used before commit 2c03216d83 is much clearer, so go back
to that.
_bt_delitems_vacuum() comments claimed that it isn't worth another scan
of the page to avoid falsely unsetting the BTP_HAS_GARBAGE page flag
hint (this happens to be the same wording that was removed from
_bt_delitems_delete() by my recent commit fe97c61c). The comments made
little sense, though. The issue can't have much to do with performing a
second scan of the target leaf page, since an LP_DEAD test could easily
be performed in the first scan of the page anyway (the scan that takes
place in btvacuumpage() caller).
Revise the explanation. It makes much more sense to frame this as an
issue about recovery conflicts. _bt_delitems_vacuum() cannot easily
generate an XID cutoff in the same way that _bt_delitems_delete() is
designed to.
Falsely unsetting the page flag is not ideal, and is likely to happen
more often than was supposed by the original comments. Explain why it
usually isn't a problem in practice. There may be an argument for
_bt_delitems_vacuum() not clearing the BTP_HAS_GARBAGE bit, removing the
question of it being falsely unset by VACUUM (there may even be an
argument for not using a page level hint at all). This can be revisited
later.
Commit fe97c61c updated LP_DEAD item deletion comments, but missed a
minor discrepancy on the REDO side. Fix it now.
In passing, don't talk about the btree_xlog_vacuum() behavior within
btree_xlog_delete(). The reliance on XLOG_HEAP2_CLEANUP_INFO records
for recovery conflicts is already discussed within btvacuumpage() and
mentioned again in passing above btree_xlog_vacuum(), which seems
sufficient.
This follows multiple complains from Peter Geoghegan, Andres Freund and
Alvaro Herrera that this issue ought to be dug more before actually
happening, if it happens.
Discussion: https://postgr.es/m/20191226144606.GA5659@alvherre.pgsql
The part in charge of doing the vacuum on all the indexes of a relation
was duplicated, with the same handling for progress reporting done.
While on it, update the progress reporting for heap vacuuming in the
subroutine doing the actual work, keeping the status update local. This
way, any future caller of lazy_vacuum_heap() does not have to worry
about doing any progress reporting update.
Author: Justin Pryzby, Michael Paquier
Discussion: https://postgr.es/m/20191120210600.GC30362@telsasoft.com
The following renaming is done so as source files related to index
access methods are more consistent with table access methods (the
original names used for index AMs ware too generic, and could be
confused as including features related to table AMs):
- amapi.h -> indexam.h.
- amapi.c -> indexamapi.c. Here we have an equivalent with
backend/access/table/tableamapi.c.
- amvalidate.c -> indexamvalidate.c.
- amvalidate.h -> indexamvalidate.h.
- genam.c -> indexgenam.c.
- genam.h -> indexgenam.h.
This has been discussed during the development of v12 when table AM was
worked on, but the renaming never happened.
Author: Michael Paquier
Reviewed-by: Fabien Coelho, Julien Rouhaud
Discussion: https://postgr.es/m/20191223053434.GF34339@paquier.xyz
Using \ is unnecessary and ugly, so remove that. While at it, stitch
the literals back into a single line: we've long discouraged splitting
error message literals even when they go past the 80 chars line limit,
to improve greppability.
Leave contrib/tablefunc alone.
Discussion: https://postgr.es/m/20191223195156.GA12271@alvherre.pgsql
Comments about the consequences of clearing the BTP_HAS_GARBAGE page
flag bit that apply only to VACUUM were added to code that deals with
opportunistic deletion of LP_DEAD items by commit a760893d. The same
comment block was added to both _bt_delitems_vacuum() and
_bt_delitems_delete(). Correct _bt_delitems_delete()'s copy of the
comment block.
_bt_delitems_delete() reliably deletes items that were found by caller
to have their LP_DEAD bit set. There is no question about whether or
not unsetting the BTP_HAS_GARBAGE bit can miss some LP_DEAD items that
were set recently.
Also tweak a related section of the nbtree README.
The REDO routine for nbtree's xl_btree_vacuum record type hasn't
performed a "pin scan" since commit 3e4b7d87 went in, so clearly there
isn't any point in VACUUM WAL-logging information that won't actually be
used. Finish off the work of commit 3e4b7d87 (and the closely related
preceding commit 687f2cd7) by removing the code that generates this
unused information. Also remove the REDO routine code disabled by
commit 3e4b7d87.
Replace the unneeded lastBlockVacuumed field in xl_btree_vacuum with a
new "ndeleted" field. The new field isn't actually needed right now,
since we could continue to infer the array length from the overall
record length. However, an upcoming patch to add deduplication to
nbtree needs to add an "items updated" field to xl_btree_vacuum, so we
might as well start being explicit about the number of items now.
(Besides, it doesn't seem like a good idea to leave the xl_btree_vacuum
struct without any fields; the C standard says that that's undefined.)
nbtree VACUUM no longer forces writing a WAL record for the last block
in the index. Writing out a WAL record with no items for the final
block was supposed to force processing of a lastBlockVacuumed field by a
pin scan.
Bump XLOG_PAGE_MAGIC because xl_btree_vacuum changed.
Discussion: https://postgr.es/m/CAH2-WzmY_mT7UnTzFB5LBQDBkKpdV5UxP3B5bLb7uP%3D%3D6UQJRQ%40mail.gmail.com
The previous coding imagined that it could call before_shmem_exit()
when a non-exclusive backup began and then remove the previously-added
handler by calling cancel_before_shmem_exit() when that backup
ended. However, this only works provided that nothing else in the
system has registered a before_shmem_exit() hook in the interim,
because cancel_before_shmem_exit() is documented to remove a callback
only if it is the latest callback registered. It also only works
if nothing can ERROR out between the time that sessionBackupState
is reset and the time that cancel_before_shmem_exit(), which doesn't
seem to be strictly true.
To fix, leave the handler installed for the lifetime of the session,
arrange to install it just once, and teach it to quietly do nothing if
there isn't a non-exclusive backup in process.
This is a bug, but for now I'm not going to back-patch, because the
consequences are minor. It's possible to cause a spurious warning
to be generated, but that doesn't really matter. It's also possible
to trigger an assertion failure, but production builds shouldn't
have assertions enabled.
Patch by me, reviewed by Kyotaro Horiguchi, Michael Paquier (who
preferred a different approach, but got outvoted), Fujii Masao,
and Tom Lane, and with comments by various others.
Discussion: http://postgr.es/m/CA+TgmobMjnyBfNhGTKQEDbqXYE3_rXWpc4CM63fhyerNCes3mA@mail.gmail.com
The new function, heap_fetch_toast_slice, is shared between
toast_fetch_datum_slice and toast_fetch_datum, and does all the
work of scanning the TOAST table, fetching chunks, and storing
them into the space allocated for the result varlena.
As an incidental side effect, this allows toast_fetch_datum_slice
to perform the scan with only a single scankey if all chunks are
being fetched, which might have some tiny performance benefit.
Discussion: http://postgr.es/m/CA+TgmobBzxwFojJ0zV0Own3dr09y43hp+OzU2VW+nos4PMXWEg@mail.gmail.com
Tuple conversion support in tupconvert.c is able to convert rowtypes
between two relations, inner and outer, which are logically equivalent
but have a different ordering or even dropped columns (used mainly for
inheritance tree and partitions). This makes use of attribute mappings,
which are simple arrays made of AttrNumber elements with a length
matching the number of attributes of the outer relation. The length of
the attribute mapping has been treated as completely independent of the
mapping itself until now, making it easy to pass down an incorrect
mapping length.
This commit refactors the code related to attribute mappings and moves
it into an independent facility called attmap.c, extracted from
tupconvert.c. This merges the attribute mapping with its length,
avoiding to try to guess what is the length of a mapping to use as this
is computed once, when the map is built.
This will avoid mistakes like what has been fixed in dc816e58, which has
used an incorrect mapping length by matching it with the number of
attributes of an inner relation (a child partition) instead of an outer
relation (a partitioned table).
Author: Michael Paquier
Reviewed-by: Amit Langote
Discussion: https://postgr.es/m/20191121042556.GD153437@paquier.xyz
This changes the routines in charge of recycling WAL segments past the
last redo LSN to not use anymore "RedoRecPtr" as a local variable, which
is also available in the context of the session as a static declaration,
replacing it with "lastredoptr". This confusion has been introduced by
d9fadbf, so backpatch down to v11 like the other commit.
Thanks to Tom Lane, Robert Haas, Alvaro Herrera, Mark Dilger and Kyotaro
Horiguchi for the input provided.
Author: Ranier Vilela
Discussion: https://postgr.es/m/MN2PR18MB2927F7B5F690065E1194B258E35D0@MN2PR18MB2927.namprd18.prod.outlook.com
Backpatch-through: 11
Commit d5406dea25 used a slightly
novel, and wrong, approach to compute the length of the last
toast chunk. It worked fine unless the last chunk happened to
have the largest possible size.
Rework some of the checks for bad TOAST chunks to be a bit simpler
and easier to understand. These checks verify that (1) we get all
and only the chunk numbers we expect to see and (2) each chunk has
the expected size. However, the existing code was a bit hard to
understand, at least for me; try to make it clearer.
As part of that, have toast_fetch_datum_slice check the relationship
between endchunk and totalchunks only with an Assert() rather than
checking every chunk number against both values. There's no need to
check that relationship in production builds because it's not a
function of whether on-disk corruption is present; it's just a
question of whether the code does the right math.
Also, have toast_fetch_datum_slice() use ereport(ERROR) rather than
elog(ERROR). Commit fd6ec93bf8 made
the two functions inconsistent with each other.
Rename assorted variables for better clarity and consistency, and
move assorted variables from function scope to the function's main
loop. Remove a few variables that are used only once entirely.
Patch by me, reviewed by Peter Eisentraut.
Discussion: http://postgr.es/m/CA+TgmobBzxwFojJ0zV0Own3dr09y43hp+OzU2VW+nos4PMXWEg@mail.gmail.com
This Assert thought that an overflowed transaction can never get registered
for the group update. But that is not true, because even when the number
of children for a transaction got reduced, the overflow flag is not
changed. And, for group update, we only care about the current number of
children for a transaction that is being committed.
Based on comments by Andres Freund, remove a redundant Assert in
TransactionIdSetPageStatus as we already had a static Assert for the same
condition a few lines earlier.
Reported-by: Vignesh C
Author: Dilip Kumar
Reviewed-by: Amit Kapila
Backpatch-through: 11
Discussion: https://postgr.es/m/CAFiTN-s5=uJw-Z6JC9gcqtBSjXsrHnU63PXBrA=pnBjqnkm5UA@mail.gmail.com
recoveryDelayUntilTime was introduced by commit 36da3cfb45 as a global
because its method of operation was devilishly intrincate. Commit
c945af80cf removed all that complexity and could have turned it into a
local variable, but didn't. Do so now.
Discussion: https://postgr.es/m/20191213200751.GA10731@alvherre.pgsql
Reviewed-by: Michaël Paquier, Daniel Gustafsson
Commit a7ee7c8513 fixed a bug in GiST page split during index creation,
where we failed to re-find the position of a downlink after the page
containing it was split. However, that fix was incomplete; the other call
to gistinserttuples() in the same function needs to also clear
'downlinkoffnum'.
Fixes bug #16134 reported by Alexander Lakhin, for real this time. The
previous fix was enough to fix the crash with the reproducer script for
bug #16162, but the original script for #16134 was still crashing.
Backpatch to v12, like the previous incomplete fix.
Discussion: https://www.postgresql.org/message-id/d869f537-abe4-d2ea-0510-38cd053f5152%40gmail.com
This has been introduced by c16dc1a since progress reporting for VACUUM
has been added. As this issue just causes some extra work and is
harmless, no backpatch is done.
Author: Justin Pryzby
Discussion: https://postgr.es/m/20191213030831.GT2082@telsasoft.com
The bug was similar to the one that was fixed in commit 22251686f0. When
we split page X and insert the downlink for the new page, the parent page
might also need to be split. When that happens, the downlink offset number
we remembered for X is no longer valid. We correctly called
gistFindCorrectParent() to re-find it, but gistFindCorrectParent() doesn't
do anything if the LSN of the page hasn't changed, and we stopped updating
LSNs during index build in commit 9155580fd5. The buggy codepath was taken
if the page was split into three or more pages, and inserting the downlink
caused the parent page to split. To fix, explicitly mark the downlink
offset number as invalid, to force gistFindCorrectParent() to re-find it.
Fixes bug #16134 reported by Alexander Lakhin, reported again as #16162 by
Andreas Kunert. Thanks to Jeff Janes, Tom Lane and Tomas Vondra for
debugging. Backpatch to v12, where we stopped WAL-logging during index
build.
Discussion: https://www.postgresql.org/message-id/16134-0423f729671dec64%40postgresql.org
Discussion: https://www.postgresql.org/message-id/16162-45d21b7b6c1a3105%40postgresql.org
Error messages referring to incorrect WAL segment names could have been
generated for a fsync() failure or when creating a new segment at the
end of recovery.
XLogFileNameP() is a wrapper routine able to build a palloc'd string for
a WAL segment name, which is used for error string generation. There
were several code paths where it gets called in a critical section,
where memory allocation is not allowed. This results in triggering
an assertion failure instead of generating the wanted error message.
Another, more annoying, problem is that if the allocation to generate
the WAL segment name fails on OOM, then the failure would be escalated
to a PANIC.
This removes the routine and all its callers are replaced with a logic
using a fixed-size buffer. This way, all the existing mistakes are
fixed and future ones are prevented.
Author: Masahiko Sawada
Reviewed-by: Michael Paquier, Álvaro Herrera
Discussion: https://postgr.es/m/CA+fd4k5gC9H4uoWMLg9K_QfNrnkkdEw+-AFveob9YX7z8JnKTA@mail.gmail.com
XLogReader, walsender and pg_waldump all had their own routines to read
data from WAL files to memory, with slightly different approaches
according to the particular conditions of each environment. There's a
lot of commonality, so we can refactor that into a single routine
WALRead in XLogReader, and move the differences to a separate (simpler)
callback that just opens the next WAL-segment. This results in a
clearer (ahem) code flow.
The error reporting needs are covered by filling in a new error-info
struct, WALReadError, and it's the caller's responsibility to act on it.
The backend has WALReadRaiseError() to do so.
We no longer ever need to seek in this interface; switch to using
pg_pread().
Author: Antonin Houska, with contributions from Álvaro Herrera
Reviewed-by: Michaël Paquier, Kyotaro Horiguchi
Discussion: https://postgr.es/m/14984.1554998742@spoje.net
This reworks the reloption parsing and build of a couple of index AMs by
creating new structures for each index AM's options. This split was
already done for BRIN, GIN and GiST (which actually has a fillfactor
parameter), but not for hash, B-tree and SPGiST which relied on
StdRdOptions due to an overlap with the default option set.
This saves a couple of bytes for rd_options in each relcache entry with
indexes making use of relation options, and brings more consistency
between all index AMs. While on it, add a couple of AssertMacro() calls
to make sure that utility macros to grab values of reloptions are used
with the expected index AM.
Author: Nikolay Shaplov
Reviewed-by: Amit Langote, Michael Paquier, Álvaro Herrera, Dent John
Discussion: https://postgr.es/m/4127670.gFlpRb6XCm@x200m
If LISTEN is the only action in a serializable-mode transaction,
and the session was not previously listening, and the notify queue
is not empty, predicate.c reported an assertion failure. That
happened because we'd acquire the transaction's initial snapshot
during PreCommit_Notify, which was called *after* predicate.c
expects any such snapshot to have been established.
To fix, just swap the order of the PreCommit_Notify and
PreCommit_CheckForSerializationFailure calls during CommitTransaction.
This will imply holding the notify-insertion lock slightly longer,
but the difference could only be meaningful in serializable mode,
which is an expensive option anyway.
It appears that this is just an assertion failure, with no
consequences in non-assert builds. A snapshot used only to scan
the notify queue could not have been involved in any serialization
conflicts, so there would be nothing for
PreCommit_CheckForSerializationFailure to do except assign it a
prepareSeqNo and set the SXACT_FLAG_PREPARED flag. And given no
conflicts, neither of those omissions affect the behavior of
ReleasePredicateLocks. This admittedly once-over-lightly analysis
is backed up by the lack of field reports of trouble.
Per report from Mark Dilger. The bug is old, so back-patch to all
supported branches; but the new test case only goes back to 9.6,
for lack of adequate isolationtester infrastructure before that.
Discussion: https://postgr.es/m/3ac7f397-4d5f-be8e-f354-440020675694@gmail.com
Discussion: https://postgr.es/m/13881.1574557302@sss.pgh.pa.us
Previously DROP DATABASE generated as many XLOG_DBASE_DROP WAL records
as the number of tablespaces that the database to drop uses. This caused
the scans of shared_buffers as many times as the number of the tablespaces
during recovery because WAL replay of one XLOG_DBASE_DROP record needs
that full scan. This could make the recovery time longer especially
when shared_buffers is large.
This commit changes DROP DATABASE so that it generates only one
XLOG_DBASE_DROP record, and registers the information of all the tablespaces
into it. Then, WAL replay of XLOG_DBASE_DROP record needs full scan of
shared_buffers only once, and which may improve the recovery performance.
Author: Fujii Masao
Reviewed-by: Kirk Jamison, Simon Riggs
Discussion: https://postgr.es/m/CAHGQGwF8YwNH0ZaL+2wjZPkj+ji9UhC+Z4ScnG97WKtVY5L9iw@mail.gmail.com
By oversight 52ac6cd2d0 makes ginDeletePage() sets pd_prune_xid of page to be
deleted before entering the critical section. It appears that only versions 11
and later were affected by this oversight.
Backpatch-through: 11
We find GIN concurrency bugs from time to time. One of the problems here is
that concurrency of GIN isn't well-documented in README. So, it might be even
hard to distinguish design bugs from implementation bugs.
This commit revised concurrency section in GIN README providing more details.
Some examples are illustrated in ASCII art.
Also, this commit add the explanation of how is tuple layout in internal GIN
B-tree page different in comparison with nbtree.
Discussion: https://postgr.es/m/CAPpHfduXR_ywyaVN4%2BOYEGaw%3DcPLzWX6RxYLBncKw8de9vOkqw%40mail.gmail.com
Author: Alexander Korotkov
Reviewed-by: Peter Geoghegan
Backpatch-through: 9.4
Current GIN code appears to don't handle traversing to the deleted page via
downlink. This commit fixes that by stepping right from the delete page like
we do in nbtree.
This commit also fixes setting 'deleted' flag to the GIN pages. Now other page
flags are not erased once page is deleted. That helps to keep our assertions
true if we arrive deleted page via downlink.
Discussion: https://postgr.es/m/CAPpHfdvMvsw-NcE5bRS7R1BbvA4BxoDnVVjkXC5W0Czvy9LVrg%40mail.gmail.com
Author: Alexander Korotkov
Reviewed-by: Peter Geoghegan
Backpatch-through: 9.4
When ginDeletePage() is about to delete page it locks its left sibling to revise
the rightlink. So, it locks pages in right to left manner. Int he same time
ginStepRight() locks pages in left to right manner, and that could cause a
deadlock.
This commit makes ginScanToDelete() keep exclusive lock on left siblings of
currently investigated path. That elimites need to relock left sibling in
ginDeletePage(). Thus, deadlock with ginStepRight() can't happen anymore.
Reported-by: Chen Huajun
Discussion: https://postgr.es/m/5c332bd1.87b6.16d7c17aa98.Coremail.chjischj%40163.com
Author: Alexander Korotkov
Reviewed-by: Peter Geoghegan
Backpatch-through: 10
Make it clear that _bt_pgaddtup() truncates the first data item on an
internal page because its key is supposed to be treated as minus
infinity within _bt_compare().
In detoast_attr_slice, VARSIZE_ANY was used to compute compressed length
of on-disk TOAST values. That's incorrect, because the varlena value may
be just a TOAST pointer, producing either bogus value or crashing.
This is likely why the code was crashing on big-endian machines before
540f316809 replaced the VARSIZE with VARSIZE_ANY, which however only
masked the issue.
Reported-by: Rushabh Lathia
Discussion: https://postgr.es/m/CAL-OGkthU9Gs7TZchf5OWaL-Gsi=hXqufTxKv9qpNG73d5na_g@mail.gmail.com
reloptions.h includes since ba748f7 a set of macros to handle reloption
types in a way similar to how parseRelOptions() works. They have never
been used in the core code, and we have more simple methods now to parse
and fill in rd_options for a given relation depending on its relkind, so
remove this interface to simplify things.
Per discussion between Amit Langote, Álvaro Herrera and me.
Discussion: https://postgr.es/m/CA+HiwqE6zbNO92az6pp5GiTw4tr-9rfCE0t84whQSP+YwSKjMQ@mail.gmail.com
Partitioned tables do not have relation options yet, but, similarly to
what's done for views which have their own parsing table, it could make
sense to introduce new parameters for some of the existing default ones
like fillfactor, autovacuum, etc. Splitting things has the advantage to
make the information stored in rd_options include only the necessary
information, reducing the amount of memory used for a relcache entry
with partitioned tables if new reloptions are introduced at this level.
Author: Nikolay Shaplov
Reviewed-by: Amit Langote, Michael Paquier
Discussion: https://postgr.es/m/1627387.Qykg9O6zpu@x200m
This commit changes xact_desc() so that it reports the detail information about
PREPARE TRANSACTION record, like GID (global transaction identifier),
timestamp at prepare transaction, delete-on-abort/commit relations,
XID of subtransactions, and invalidation messages. These are helpful
when diagnosing 2PC-related troubles.
Author: Fujii Masao
Reviewed-by: Michael Paquier, Andrey Lepikhov, Kyotaro Horiguchi, Julien Rouhaud, Alvaro Herrera
Discussion: https://postgr.es/m/CAHGQGwEvhASad4JJnCv=0dW2TJypZgW_Vpb-oZik2a3utCqcrA@mail.gmail.com
An upcoming patch that adds deduplication to the nbtree AM will rely on
_bt_keep_natts_fast() understanding that differences in TOAST input
state can never affect its answer. In particular, two opclass-equal
datums (with opclasses deemed safe for deduplication) should never be
treated as unequal by _bt_keep_natts_fast() due to TOAST input
differences.
This also seems like a good idea on general principle. nbtsplitloc.c
will now occasionally make better decisions about where to split a leaf
page. The behavior of _bt_keep_natts_fast() is now somewhat closer to
the behavior of _bt_keep_natts().
Discussion: https://postgr.es/m/CAH2-Wzn3Ee49Gmxb7V1VJ3-AC8fWn-Fr8pfWQebHe8rYRxt5OQ@mail.gmail.com
If the passed in xid is the current top transaction, we can do a fast
check and exit early. This should work well for the current heap but
also works very well for proposed AMs that don't use a separate xid
for subtransactions.
Author: Ashwin Agrawal, based on a suggestion from Andres Freund
Reviewed-by: Thomas Munro
Discussion: https://postgr.es/m/CALfoeiv0k3hkEb3Oqk%3DziWqtyk2Jys1UOK5hwRBNeANT_yX%2Bng%40mail.gmail.com
nbtree index builds once stashed the "minimum key" for a page, which was
used as the basis of the pivot tuple that gets placed in the next level
up (i.e. the tuple that stores the downlink to the page in question).
It doesn't quite work that way anymore, so the "minimum key" terminology
now seems misleading (these days the minimum key is actually a straight
copy of the high key from the left sibling, which is a distinct thing in
subtle but important ways). Rename this concept to "low key". This
name is a lot clearer given that there is now a sharp distinction
between pivot and non-pivot tuples. Also remove comments that describe
obsolete details about how the minimum key concept used to work.
Rather than generating the minus infinity item for the leftmost page on
a level by copying the new item and truncating that copy, simply
allocate a small buffer. The old approach confusingly created the
impression that the new item had some kind of significance. This was
another artifact of how things used to work before commits 8224de4f and
dd299df8.
If there is the WAL page that the continuation WAL record just fits within
(i.e., the continuation record ends just at the end of the page) and
the LSN in such page is specified with -s option, previously pg_waldump
caused an assertion failure. The cause of this assertion failure was that
XLogFindNextRecord() that pg_waldump -s calls mistakenly handled
such special WAL page.
This commit changes XLogFindNextRecord() so that it can handle
such WAL page correctly.
Back-patch to all supported versions.
Author: Andrey Lepikhov
Reviewed-by: Fujii Masao, Michael Paquier
Discussion: https://postgr.es/m/99303554-5dd5-06e6-f943-b3005ccd6edd@postgrespro.ru
When maintaining or merging patches, one of the most common sources
for conflicts are the list of objects in makefiles. Especially when
the split across lines has been changed on both sides, which is
somewhat common due to attempting to stay below 80 columns, those
conflicts are unnecessarily laborious to resolve.
By splitting, and alphabetically sorting, OBJS style lines into one
object per line, conflicts should be less frequent, and easier to
resolve when they still occur.
Author: Andres Freund
Discussion: https://postgr.es/m/20191029200901.vww4idgcxv74cwes@alap3.anarazel.de
Historically, the code to build relation options has been shaped the
same way in multiple code paths by using a set of datums in input with
the options parsed with a static table which is then filled with the
option values. This introduces a new common routine in reloptions.c to
do most of the legwork for the in-core code paths.
Author: Amit Langote
Reviewed-by: Michael Paquier
Discussion: https://postgr.es/m/CA+HiwqGsoSn_uTPPYT19WrtR7oYpYtv4CdS0xuedTKiHHWuk_g@mail.gmail.com
As the code stands, nEntries counts the number of ginEntryInsert()
calls, so that's what you end up with at the end of a GIN index build.
However, ginvacuumcleanup() recomputes nEntries as the number of
surviving leaf tuples, and that's generally consistent with the way that
gincostestimate() uses the value. So let's clearly define nEntries
as the number of leaf tuples, and therefore adjust ginEntryInsert() to
increment it only when we make a new one, not when we add TIDs into an
existing tuple or posting tree.
In practice this inconsistency probably has little impact, so I don't
feel a need to back-patch.
Insung Moon and Keisuke Kuroda
Discussion: https://postgr.es/m/CAEMmqBuH_O-oXL+3_ArQ6F5cJ7kXVow2SGQB3HRacku_T+xkmA@mail.gmail.com
9155580 has changed the value of the first fake LSN for unlogged
relations from 1 to FirstNormalUnloggedLSN (aka 1000), GiST requiring a
non-zero LSN on some pages to allow an interlocking logic to work, but
its value was still initialized to 1 at the beginning of recovery or
after running pg_resetwal. This fixes the initialization for both code
paths.
Author: Takayuki Tsunakawa
Reviewed-by: Dilip Kumar, Kyotaro Horiguchi, Michael Paquier
Discussion: https://postgr.es/m/OSBPR01MB2503CE851940C17DE44AE3D9FE6F0@OSBPR01MB2503.jpnprd01.prod.outlook.com
Backpatch-through: 12
We memorize all internal and empty leaf pages in the 1st vacuum stage for
gist indexes. They are used in the 2nd stage, to delete all the empty
pages. There was a memory context page_set_context for this purpose, but
we never used it.
Reported-by: Amit Kapila
Author: Dilip Kumar
Reviewed-by: Amit Kapila
Backpatch-through: 12, where it got introduced
Discussion: https://postgr.es/m/CAA4eK1LGr+MN0xHZpJ2dfS8QNQ1a_aROKowZB+MPNep8FVtwAA@mail.gmail.com
recovery_min_apply_delay parameter is intended for use with streaming
replication deployments. However, the document clearly explains that
the parameter will be honored in all cases if it's specified. So it should
take effect even if in archive recovery. But, previously, archive recovery
with recovery_min_apply_delay enabled always failed, and caused assertion
failure if --enable-caasert is enabled.
The cause of this problem is that; the ownership of recoveryWakeupLatch
that recovery_min_apply_delay uses was taken only when standby mode
is requested. So unowned latch could be used in archive recovery, and
which caused the failure.
This commit changes recovery code so that the ownership of
recoveryWakeupLatch is taken even in archive recovery. Which prevents
archive recovery with recovery_min_apply_delay from failing.
Back-patch to v9.4 where recovery_min_apply_delay was added.
Author: Fujii Masao
Reviewed-by: Michael Paquier
Discussion: https://postgr.es/m/CAHGQGwEyD6HdZLfdWc+95g=VQFPR4zQL4n+yHxQgGEGjaSVheQ@mail.gmail.com
In v11 or before, this setting could not take effect in crash recovery
because it's specified in recovery.conf and crash recovery always
starts without recovery.conf. But commit 2dedf4d9a8 integrated
recovery.conf into postgresql.conf and which unexpectedly allowed
this setting to take effect even in crash recovery. This is definitely
not good behavior.
To fix the issue, this commit makes crash recovery always ignore
recovery_min_apply_delay setting.
Back-patch to v12 where the issue was added.
Author: Fujii Masao
Reviewed-by: Michael Paquier
Discussion: https://postgr.es/m/CAHGQGwEyD6HdZLfdWc+95g=VQFPR4zQL4n+yHxQgGEGjaSVheQ@mail.gmail.com
Discussion: https://postgr.es/m/e445616d-023e-a268-8aa1-67b8b335340c@pgmasters.net
In v11 or before, those settings could not take effect in crash recovery
because they are specified in recovery.conf and crash recovery always
starts without recovery.conf. But commit 2dedf4d9a8 integrated
recovery.conf into postgresql.conf and which unexpectedly allowed
those settings to take effect even in crash recovery. This is definitely
not good behavior.
To fix the issue, this commit makes crash recovery always ignore
restore_command and recovery_end_command settings.
Back-patch to v12 where the issue was added.
Author: Fujii Masao
Reviewed-by: Peter Eisentraut
Discussion: https://postgr.es/m/e445616d-023e-a268-8aa1-67b8b335340c@pgmasters.net
The file descriptor was opened with read-only to fsync a regular file,
which would cause EBADFD errors on some platforms.
This is similar to the recent fix done by a586cc4b (which was broken by
me with 82a5649), except that I noticed this issue while monitoring the
backend code for similar mistakes. Backpatch to 9.4, as this has been
introduced since logical decoding exists as of b89e151.
Author: Michael Paquier
Reviewed-by: Andres Freund
Discussion: https://postgr.es/m/20191006045548.GA14532@paquier.xyz
Backpatch-through: 9.4
The old names for the attribute-detoasting functions names included
the word "heap," which seems outdated now that the heap is only one of
potentially many table access methods.
On the other hand, toast_insert_or_update and toast_delete are
heap-specific, so rename them by adding "heap_" as a prefix.
Not all of the work of making the TOAST system fully accessible to AMs
other than the heap is done yet, but there seems to be little harm in
getting this renaming out of the way now. Commit
8b94dab066 already divided up the
functions among various files partially according to whether it was
intended that they should be heap-specific or AM-agnostic, so this is
just clarifying the division contemplated by that commit.
Patch by me, reviewed and tested by Prabhat Sabu, Thomas Munro,
Andres Freund, and Álvaro Herrera.
Discussion: http://postgr.es/m/CA+TgmoZv-=2iWM4jcw5ZhJeL18HF96+W1yJeYrnGMYdkFFnEpQ@mail.gmail.com
cbc55da has reworked the order of some actions at the end of archive
recovery. Unfortunately this overlooked the fact that the startup
process needs to remove RECOVERYXLOG (for temporary WAL segment newly
recovered from archives) and RECOVERYHISTORY (for temporary history
file) at this step, leaving the files around even after recovery ended.
Backpatch to 9.5, like the previous commit.
Author: Sawada Masahiko
Reviewed-by: Fujii Masao, Michael Paquier
Discussion: https://postgr.es/m/CAD21AoBO_eDQub6zojFnWtnmutRBWvYf7=cW4Hsqj+U_R26w3Q@mail.gmail.com
Backpatch-through: 9.5
Commit 4d0e994eed added support for partial TOAST decompression, so the
decompression is interrupted after producing the requested prefix. For
prefix and slices near the beginning of the entry, this may saves a lot
of decompression work.
That however only deals with decompression - the whole compressed entry
was still fetched and re-assembled, even though the compression used
only a small fraction of it. This commit improves that by computing how
much compressed data may be needed to decompress the requested prefix,
and then fetches only the necessary part.
We always need to fetch a bit more compressed data than the requested
(uncompressed) prefix, because the prefix may not be compressible at all
and pglz itself adds a bit of overhead. That means this optimization is
most effective when the requested prefix is much smaller than the whole
compressed entry.
Author: Binguo Bao
Reviewed-by: Andrey Borodin, Tomas Vondra, Paul Ramsey
Discussion: https://www.postgresql.org/message-id/flat/CAL-OGkthU9Gs7TZchf5OWaL-Gsi=hXqufTxKv9qpNG73d5na_g@mail.gmail.com
In v11 or before, recovery target settings could not take effect in
crash recovery because they are specified in recovery.conf and
crash recovery always starts without recovery.conf. But commit
2dedf4d9a8 integrated recovery.conf into postgresql.conf and
which unexpectedly allowed recovery target settings to take effect
even in crash recovery. This is definitely not good behavior.
To fix the issue, this commit makes crash recovery always ignore
recovery target settings.
Back-patch to v12.
Author: Peter Eisentraut
Reviewed-by: Fujii Masao
Discussion: https://postgr.es/m/e445616d-023e-a268-8aa1-67b8b335340c@pgmasters.net
The code was enforcing AccessExclusiveLock for all custom relation
options, which is incorrect as the APIs allow a custom lock level to be
set.
While on it, fix a couple of inconsistencies in the tests and the README
of dummy_index_am.
Oversights in commit 773df88.
Discussion: https://postgr.es/m/20190925234152.GA2115@paquier.xyz
All our current in core relation options of type string (not many,
admittedly) behave in reality like enums. But after seeing an
implementation for enum reloptions, it's clear that strings are messier,
so introduce the new reloption type. Switch all string options to be
enums instead.
Fortunately we have a recently introduced test module for reloptions, so
we don't lose coverage of string reloptions, which may still be used by
third-party modules.
Authors: Nikolay Shaplov, Álvaro Herrera
Reviewed-by: Nikita Glukhov, Aleksandr Parfenov
Discussion: https://postgr.es/m/43332102.S2V5pIjXRx@x200m
Relation options can define a lock mode other than AccessExclusiveMode
since 47167b7, but modules defining custom relation options did not
really have a way to enforce that. Correct that by extending the
current API set so as modules can define a custom lock mode.
Author: Michael Paquier
Reviewed-by: Kuntal Ghosh
Discussion: https://postgr.es/m/20190920013831.GD1844@paquier.xyz
In-core relation options can use a custom lock mode since 47167b7, that
has lowered the lock available for some autovacuum parameters. However
it forgot to consider custom relation options. This causes failures
with ALTER TABLE SET when changing a custom relation option, as its lock
is not defined. The existing APIs to define a custom reloption does not
allow to define a custom lock mode, so enforce its initialization to
AccessExclusiveMode which should be safe enough in all cases. An
upcoming patch will extend the existing APIs to allow a custom lock mode
to be defined.
The problem can be reproduced with bloom indexes, so add a test there.
Reported-by: Nikolay Sharplov
Analyzed-by: Thomas Munro, Michael Paquier
Author: Michael Paquier
Reviewed-by: Kuntal Ghosh
Discussion: https://postgr.es/m/20190920013831.GD1844@paquier.xyz
Backpatch-through: 9.6
The state-tracking of WAL reading in various places was pretty messy,
mostly because the ancient physical-replication WAL reading code wasn't
using the XLogReader abstraction. This led to some untidy code. Make
it prettier by creating two additional supporting structs,
WALSegmentContext and WALOpenSegment which keep track of WAL-reading
state. This makes code cleaner, as well as supports more future
cleanup.
Author: Antonin Houska
Reviewed-by: Álvaro Herrera and (older versions) Robert Haas
Discussion: https://postgr.es/m/14984.1554998742@spoje.net
When a relation is truncated, shared_buffers needs to be scanned
so that any buffers for the relation forks are invalidated in it.
Previously, shared_buffers was scanned for each relation forks, i.e.,
MAIN, FSM and VM, when VACUUM truncated off any empty pages
at the end of relation or TRUNCATE truncated the relation in place.
Since shared_buffers needed to be scanned multiple times,
it could take a long time to finish those commands especially
when shared_buffers was large.
This commit changes the logic so that shared_buffers is scanned only
one time for those three relation forks.
Author: Kirk Jamison
Reviewed-by: Masahiko Sawada, Thomas Munro, Alvaro Herrera, Takayuki Tsunakawa and Fujii Masao
Discussion: https://postgr.es/m/D09B13F772D2274BB348A310EE3027C64E2067@g01jpexmbkw24
This commit improves subject in two ways:
* It removes ugliness of 02f90879e7, which stores distance values and null
flags in two separate arrays after GISTSearchItem struct. Instead we pack
both distance value and null flag in IndexOrderByDistance struct. Alignment
overhead should be negligible, because we typically deal with at most few
"col op const" expressions in ORDER BY clause.
* It fixes handling of "col op NULL" expression in KNN-SP-GiST. Now, these
expression are not passed to support functions, which can't deal with them.
Instead, NULL result is implicitly assumed. It future we may decide to
teach support functions to deal with NULL arguments, but current solution is
bugfix suitable for backpatch.
Reported-by: Nikita Glukhov
Discussion: https://postgr.es/m/826f57ee-afc7-8977-c44c-6111d18b02ec%40postgrespro.ru
Author: Nikita Glukhov
Reviewed-by: Alexander Korotkov
Backpatch-through: 9.4
Include newitemoff in rmgr desc output for nbtree page split records.
In passing, correct an obsolete comment that claimed that newitemoff is
only logged for _L variant nbtree page split WAL records.
Both issues were oversights in commit 2c03216d83, which revamped the
WAL format.
Author: Peter Geoghegan
Backpatch: 9.5-, where the WAL format was revamped.
Assert that _bt_binsrch() binary searches with scantid set in insertion
scankey cannot be performed on leaf pages. Leaf-level binary searches
where scantid is set must use _bt_binsrch_insert() instead.
_bt_binsrch_insert() is likely to have additional responsibilities in
the future, such as searching within GIN-style posting lists using
scantid. It seems like a good idea to tighten things up now.
In order to implement NULL LAST semantic GiST previously assumed distance to
the NULL value to be Inf. However, our distance functions can return Inf and
NaN for non-null values. In such cases, NULL LAST semantic appears to be
broken. This commit fixes that by introducing separate array of null flags for
distances.
Backpatch to all supported versions.
Discussion: https://postgr.es/m/CAPpHfdsNvNdA0DBS%2BwMpFrgwT6C3-q50sFVGLSiuWnV3FqOJuQ%40mail.gmail.com
Author: Alexander Korotkov
Backpatch-through: 9.4
Previously plain float comparison was used in GiST pairing heap. Such
comparison doesn't provide proper ordering for value sets containing Inf and Nan
values. This commit fixes that by usage of float8_cmp_internal(). Note, there
is remaining problem with NULL distances, which are represented as Inf in
pairing heap. It would be fixes in subsequent commit.
Backpatch to all supported versions.
Reported-by: Andrey Borodin
Discussion: https://postgr.es/m/CAPpHfdsNvNdA0DBS%2BwMpFrgwT6C3-q50sFVGLSiuWnV3FqOJuQ%40mail.gmail.com
Author: Alexander Korotkov
Reviewed-by: Heikki Linnakangas
Backpatch-through: 9.4
When using COMMIT AND CHAIN or ROLLBACK AND CHAIN not in an explicit
transaction block, the previous implementation would leave a
transaction block active in the ROLLBACK case but not the COMMIT case.
To fix for now, error out when using these commands not in an explicit
transaction block. This restriction could be lifted if a sensible
definition and implementation is found.
Bug: #15977
Author: fn ln <emuser20140816@gmail.com>
Reviewed-by: Fabien COELHO <coelho@cri.ensmp.fr>
This moves much of the non-heap-specific logic from toast_delete and
toast_insert_or_update into a helper functions accessible via a new
header, toast_helper.h. Using the functions in this module, a table
AM can implement creation and deletion of TOAST table rows with
much less code duplication than was possible heretofore. Some
table AMs won't want to use the TOAST logic at all, but for those
that do this will make that easier.
Patch by me, reviewed and tested by Prabhat Sabu, Thomas Munro,
Andres Freund, and Álvaro Herrera.
Discussion: http://postgr.es/m/CA+TgmoZv-=2iWM4jcw5ZhJeL18HF96+W1yJeYrnGMYdkFFnEpQ@mail.gmail.com
Previously even if postmaster died and WaitLatch() woke up with that event
while pg_promote() was waiting for the standby promotion to finish,
pg_promote() did nothing special and kept waiting until timeout occurred.
This could cause a busy loop.
This patch make pg_promote() return false immediately when postmaster
dies, to avoid such a busy loop.
Back-patch to v12 where pg_promote() was added.
Author: Fujii Masao
Reviewed-by: Michael Paquier
Discussion: https://postgr.es/m/CAHGQGwEs9ROgSp+QF+YdDU+xP8W=CY1k-_Ov-d_Z3JY+to3eXA@mail.gmail.com
detoast.c/h contain functions required to detoast a datum, partially
or completely, plus a few other utility functions for examining the
size of toasted datums.
toast_internals.c/h contain functions that are used internally to the
TOAST subsystem but which (mostly) do not need to be accessed from
outside.
heaptoast.c/h contains code that is intrinsically specific to the
heap AM, either because it operates on HeapTuples or is based on the
layout of a heap page.
detoast.c and toast_internals.c are placed in
src/backend/access/common rather than src/backend/access/heap. At
present, both files still have dependencies on the heap, but that will
be improved in a future commit.
Patch by me, reviewed and tested by Prabhat Sabu, Thomas Munro,
Andres Freund, and Álvaro Herrera.
Discussion: http://postgr.es/m/CA+TgmoZv-=2iWM4jcw5ZhJeL18HF96+W1yJeYrnGMYdkFFnEpQ@mail.gmail.com
This function is only used by xlogreader.c itself, so there's no need to
export it. It was introduced by commit 3b02ea4f07 with the apparent
intention that it could be used externally, but I couldn't find any
external code calling it.
I (Álvaro) couldn't resist the urge to sort nearby function prototypes
properly while at it.
Author: Antonin Houska
Discussion: https://postgr.es/m/14984.1554998742@spoje.net
The message was included as a parameter when this function was added in
dcb2bda9b7, but I don't think it has ever served any useful purpose.
Let's stop spreading it pointlessly.
Reviewed by Amit Langote and Peter Eisentraut.
Discussion: https://postgr.es/m/20190806224728.GA17233@alvherre.pgsql
In what seems like a fit of misplaced optimization,
ExtractReplicaIdentity() accessed the relation's replica-identity
index without taking any lock on it. Usually, the surrounding query
already holds some lock so this is safe enough ... but in the case
of a previously-planned delete, there might be no existing lock.
Given a suitable test case, this is exposed in v12 and HEAD by an
assertion added by commit b04aeb0a0.
The whole thing's rather poorly thought out anyway; rather than
looking directly at the index, we should use the index-attributes
bitmap that's held by the parent table's relcache entry, as the
caller functions do. This is more consistent and likely a bit
faster, since it avoids a cache lookup. Hence, change to doing it
that way.
While at it, rather than blithely assuming that the identity
columns are non-null (with catastrophic results if that's wrong),
add assertion checks that they aren't null. Possibly those should
be actual test-and-elog, but I'll leave it like this for now.
In principle, this is a bug that's been there since this code was
introduced (in 9.4). In practice, the risk seems quite low, since
we do have a lock on the index's parent table, so concurrent
changes to the index's catalog entries seem unlikely. Given the
precedent that commit 9c703c169 wasn't back-patched, I won't risk
back-patching this further than v12.
Per report from Hadi Moshayedi.
Discussion: https://postgr.es/m/CAK=1=Wrek44Ese1V7LjKiQS-Nd-5LgLi_5_CskGbpggKEf3tKQ@mail.gmail.com
The comment did not match what the code actually did for integers with
the 43rd bit set. You get an integer like that, if you have a posting
list with two adjacent TIDs that are more than 2^31 blocks apart.
According to the comment, we would store that in 6 bytes, with no
continuation bit on the 6th byte, but in reality, the code encodes it
using 7 bytes, with a continuation bit on the 6th byte as normal.
The decoding routine also handled these 7-byte integers correctly, except
for an overflow check that assumed that one integer needs at most 6 bytes.
Fix the overflow check, and fix the comment to match what the code
actually does. Also fix the comment that claimed that there are 17 unused
bits in the 64-bit representation of an item pointer. In reality, there
are 64-32-11=21.
Fitting any item pointer into max 6 bytes was an important property when
this was written, because in the old pre-9.4 format, item pointers were
stored as plain arrays, with 6 bytes for every item pointer. The maximum
of 6 bytes per integer in the new format guaranteed that we could convert
any page from the old format to the new format after upgrade, so that the
new format was never larger than the old format. But we hardly need to
worry about that anymore, and running into that problem during upgrade,
where an item pointer is expanded from 6 to 7 bytes such that the data
doesn't fit on a page anymore, is implausible in practice anyway.
Backpatch to all supported versions.
This also includes a little test module to test these large distances
between item pointers, without requiring a 16 TB table. It is not
backpatched, I'm including it more for the benefit of future development
of new posting list formats.
Discussion: https://www.postgresql.org/message-id/33bfc20a-5c86-f50c-f5a5-58e9925d05ff%40iki.fi
Reviewed-by: Masahiko Sawada, Alexander Korotkov
Commit efada2b8e9, which made the nbtree page deletion algorithm more
robust, removed the concept of a half-dead internal page. Remove a
comment about half dead parent pages that was overlooked.
The Postgres approach to coupling locks during an ascent of the tree is
slightly different to the approach taken by Lehman and Yao. Add a new
paragraph to the "Differences to the Lehman & Yao algorithm" section of
the nbtree README that explains the similarities and differences.
Adjust the struct comment that describes how page splits use their
descent stack to cascade up the tree from the leaf level.
In passing, fix up some unrelated nbtree comments that had typos or were
obsolete.
The initial value of the nbtree stack downlink block number field
recorded during an initial descent of the tree wasn't actually used.
Both _bt_getstackbuf() callers overwrote the value with their own value.
Remove the block number field from the stack struct, and add a child
block number argument to _bt_getstackbuf() in its place. This makes the
overall design of _bt_getstackbuf() clearer.
Author: Peter Geoghegan
Reviewed-By: Anastasia Lubennikova
Discussion: https://postgr.es/m/CAH2-Wzmx+UbXt2YNOUCZ-a04VdXU=S=OHuAuD7Z8uQq-PXTYUg@mail.gmail.com
Commit d2086b08b0 removed almost all cases where nbtree must release a
read buffer lock and acquire a write buffer lock instead, so remaining
cases in which that's still necessary are not notable enough to appear
in the nbtree README.
More importantly, holding on to a buffer pin in cases where nbtree must
trade a read lock for a write lock is very unlikely to save any I/O.
This seems to have been a long overlooked throwback to a time when
nbtree cared about write-ordering dependencies, and performed
synchronous buffer writes. It hasn't worked that way in many years.
Use the PageIndexTupleOverwrite() bufpage.c routine within nbtree
instead of deleting a tuple and re-inserting its replacement. This
makes the intent of affected code slightly clearer. It also makes
CREATE INDEX slightly faster, since there is no longer a need to shift
every leaf page's line pointer array back and forth during index builds.
Author: Peter Geoghegan, Anastasia Lubennikova
Reviewed-By: Anastasia Lubennikova
Discussion: https://postgr.es/m/CAH2-Wz=Zk=B9+Vwm376WuO7YTjFc2SSskifQm4Nme3RRRPtOSQ@mail.gmail.com
In serializable mode, heap_hot_search_buffer() incorrectly acquired a
predicate lock on the root tuple, not the returned tuple that satisfied
the visibility checks. As explained in README-SSI, the predicate lock does
not need to be copied or extended to other tuple versions, but for that to
work, the correct, visible, tuple version must be locked in the first
place.
The original SSI commit had this bug in it, but it was fixed back in 2013,
in commit 81fbbfe335. But unfortunately, it was reintroduced a few months
later in commit b89e151054. Wising up from that, add a regression test
to cover this, so that it doesn't get reintroduced again. Also, move the
code that sets 't_self', so that it happens at the same time that the
other HeapTuple fields are set, to make it more clear that all the code in
the loop operate on the "current" tuple in the chain, not the root tuple.
Bug spotted by Andres Freund, analysis and original fix by Thomas Munro,
test case and some additional changes to the fix by Heikki Linnakangas.
Backpatch to all supported versions (9.4).
Discussion: https://www.postgresql.org/message-id/20190731210630.nqhszuktygwftjty%40alap3.anarazel.de
Otherwise, after a deleted page gets even older, it becomes unrecyclable
again. B-tree has the same problem, and has had since time immemorial,
but let's at least fix this in GiST, where this is new.
Backpatch to v12, where GiST page deletion was introduced.
Reviewed-by: Andrey Borodin
Discussion: https://www.postgresql.org/message-id/835A15A5-F1B4-4446-A711-BF48357EB602%40yandex-team.ru
The explicit check in gistScanPage() isn't currently really necessary, as
a deleted page is always empty, so the loop would fall through without
doing anything, anyway. But it's a marginal optimization, and it gives a
nice place to attach a comment to explain how it works.
Backpatch to v12, where GiST page deletion was introduced.
Reviewed-by: Andrey Borodin
Discussion: https://www.postgresql.org/message-id/835A15A5-F1B4-4446-A711-BF48357EB602%40yandex-team.ru
Commit 857f9c36cd, which taught nbtree VACUUM to avoid unnecessary
index scans, bumped the nbtree version number from 2 to 3, while adding
the ability for nbtree indexes to be upgraded on-the-fly. Various
assertions that assumed that an nbtree index was always on version 2 had
to be changed to accept any supported version (version 2 or 3 on
Postgres 11).
However, a few assertions were missed in the initial commit, all of
which were in code paths that cache a local copy of the metapage
metadata, where the index had been expected to be on the current version
(no longer version 2) as a generic sanity check. Rather than simply
update the assertions, follow-up commit 0a64b45152 intentionally made
the metapage caching code update the per-backend cached metadata version
without changing the on-disk version at the same time. This could even
happen when the planner needed to determine the height of a B-Tree for
costing purposes. The assertions only fail on Postgres v12 when
upgrading from v10, because they were adjusted to use the authoritative
shared memory metapage by v12's commit dd299df8.
To fix, remove the cache-only upgrade mechanism entirely, and update the
assertions themselves to accept any supported version (go back to using
the cached version in v12). The fix is almost a full revert of commit
0a64b45152 on the v11 branch.
VACUUM only considers the authoritative metapage, and never bothers with
a locally cached version, whereas everywhere else isn't interested in
the metapage fields that were added by commit 857f9c36cd. It seems
unlikely that this bug has affected any user on v11.
Reported-By: Christoph Berg
Bug: #15896
Discussion: https://postgr.es/m/15896-5b25e260fdb0b081%40postgresql.org
Backpatch: 11-, where VACUUM was taught to avoid unnecessary index scans.
Formerly, lcons was about the same speed as lappend, but with the new
List implementation, that's not so; with a long List, data movement
imposes an O(N) cost on lcons and list_delete_first, but not lappend.
Hence, invent list_delete_last with semantics parallel to
list_delete_first (but O(1) cost), and change various places to use
lappend and list_delete_last where this can be done without much
violence to the code logic.
There are quite a few places that construct result lists using lcons not
lappend. Some have semantic rationales for that; I added comments about
it to a couple that didn't have them already. In many such places though,
I think the coding is that way only because back in the dark ages lcons
was faster than lappend. Hence, switch to lappend where this can be done
without causing semantic changes.
In ExecInitExprRec(), this results in aggregates and window functions that
are in the same plan node being executed in a different order than before.
Generally, the executions of such functions ought to be independent of
each other, so this shouldn't result in visibly different query results.
But if you push it, as one regression test case does, you can show that
the order is different. The new order seems saner; it's closer to
the order of the functions in the query text. And we never documented
or promised anything about this, anyway.
Also, in gistfinishsplit(), don't bother building a reverse-order list;
it's easy now to iterate backwards through the original list.
It'd be possible to go further towards removing uses of lcons and
list_delete_first, but it'd require more extensive logic changes,
and I'm not convinced it's worth it. Most of the remaining uses
deal with queues that probably never get long enough to be worth
sweating over. (Actually, I doubt that any of the changes in this
patch will have measurable performance effects either. But better
to have good examples than bad ones in the code base.)
Patch by me, thanks to David Rowley and Daniel Gustafsson for review.
Discussion: https://postgr.es/m/21272.1563318411@sss.pgh.pa.us
This is numbered take 7, and addresses a set of issues around:
- Fixes for typos and incorrect reference names.
- Removal of unneeded comments.
- Removal of unreferenced functions and structures.
- Fixes regarding variable name consistency.
Author: Alexander Lakhin
Discussion: https://postgr.es/m/10bfd4ac-3e7c-40ab-2b2e-355ed15495e8@gmail.com
The logic just added by commit e3899ffd falls back on a 50:50 page split
in the event of a new item that's just to the right of our provisional
"many duplicates" split point. Fix a comment that incorrectly claimed
that the new item had to be just to the left of our provisional split
point.
Backpatch: 12-, just like commit e3899ffd.
Specific ever-decreasing insertion patterns could cause successive
unbalanced nbtree page splits. Problem cases involve a large group of
duplicates to the left, and ever-decreasing insertions to the right.
To fix, detect the situation by considering the newitem offset before
performing a split using nbtsplitloc.c's "many duplicates" strategy. If
the new item was inserted just to the right of our provisional "many
duplicates" split point, infer ever-decreasing insertions and fall back
on a 50:50 (space delta optimal) split. This seems to barely affect
cases that already had acceptable space utilization.
An alternative fix also seems possible. Instead of changing
nbtsplitloc.c split choice logic, we could instead teach _bt_truncate()
to generate a new value for new high keys by interpolating from the
lastleft and firstright key values. That would certainly be a more
elegant fix, but it isn't suitable for backpatching.
Discussion: https://postgr.es/m/CAH2-WznCNvhZpxa__GqAa1fgQ9uYdVc=_apArkW2nc-K3O7_NA@mail.gmail.com
Backpatch: 12-, where the nbtree page split enhancements were introduced.
Originally, Postgres Lists were a more or less exact reimplementation of
Lisp lists, which consist of chains of separately-allocated cons cells,
each having a value and a next-cell link. We'd hacked that once before
(commit d0b4399d8) to add a separate List header, but the data was still
in cons cells. That makes some operations -- notably list_nth() -- O(N),
and it's bulky because of the next-cell pointers and per-cell palloc
overhead, and it's very cache-unfriendly if the cons cells end up
scattered around rather than being adjacent.
In this rewrite, we still have List headers, but the data is in a
resizable array of values, with no next-cell links. Now we need at
most two palloc's per List, and often only one, since we can allocate
some values in the same palloc call as the List header. (Of course,
extending an existing List may require repalloc's to enlarge the array.
But this involves just O(log N) allocations not O(N).)
Of course this is not without downsides. The key difficulty is that
addition or deletion of a list entry may now cause other entries to
move, which it did not before.
For example, that breaks foreach() and sister macros, which historically
used a pointer to the current cons-cell as loop state. We can repair
those macros transparently by making their actual loop state be an
integer list index; the exposed "ListCell *" pointer is no longer state
carried across loop iterations, but is just a derived value. (In
practice, modern compilers can optimize things back to having just one
loop state value, at least for simple cases with inline loop bodies.)
In principle, this is a semantics change for cases where the loop body
inserts or deletes list entries ahead of the current loop index; but
I found no such cases in the Postgres code.
The change is not at all transparent for code that doesn't use foreach()
but chases lists "by hand" using lnext(). The largest share of such
code in the backend is in loops that were maintaining "prev" and "next"
variables in addition to the current-cell pointer, in order to delete
list cells efficiently using list_delete_cell(). However, we no longer
need a previous-cell pointer to delete a list cell efficiently. Keeping
a next-cell pointer doesn't work, as explained above, but we can improve
matters by changing such code to use a regular foreach() loop and then
using the new macro foreach_delete_current() to delete the current cell.
(This macro knows how to update the associated foreach loop's state so
that no cells will be missed in the traversal.)
There remains a nontrivial risk of code assuming that a ListCell *
pointer will remain good over an operation that could now move the list
contents. To help catch such errors, list.c can be compiled with a new
define symbol DEBUG_LIST_MEMORY_USAGE that forcibly moves list contents
whenever that could possibly happen. This makes list operations
significantly more expensive so it's not normally turned on (though it
is on by default if USE_VALGRIND is on).
There are two notable API differences from the previous code:
* lnext() now requires the List's header pointer in addition to the
current cell's address.
* list_delete_cell() no longer requires a previous-cell argument.
These changes are somewhat unfortunate, but on the other hand code using
either function needs inspection to see if it is assuming anything
it shouldn't, so it's not all bad.
Programmers should be aware of these significant performance changes:
* list_nth() and related functions are now O(1); so there's no
major access-speed difference between a list and an array.
* Inserting or deleting a list element now takes time proportional to
the distance to the end of the list, due to moving the array elements.
(However, it typically *doesn't* require palloc or pfree, so except in
long lists it's probably still faster than before.) Notably, lcons()
used to be about the same cost as lappend(), but that's no longer true
if the list is long. Code that uses lcons() and list_delete_first()
to maintain a stack might usefully be rewritten to push and pop at the
end of the list rather than the beginning.
* There are now list_insert_nth...() and list_delete_nth...() functions
that add or remove a list cell identified by index. These have the
data-movement penalty explained above, but there's no search penalty.
* list_concat() and variants now copy the second list's data into
storage belonging to the first list, so there is no longer any
sharing of cells between the input lists. The second argument is
now declared "const List *" to reflect that it isn't changed.
This patch just does the minimum needed to get the new implementation
in place and fix bugs exposed by the regression tests. As suggested
by the foregoing, there's a fair amount of followup work remaining to
do.
Also, the ENABLE_LIST_COMPAT macros are finally removed in this
commit. Code using those should have been gone a dozen years ago.
Patch by me; thanks to David Rowley, Jesper Pedersen, and others
for review.
Discussion: https://postgr.es/m/11587.1550975080@sss.pgh.pa.us
In order to be able to work with FullTransactionId values during replay
without increasing the size of the WAL, infer the epoch. In general we
can't do that safely, but during replay we can because we know that
nextFullXid can't advance concurrently.
Prevent frontend code from seeing this new function, due to the above
restriction. Perhaps in future it will be possible to extract the value
entirely from independent WAL records, and then this restriction can be
lifted.
Author: Thomas Munro, based on earlier code from Andres Freund
Discussion: https://postgr.es/m/CA%2BhUKG%2BmLmuDjMi6o1dxkKvGRL56Y2Rz%2BiXAcrZV03G9ZuFQ8Q%40mail.gmail.com
Index-based calculation of this operator is exact. So, signature of
gist_bbox_distance() function is changes so that caller is responsible for
setting *recheck flag.
Discussion: https://postgr.es/m/f71ba19d-d989-63b6-f04a-abf02ad9345d%40postgrespro.ru
Author: Nikita Glukhov
Reviewed-by: Tom Lane, Alexander Korotkov
This can cause valgrind to complain, as the flag marking a buffer as a
temporary copy was not getting initialized.
While on it, fill in with zeros newly-created buffer pages. This does
not matter when loading a block from a temporary file, but it makes the
push of an index tuple into a new buffer page safer.
This has been introduced by 1d27dcf, so backpatch all the way down to
9.4.
Author: Alexander Lakhin
Discussion: https://postgr.es/m/15899-0d24fb273b3dd90c@postgresql.org
Backpatch-through: 9.4
Most block-based table AMs will need the exact same implementation of
the relation_size callback as the heap, and if they use a standard
page layout, they will likely need an implementation of the
relation_estimate_size callback that is very similar to that of the
heap. Rearrange to facilitate code reuse.
Patch by me, reviewed by Michael Paquier, Daniel Gustafsson, and
Álvaro Herrera.
Discussion: http://postgr.es/m/CA+TgmoZ6DBPnP1E-vRpQZUJQijJFD54F+SR_pxGiAAS-MyrigA@mail.gmail.com
This addresses a couple of issues in the code:
- Typos and inconsistencies in comments and function declarations.
- Removal of unreferenced function declarations.
- Removal of unnecessary compile flags.
- A cleanup error in regressplans.sh.
Author: Alexander Lakhin
Discussion: https://postgr.es/m/0c991fdf-2670-1997-c027-772a420c4604@gmail.com
Because there is no portable int64/uint64 format specifier and we
can't stick macros like INT64_FORMAT into the middle of a translatable
string, we have been using various workarounds that put the number to
be printed into a string buffer first. Now that we always use our own
sprintf(), we can rely on %lld and %llu to work, so we can use those.
This patch undoes this workaround in a few places where it was
egregiously verbose.
Reviewed-by: Tom Lane <tgl@sss.pgh.pa.us>
Discussion: https://www.postgresql.org/message-id/flat/CAH2-Wz%3DWbNxc5ob5NJ9yqo2RMJ0q4HXDS30GVCobeCvC9A1L9A%40mail.gmail.com
This changes various places where appendPQExpBuffer was used in places
where it was possible to use appendPQExpBufferStr, and likewise for
appendStringInfo and appendStringInfoString. This is really just a
stylistic improvement, but there are also small performance gains to be
had from doing this.
Discussion: http://postgr.es/m/CAKJS1f9P=M-3ULmPvr8iCno8yvfDViHibJjpriHU8+SXUgeZ=w@mail.gmail.com
Instead of calling pg_lsn_in() in check_recovery_target_lsn and
timestamptz_in() in check_recovery_target_time, reorganize the
respective code so that we don't raise any errors in the check hooks.
The previous code tried to use PG_TRY/PG_CATCH to handle errors in a
way that is not safe, so now the code contains no ereport() calls and
can operate safely within the GUC error handling system.
Moreover, since the interpretation of the recovery_target_time string
may depend on the time zone, we cannot do the final processing of that
string until all the GUC processing is done. Instead,
check_recovery_target_time() now does some parsing for syntax
checking, but the actual conversion to a timestamptz value is done
later in the recovery code that uses it.
Reported-by: Andres Freund <andres@anarazel.de>
Reviewed-by: Michael Paquier <michael@paquier.xyz>
Discussion: https://www.postgresql.org/message-id/flat/20190611061115.njjwkagvxp4qujhp%40alap3.anarazel.de
a96c41f has introduced the option for heap, but it still lacked the
variant to control the behavior for toast relations.
While on it, refactor the tests so as they stress more scenarios with
the various values that vacuum_index_cleanup can use. It would be
useful to couple those tests with pageinspect to check that pages are
actually cleaned up, but this is left for later.
Author: Masahiko Sawada, Michael Paquier
Reviewed-by: Peter Geoghegan
Discussion: https://postgr.es/m/CAD21AoCqs8iN04RX=i1KtLSaX5RrTEM04b7NHYps4+rqtpWNEg@mail.gmail.com
This record uses one metadata buffer and registers some data associated
to the buffer, but when parsing the record for its description a direct
access to the record data was done, but there is none. This leads
usually to an incorrect description, but can also cause crashes like in
pg_waldump. Instead, fix things so as the parsing uses the data
associated to the metadata block.
This is an oversight from 3d92796, so backpatch down to 11.
Author: Michael Paquier
Description: https://postgr.es/m/20190617013059.GA3153@paquier.xyz
Backpatch-through: 11
This fixes an embarrassing oversight I (Andres) made in 737a292b,
namely missing two place where liverows/deadrows were used when
converting those variables to pointers, leading to incrementing the
pointer, rather than the value.
It's not that actually that easy to trigger a crash: One needs tuples
deleted by the current transaction, followed by a tuple deleted in
another session, all in one page. Which is presumably why this hasn't
been noticed before.
Reported-By: Steve Singer
Author: Steve Singer
Discussion: https://postgr.es/m/c7988239-d42c-ddc4-41db-171b23b35e4f@ssinger.info
This puts back reverted commit de87a084c0, with some bug fixes.
When two (or more) transactions are waiting for transaction T1 to release a
tuple-level lock, and transaction T1 upgrades its lock to a higher level, a
spurious deadlock can be reported among the waiting transactions when T1
finishes. The simplest example case seems to be:
T1: select id from job where name = 'a' for key share;
Y: select id from job where name = 'a' for update; -- starts waiting for T1
Z: select id from job where name = 'a' for key share;
T1: update job set name = 'b' where id = 1;
Z: update job set name = 'c' where id = 1; -- starts waiting for T1
T1: rollback;
At this point, transaction Y is rolled back on account of a deadlock: Y
holds the heavyweight tuple lock and is waiting for the Xmax to be released,
while Z holds part of the multixact and tries to acquire the heavyweight
lock (per protocol) and goes to sleep; once T1 releases its part of the
multixact, Z is awakened only to be put back to sleep on the heavyweight
lock that Y is holding while sleeping. Kaboom.
This can be avoided by having Z skip the heavyweight lock acquisition. As
far as I can see, the biggest downside is that if there are multiple Z
transactions, the order in which they resume after T1 finishes is not
guaranteed.
Backpatch to 9.6. The patch applies cleanly on 9.5, but the new tests don't
work there (because isolationtester is not smart enough), so I'm not going
to risk it.
Author: Oleksii Kliukin
Discussion: https://postgr.es/m/B9C9D7CD-EB94-4635-91B6-E558ACEC0EC3@hintbits.com
Discussion: https://postgr.es/m/2815.1560521451@sss.pgh.pa.us
This reverts commits 3da73d6839 and de87a084c0.
This code has some tricky corner cases that I'm not sure are correct and
not properly tested anyway, so I'm reverting the whole thing for next
week's releases (reintroducing the deadlock bug that we set to fix).
I'll try again afterwards.
Discussion: https://postgr.es/m/E1hbXKQ-0003g1-0C@gemulon.postgresql.org
When two (or more) transactions are waiting for transaction T1 to release a
tuple-level lock, and transaction T1 upgrades its lock to a higher level, a
spurious deadlock can be reported among the waiting transactions when T1
finishes. The simplest example case seems to be:
T1: select id from job where name = 'a' for key share;
Y: select id from job where name = 'a' for update; -- starts waiting for X
Z: select id from job where name = 'a' for key share;
T1: update job set name = 'b' where id = 1;
Z: update job set name = 'c' where id = 1; -- starts waiting for X
T1: rollback;
At this point, transaction Y is rolled back on account of a deadlock: Y
holds the heavyweight tuple lock and is waiting for the Xmax to be released,
while Z holds part of the multixact and tries to acquire the heavyweight
lock (per protocol) and goes to sleep; once X releases its part of the
multixact, Z is awakened only to be put back to sleep on the heavyweight
lock that Y is holding while sleeping. Kaboom.
This can be avoided by having Z skip the heavyweight lock acquisition. As
far as I can see, the biggest downside is that if there are multiple Z
transactions, the order in which they resume after X finishes is not
guaranteed.
Backpatch to 9.6. The patch applies cleanly on 9.5, but the new tests don't
work there (because isolationtester is not smart enough), so I'm not going
to risk it.
Author: Oleksii Kliukin
Discussion: https://postgr.es/m/B9C9D7CD-EB94-4635-91B6-E558ACEC0EC3@hintbits.com
Vignesh found this bug in the check function for
default_table_access_method's check hook, but that was just copied
from older GUCs. Investigation by Michael and me then found the bug in
further places.
When not connected to a database (e.g. in a walsender connection), we
cannot perform (most) GUC checks that need database access. Even when
only shared tables are needed, unless they're
nailed (c.f. RelationCacheInitializePhase2()), they cannot be accessed
without pg_class etc. being present.
Fix by extending the existing IsTransactionState() checks to also
check for MyDatabaseOid.
Reported-By: Vignesh C, Michael Paquier, Andres Freund
Author: Vignesh C, Andres Freund
Discussion: https://postgr.es/m/CALDaNm1KXK9gbZfY-p_peRFm_XrBh1OwQO1Kk6Gig0c0fVZ2uw%40mail.gmail.com
Backpatch: 9.4-
There were a number of issues in the recent commits which include typos,
code and comments mismatch, leftover function declarations. Fix them.
Reported-by: Alexander Lakhin
Author: Alexander Lakhin, Amit Kapila and Amit Langote
Reviewed-by: Amit Kapila
Discussion: https://postgr.es/m/ef0c0232-0c1d-3a35-63d4-0ebd06e31387@gmail.com
Some of the wrapper functions didn't match the callback names. Many of
them due to staying "consistent" with historic naming of the wrapped
functionality. We decided that for most cases it's more important to
be for tableam to be consistent going forward, than with the past.
The one exception is beginscan/endscan/... because it'd have looked
odd to have systable_beginscan/endscan/... with a different naming
scheme, and changing the systable_* APIs would have caused way too
much churn (including breaking a lot of external users).
Author: Ashwin Agrawal, with some small additions by Andres Freund
Reviewed-By: Andres Freund
Discussion: https://postgr.es/m/CALfoeiugyrXZfX7n0ORCa4L-m834dzmaE8eFdbNR6PMpetU4Ww@mail.gmail.com
This is still using the 2.0 version of pg_bsd_indent.
I thought it would be good to commit this separately,
so as to document the differences between 2.0 and 2.1 behavior.
Discussion: https://postgr.es/m/16296.1558103386@sss.pgh.pa.us
Before this commit, when ANALYZE was run on a table and serializable
was used (either by virtue of an explicit BEGIN TRANSACTION ISOLATION
LEVEL SERIALIZABLE, or default_transaction_isolation being set to
serializable) a null pointer dereference lead to a crash.
The analyze scan doesn't need a snapshot (nor predicate locking), but
before this commit a scan only contained information about being a
bitmap or sample scan.
Refactor the option passing to the scan_begin callback to use a
bitmask instead. Alternatively we could have added a new boolean
parameter, but that seems harder to read. Even before this issue
various people (Heikki, Tom, Robert) suggested doing so.
These changes don't change the scan APIs outside of tableam. The flags
argument could be exposed, it's not necessary to fix this
problem. Also the wrapper table_beginscan* functions encapsulate most
of that complexity.
After these changes fixing the bug is trivial, just don't acquire
predicate lock for analyze style scans. That was already done for
bitmap heap scans. Add an assert that a snapshot is passed when
acquiring the predicate lock, so this kind of bug doesn't require
running with serializable.
Also add a comment about sample scans currently requiring predicate
locking the entire relation, that previously wasn't remarked upon.
Reported-By: Joe Wildish
Author: Andres Freund
Discussion:
https://postgr.es/m/4EA80A20-E9BF-49F1-9F01-5B66CAB21453@elusive.cxhttps://postgr.es/m/20190411164947.nkii4gaeilt4bui7@alap3.anarazel.dehttps://postgr.es/m/20190518203102.g7peu2fianukjuxm@alap3.anarazel.de
Instead add a tableam callback to do so. To avoid adding per
validation overhead, pass a scan to tuple_tid_valid. In heap's case
we'd otherwise incurred a RelationGetNumberOfBlocks() call for each
tid - which'd have added noticable overhead to nodeTidscan.c.
Author: Andres Freund
Reviewed-By: Ashwin Agrawal
Discussion: https://postgr.es/m/20190515185447.gno2jtqxyktylyvs@alap3.anarazel.de
Previously various parts of the code routed size requests through
RelationGetNumberOfBlocks[InFork]. That works if md.c is used by the
AM, but not otherwise.
Add a tableam callback to return the size of the table. As not every
AM will use postgres' BLCKSZ, have it return bytes, and have
RelationGetNumberOfBlocksInFork() round the byte size up into blocks.
To allow code outside of the AM to determine the actual relation size
map InvalidForkNumber the total size of a relation, as not every AM
might just need the postgres defined forks.
A few users of RelationGetNumberOfBlocks() ought to be converted away
from that. One case, the use of it to determine whether a tid is
valid, will be fixed in a follow up commit. Others will have to wait
for v13.
Author: Andres Freund
Discussion: https://postgr.es/m/20190423225201.3bbv6tbqzkb5w7cw@alap3.anarazel.de
It's not safe for nbtree VACUUM to attempt to delete a target page whose
right sibling is already half-dead, since that would fail the
cross-check when VACUUM attempts to re-find a downlink to the right
sibling in the parent page. Logic to prevent this from happening was
added by commit 8da3183780, which addressed a bug in the overhaul of
page deletion that went into PostgreSQL 9.4 (commit efada2b8e9).
VACUUM was made to check the right sibling page, and back off when it
happened to be half-dead already.
However, it is only truly necessary to do the right sibling check on the
leaf level, since that transitively determines if the deletion target's
parent's right sibling page is itself undergoing deletion. Remove the
internal page level check, and add a comment explaining why the leaf
level check alone suffices.
The extra check is also unnecessary due to the fact that internal pages
that are marked half-dead are generally considered corrupt. Commit
efada2b8e9 established the principle that there should never be
half-dead internal pages (internal pages pending deletion are possible,
but that status is never directly represented in the internal page).
VACUUM will complain about corruption when it encounters half-dead
internal pages, so VACUUM is bound to raise an error one way or another
when an nbtree index has a half-dead internal page (contrib/amcheck will
also report that the page is corrupt).
It's possible that a pg_upgrade'd 9.3 database will still have half-dead
internal pages, so it may seem like there is an argument for leaving the
check in place to reliably get a cleaner error message that advises the
user to REINDEX. However, leaf pages are also deleted in the first
phase of deletion prior to PostgreSQL 9.4, so I believe we won't even
attempt to re-find the parent page anyway (we won't have the fully
deleted leaf page as the right sibling of our target page, so we won't
even try to find a downlink for it).
Discussion: https://postgr.es/m/CAH2-Wzm_ntmqJjWLRyKzimFmFvk+BnVAvUpaA4s1h9Ja58woaQ@mail.gmail.com
Remove a Berkeley-era comment above _bt_insertonpg() that admonishes the
reader to grok Lehman and Yao's paper before making any changes. This
made a certain amount of sense back when _bt_insertonpg() was
responsible for most of the things that are now spread across
_bt_insertonpg(), _bt_findinsertloc(), _bt_insert_parent(), and
_bt_split(), but it doesn't work like that anymore.
I believe that this comment alludes to the need to "couple" or "crab"
buffer locks as we ascend the tree as page splits cascade upwards. The
nbtree README already explains this in detail, which seems sufficient.
Besides, the changes to page splits made by commit 40dae7ec53 altered
the exact details of how buffer locks are retained during splits; Lehman
and Yao's original algorithm seems to release the lock on the left child
page/buffer slightly earlier than _bt_insertonpg()/_bt_insert_parent()
can.
Commit fab25024, which taught nbtree to choose candidate split points
more carefully, had _bt_findsplitloc() record all possible split points
in an initial pass over a page that is about to be split. The order
that candidate split points were processed and stored in was assumed to
match the offset number order of split points on an imaginary version of
the page that contains the same items as the original, but also fits
newitem (the item that provoked the split precisely because it didn't
fit).
However, the order of split points in the final array was not quite what
was expected: the split point that makes newitem the firstright item
came after the split point that makes newitem the lastleft item -- not
before. As a result, _bt_findsplitloc() could get confused about the
leftmost and rightmost tuples among all possible split points recorded
for the page. This seems to have no appreciable impact on the quality
of the final split point chosen by _bt_findsplitloc(), but it's still
wrong.
To fix, switch the order in which newitem candidate splits are recorded
in. This also makes it possible to describe candidate split points in
terms of which pair of adjoining tuples enclose the split point within
_bt_findsplitloc(), making it clearer why it's generally safe for
_bt_split() to expect lastleft and firstright tuples.
For some reason both callsite and the implementation for heapam had
the meaning inverted (i.e. succeeded == true was passed in case of
conflict). That's confusing.
I (Andres) briefly pondered whether it'd be better to rename
table_complete_speculative's argument to 'bool specConflict' or such,
but decided not to. The 'complete' in the function name for me makes
`succeeded` sound a bit better.
Reported-By: Ashwin Agrawal, Melanie Plageman, Heikki Linnakangas
Discussion:
https://postgr.es/m/CALfoeitk7-TACwYv3hCw45FNPjkA86RfXg4iQ5kAOPhR+F1Y4w@mail.gmail.comhttps://postgr.es/m/97673451-339f-b21e-a781-998d06b1067c@iki.fi
As we descend the GiST tree during insertion, we modify any downlinks on
the way down to include the new tuple we're about to insert (if they don't
cover it already). Modifying an existing downlink might cause an internal
page to split, if the new downlink tuple is larger than the old one. If
that happens, we need to back up to the parent and re-choose a page to
insert to. We used to detect that situation, thanks to the NSN-LSN
interlock normally used to detect concurrent page splits, but that got
broken by commit 9155580fd5. With that commit, we now use a dummy constant
LSN value for every page during index build, so the LSN-NSN interlock no
longer works. I thought that was OK because there can't be any other
backends modifying the index during index build, but missed that the
insertion itself can modify the page we're inserting to. The consequence
was that we would sometimes insert the new tuple to an incorrect page, one
whose downlink doesn't cover the new tuple.
To fix, add a flag to the stack that keeps track of the state while
descending tree, to indicate that a page was split, and that we need to
retry the descend from the parent.
Thomas Munro first reported that the contrib/intarray regression test was
failing occasionally on the buildfarm after commit 9155580fd5. The failure
was intermittent, because the gistchoose() function is not deterministic,
and would only occasionally create the right circumstances for this bug to
cause the failure.
Patch by Anastasia Lubennikova, with some changes by me to make it work
correctly also when the internal page split also causes the "grandparent"
to be split.
Discussion: https://www.postgresql.org/message-id/CA%2BhUKGJRzLo7tZExWfSbwM3XuK7aAK7FhdBV0FLkbUG%2BW0v0zg%40mail.gmail.com
The conditions listed in this comment have changed several times, and at
some point the thing that the "if so" referred to was negated.
The text was OK up to 9.6. It was differently wrong in v10, v11 and
master, so fix in all those versions.
The term "item pointer" should not be used to refer to ItemIdData
variables, since that is needlessly ambiguous. Only
ItemPointerData/ItemPointer variables should be called item pointers.
To fix, establish the convention that ItemIdData variables should always
be referred to either as "item identifiers" or "line pointers". The
term "item identifier" already predominates in docs and translatable
messages, and so should be the preferred alternative there.
Discussion: https://postgr.es/m/CAH2-Wz=c=MZQjUzde3o9+2PLAPuHTpVZPPdYxN=E4ndQ2--8ew@mail.gmail.com
Commit 8fa30f906b reduced the elevel of a number of "can't happen"
_bt_split() errors from PANIC to ERROR. At the same time, the new right
page buffer for the split could continue to be acquired well before the
critical section. This was possible because it was relatively
straightforward to make sure that _bt_split() could not throw an error,
with a few specific exceptions. The exceptional cases were safe because
they involved specific, well understood errors, making it possible to
consistently zero the right page before actually raising an error using
elog(). There was no danger of leaving around a junk page, provided
_bt_split() stuck to this coding rule.
Commit 8224de4f, which introduced INCLUDE indexes, added code to make
_bt_split() truncate away non-key attributes. This happened at a point
that broke the rule around zeroing the right page in _bt_split(). If
truncation failed (perhaps due to palloc() failure), that would result
in an errant right page buffer with junk contents. This could confuse
VACUUM when it attempted to delete the page, and should be avoided on
general principle.
To fix, reorganize _bt_split() so that truncation occurs before the new
right page buffer is even acquired. A junk page/buffer will not be left
behind if _bt_nonkey_truncate()/_bt_truncate() raise an error.
Discussion: https://postgr.es/m/CAH2-WzkcWT_-NH7EeL=Az4efg0KCV+wArygW8zKB=+HoP=VWMw@mail.gmail.com
Backpatch: 11-, where INCLUDE indexes were introduced.
The right way for IsCatalogRelation/Class to behave is to return true
for OIDs less than FirstBootstrapObjectId (not FirstNormalObjectId),
without any of the ad-hoc fooling around with schema membership.
The previous code was wrong because (1) it claimed that
information_schema tables were not catalog relations but their toast
tables were, which is silly; and (2) if you dropped and recreated
information_schema, which is a supported operation, the behavior
changed. That's even sillier. With this definition, "catalog
relations" are exactly the ones traceable to the postgres.bki data,
which seems like what we want.
With this simplification, we don't actually need access to the pg_class
tuple to identify a catalog relation; we only need its OID. Hence,
replace IsCatalogClass with "IsCatalogRelationOid(oid)". But keep
IsCatalogRelation as a convenience function.
This allows fixing some arguably-wrong semantics in contrib/sepgsql and
ReindexRelationConcurrently, which were using an IsSystemNamespace test
where what they really should be using is IsCatalogRelationOid. The
previous coding failed to protect toast tables of system catalogs, and
also was not on board with the general principle that user-created tables
do not become catalogs just by virtue of being renamed into pg_catalog.
We can also get rid of a messy hack in ReindexMultipleTables.
While we're at it, also rename IsSystemNamespace to IsCatalogNamespace,
because the previous name invited confusion with the more expansive
semantics used by IsSystemRelation/Class.
Also improve the comments in catalog.c.
There are a few remaining places in replication-related code that are
special-casing OIDs below FirstNormalObjectId. I'm inclined to think
those are wrong too, and if there should be any special case it should
just extend to FirstBootstrapObjectId. But first we need to debate
whether a FOR ALL TABLES publication should include information_schema.
Discussion: https://postgr.es/m/21697.1557092753@sss.pgh.pa.us
Discussion: https://postgr.es/m/15150.1557257111@sss.pgh.pa.us
Commit dd299df818, which added suffix truncation to nbtree, simplified
the WAL record format used by page splits. It became necessary to
explicitly WAL-log the new high key for the left half of a split in all
cases, which relieved the REDO routine from having to reconstruct a new
high key for the left page by copying the first item from the right
page. Remove a comment that referred to the previous practice.
It is no longer possible under any circumstances for nbtree code to
reconstruct a strict lower bound key (parent page's pivot tuple key) for
a right sibling page by retrieving the first item in the right sibling
page.
This commit adds new parameter to VACUUM command, TRUNCATE,
which specifies that VACUUM should attempt to truncate off
any empty pages at the end of the table and allow the disk space
for the truncated pages to be returned to the operating system.
This parameter, if specified, overrides the vacuum_truncate
reloption. If neither the reloption nor the VACUUM option is
used, the default is true, as before.
Author: Fujii Masao
Reviewed-by: Julien Rouhaud, Masahiko Sawada
Discussion: https://postgr.es/m/CAD21AoD+qtrSDL=GSma4Wd3kLYLeRC0hPna-YAdkDeV4z156vg@mail.gmail.com
This feature was using a process local map to track the first few blocks
in the relation. The map was reset each time we get the block with enough
freespace. It was discussed that it would be better to track this map on
a per-relation basis in relcache and then invalidate the same whenever
vacuum frees up some space in the page or when FSM is created. The new
design would be better both in terms of API design and performance.
List of commits reverted, in reverse chronological order:
06c8a5090e Improve code comments in b0eaa4c51b.
13e8643bfc During pg_upgrade, conditionally skip transfer of FSMs.
6f918159a9 Add more tests for FSM.
9c32e4c350 Clear the local map when not used.
29d108cdec Update the documentation for FSM behavior..
08ecdfe7e5 Make FSM test portable.
b0eaa4c51b Avoid creation of the free space map for small heap relations.
Discussion: https://postgr.es/m/20190416180452.3pm6uegx54iitbt5@alap3.anarazel.de
Commit 3f342839 corrected obsolete comments about buffer locks at the
main _bt_insert_parent() call site, but missed similar obsolete comments
above _bt_insert_parent() itself. Both sets of comments were rendered
obsolete by commit 40dae7ec53, which made the nbtree page split
algorithm more robust. Fix the comments that were missed the first time
around now.
In passing, refine a related _bt_insert_parent() comment about
re-finding the parent page to insert new downlink.
Commit d2599ecfcc introduced some contorted, confused code around:
readers would think that it's possible for HeapTupleHeaderGetXmin return
a non-frozen value for some frozen tuples, which would be disastrous.
There's no actual bug, but it seems better to make it clearer.
Per gripe from Tom Lane and Andres Freund.
Discussion: https://postgr.es/m/30116.1555430496@sss.pgh.pa.us
Commit dd299df818, which made heap TID a tiebreaker nbtree index
column, introduced new rules on page space management to make suffix
truncation safe. In general, suffix truncation needs to have a small
amount of extra space available on the new left page when splitting a
leaf page. This is needed in case it turns out that truncation cannot
even "truncate away the heap TID column", resulting in a
larger-than-firstright leaf high key with an explicit heap TID
representation.
Despite all this, CREATE INDEX/nbtsort.c did not account for the
possible need for extra heap TID space on leaf pages when deciding
whether or not a new item could fit on current page. This could lead to
"failed to add item to the index page" errors when CREATE
INDEX/nbtsort.c tried to finish off a leaf page that lacked space for a
larger-than-firstright leaf high key (it only had space for firstright
tuple, which was just short of what was needed following "truncation").
Several conditions needed to be met all at once for CREATE INDEX to
fail. The problem was in the hard limit on what will fit on a page,
which tends to be masked by the soft fillfactor-wise limit. The easiest
way to recreate the problem seems to be a CREATE INDEX on a low
cardinality text column, with tuples that are of non-uniform width,
using a fillfactor of 100.
To fix, bring nbtsort.c in line with nbtsplitloc.c, which already
pessimistically assumes that all leaf page splits will have high keys
that have a heap TID appended.
Reported-By: Andreas Joseph Krogh
Discussion: https://postgr.es/m/VisenaEmail.c5.3ee7fe277d514162.16a6d785bea@tc7-visena
Peter Geoghegan
Fujii Masao
Michael Paquier
Noah Misch
Tom Lane
Thomas Munro
Peter Eisentraut
Robert Haas
Amit Kapila
Andrew Gierth
Alvaro Herrera
Heikki Linnakangas
Alexander Korotkov
Bruce Momjian
Tomas Vondra
Andres Freund
David Rowley
Magnus Hagander