The first failed connection tests the "regular" connections limit, not
the reserved limit.
In the second failed connection, the username doesn't really matter,
but since the previous successful connections used "regress_reserved",
it seems weird to switch back to "regress_regular" for the
expected-to-fail attempt.
Discussion: https://www.postgresql.org/message-id/fd5e9523-78d3-4270-86b2-fd1b1eeb4fc9@iki.fi
Presently, this section lists a couple of parallelized parts of
pg_upgrade and suggests a starting point for setting the --jobs
option. The list of parallelized tasks is not particularly
actionable, and the phrasing for the --jobs recommendation is
confusing to some readers.
This commit attempts to improve this section by eliminating the
list of parallelized tasks and instead highlighting that --jobs is
most useful for clusters with multiple databases or tablespaces.
Additionally, the recommendation for setting --jobs is simplified
to suggest starting with the number of CPU cores.
Reported-by: Magnus Hagander <magnus@hagander.net>
Reviewed-by: Daniel Gustafsson <daniel@yesql.se>
Reviewed-by: Magnus Hagander <magnus@hagander.net>
Discussion: https://postgr.es/m/Z8dBn_5iGLNuYiPo%40nathan
Commit 8f427187db improved performance by remembering relation stats
as native types rather than issuing a new query for each relation.
Using native types is fine for integers like relpages; but reltuples
is floating point. The commit controllled for that complexity by using
setlocale(LC_NUMERIC, "C"). After that, Alexander Lakhin found a
problem in pg_strtof(), fixed in 00d61a08c5.
While we aren't aware of any more problems with that approach, it
seems wise to just use a string the whole way for floating point
values, as Corey's original patch did, and get rid of the
setlocale(). Integers are still converted to native types to avoid
wasting memory.
Co-authored-by: Corey Huinker <corey.huinker@gmail.com>
Discussion: https://postgr.es/m/3049348.1740855411@sss.pgh.pa.us
Discussion: https://postgr.es/m/560cca3781740bd69881bb07e26eb8f65b09792c.camel%40j-davis.com
Per POSIX, a caller of strtol() that wishes to check for errors must
set errno to 0 beforehand. Several places in spell.c neglected that,
so that they risked delivering a false overflow error in case errno
had been ERANGE already. Given the lack of field reports, this case
may be unreachable at present --- but it's surely trouble waiting to
happen, so fix it.
Author: Jacob Brazeal <jacob.brazeal@gmail.com>
Discussion: https://postgr.es/m/CA+COZaBhsq6EromFm+knMJfzK6nTpG23zJ+K2=nfUQQXcj_xcQ@mail.gmail.com
Backpatch-through: 13
Teach parallel nbtree index scans to use an LWLock (not a spinlock) to
protect the scan's shared descriptor state.
Preparation for an upcoming patch that will add skip scan optimizations
to nbtree. That patch will create the need to occasionally allocate
memory while the scan descriptor is locked, while copying datums that
were serialized by another backend.
Author: Peter Geoghegan <pg@bowt.ie>
Reviewed-By: Matthias van de Meent <boekewurm+postgres@gmail.com>
Discussion: https://postgr.es/m/CAH2-Wz=PKR6rB7qbx+Vnd7eqeB5VTcrW=iJvAsTsKbdG+kW_UA@mail.gmail.com
The query introduced in 8b532771a0 is proving to have ordering issues
under at least the locale cs_CZ. This commit updates the query to use a
stricter ordering.
Per reports from buildfarm members hippopotamus and jay.
pg_stat_io returns a set of tuples based on a combination of three
properties (BackendType, IOObject and IOContext) and
pgstat_tracks_io_object() to decide if a BackendType should return a
tuple based on a pair made of an IOObject and an IOContext.
This commit adds a regression test to track all the combinations
supported. This is useful to know which tuples are relevant when adding
a new BackendType to the set or when touching pgstat_tracks_io_object(),
and I have noticed while playing with this area that it is not
complicated to break it without the regression tests noticing a
difference in some cases.
Reviewed-by: Bertrand Drouvot <bertranddrouvot.pg@gmail.com>
Discussion: https://postgr.es/m/Z8exfAehbVbEKXW5@paquier.xyz
The callback pgstat_backend_have_pending_cb() is used as a way for
pg_stat_report() to detect if there is any pending data for backend
statistics.
It did not include a check based on pgstat_tracks_backend_bktype(), that
discards processes whose backend types do not support backend
statistics. The logic is not a problem on HEAD, as processes that do
not support backend statistics cannot touch PendingBackendStats, so the
callback would always report that there is no pending data in this case.
However, we would run into trouble once backend statistics include
portions of pending stats that are not always zeroed, like pgWalUsage.
There is no reason for pgstat_backend_have_pending_cb() to not check
for pgstat_tracks_backend_bktype(), anyway, and this pattern is safer in
the long run, so let's update the code to do so.
While on it, this commit adds a proper initialization to
PendingBackendStats.
Author: Bertrand Drouvot <bertranddrouvot.pg@gmail.com>
Co-authored-by: Michael Paquier <michael@paquier.xyz>
Discussion: https://postgr.es/m/Z8l6EMM4ImVoWRkg@ip-10-97-1-34.eu-west-3.compute.internal
Some BF animals use very high timeouts due to their slowness. Unfortunately
postmaster/003_start_stop fails if a high timeout is configured, due to
authentication_timeout having a fairly low max.
As this test is reasonably fast, the easiest fix seems to be to cap the
timeout to 600.
Per buildfarm animal skink.
Reviewed-by: Heikki Linnakangas <hlinnaka@iki.fi>
Discussion: https://postgr.es/m/ggflhkciwdyotpoie323chu2c2idpjk5qimrn462encwx2io7s@thmcxl7i6dpw
The test occasionally failed due to unexpected connection limit errors being
encountered after having waited for FATAL errors on another connection. These
spurious failures were caused by the the backend reporting FATAL errors to the
client before detaching from the PGPROC entry. Adding a sleep(1) before
proc_exit() makes it easy to reproduce that problem.
To fix the issue, add a helper function that waits for postmaster to notice
the process having exited. For now this is implemented by waiting for the
DEBUG2 message that postmaster logs in that case. That's not the prettiest
fix, but simple. If we notice this problem elsewhere, it might be worthwhile
to make this more general, e.g. by adding an injection point.
Reported-by: Tomas Vondra <tomas@vondra.me>
Diagnosed-by: Heikki Linnakangas <hlinnaka@iki.fi>
Reviewed-by: Heikki Linnakangas <hlinnaka@iki.fi>
Tested-by: Tomas Vondra <tomas@vondra.me>
Discussion: https://postgr.es/m/ggflhkciwdyotpoie323chu2c2idpjk5qimrn462encwx2io7s@thmcxl7i6dpw
This commit adds a new --missing-only option that can be used in
conjunction with --analyze-only and --analyze-in-stages. When this
option is specified, vacuumdb will generate ANALYZE commands for a
relation if it is missing any statistics it should ordinarily have.
For example, if a table has statistics for one column but not
another, we will analyze the whole table. A similar principle
applies to extended statistics, expression indexes, and table
inheritance.
Co-authored-by: Corey Huinker <corey.huinker@gmail.com>
Reviewed-by: TODO
Discussion: https://postgr.es/m/Z5O1bpcwDrMgyrYy%40nathan
The authentication test added in c76db55c90 expects a backend to start
and wait at the injection point "init-pre-auth". A query is used to
retrieve the PID of the backend waiting at authentication, but its WHERE
clause was too soft, checking only for a backend in a "starting" state.
As proved by the CI, this WHERE clause is not enough. There is a small
window between the moment when the backend is reported as "starting" in
its backend entry and the moment when it waits in its injection point,
and it was possible for the test to return the PID of a backend process
not yet waiting in the injection point, causing spurious failures. This
issue is fixed by tweaking the query retrieving the PID of the backend
waiting before authentication so as we check for "init-pre-auth" in its
wait_event. An extra check based on the backend_type is added, based on
a suggestion by Jacob, to be more cautious.
Error spotted by the CI on Windows, but it could happen anywhere, as
long as the authentication path is slow enough compared to the TAP test.
Reported-by: Andres Freund <andres@anarazel.de>
Author: Jacob Champion <jacob.champion@enterprisedb.com>
Co-authored-by: Michael Paquier <michael@paquier.xyz>
Discussion: https://postgr.es/m/soexrl7oeyku24bj3czupxmv27ow35u6edymp5y3oyoysbe2kb@r3tgoos2xp2x
get_parallel_object_list() has no business closing a connection it did
not create. Make things more sensible by closing the connection at the
level where it is created, in reindex_one_database().
Extracted from a larger patch by the same author. However, the patch as
submitted not only was not described as containing this change, but in
addition it contained a fatal flaw whereby reindexdb would crash and
fail across all of its TAP test, which is why I list myself as
co-author.
Author: Ranier Vilela <ranier.vf@gmail.com>
Author: Álvaro Herrera <alvherre@alvh.no-ip.org>
Discussion: https://postgr.es/m/CAEudQArfqr0-s0VVPSEh=0kgOgBJvFNdGW=xSL5rBcr0WDMQYQ@mail.gmail.com
If a GIN index search had a lot of search keys (for example,
"jsonbcol ?| array[]" with tens of thousands of array elements),
both ginFillScanKey() and startScanKey() took O(N^2) time.
Worse, those loops were uncancelable for lack of CHECK_FOR_INTERRUPTS.
The problem in ginFillScanKey() is the brute-force search key
de-duplication done in ginFillScanEntry(). The most expedient
solution seems to be to just stop trying to de-duplicate once
there are "too many" search keys. We could imagine working harder,
say by using a sort-and-unique algorithm instead of brute force
compare-all-the-keys. But it seems unlikely to be worth the trouble.
There is no correctness issue here, since the code already allowed
duplicate keys if any extra_data is present.
The problem in startScanKey() is the loop that attempts to identify
the first non-required search key. In the submitted test case, that
vainly tests all the key positions, and each iteration takes O(N)
time. One part of that is that it's reinitializing the entryRes[]
array from scratch each time, which is entirely unnecessary given
that the triConsistentFn isn't supposed to scribble on its input.
We can easily adjust the array contents incrementally instead.
The other part of it is that the triConsistentFn may itself take
O(N) time (and does in this test case). This is all extremely
brute force: in simple cases with AND or OR semantics, we could
know without any looping whatever that all or none of the keys
are required. But GIN opclasses don't have any API for exposing
that knowledge, so at least in the short run there is little to
be done about that. Put in a CHECK_FOR_INTERRUPTS so that at
least the loop is cancelable.
These two changes together resolve the primary complaint that
the test query doesn't respond promptly to cancel interrupts.
Also, while they don't completely eliminate the O(N^2) behavior,
they do provide quite a nice speedup for mid-sized examples.
Bug: #18831
Reported-by: Niek <niek.brasa@hitachienergy.com>
Author: Tom Lane <tgl@sss.pgh.pa.us>
Discussion: https://postgr.es/m/18831-e845ac44ebc5dd36@postgresql.org
Backpatch-through: 13
On change of publication via ALTER PUBLICATION ... SET/ADD/DROP commands,
we were invalidating all the relations present in relation sync cache
maintained by pgoutput. We need to invalidate only the relation entries
that are changed as part of publication DDL.
We have ensured that the publication DDL execution generated the
invalidations required to invalidate impacted relation sync entries in
RelationSyncCache.
This improves the performance by avoiding building the cache entries for
the cases where a publication has many tables but only one of them is
dropped.
Author: Shlok Kyal <shlok.kyal.oss@gmail.com>
Author: Hayato Kuroda <kuroda.hayato@fujitsu.com>
Reviewed-by: Hou Zhijie <houzj.fnst@fujitsu.com>
Reviewed-by: Amit Kapila <amit.kapila16@gmail.com>
Discussion: https://postgr.es/m/OSCPR01MB14966C09AA201EFFA706576A7F5C92@OSCPR01MB14966.jpnprd01.prod.outlook.com
Clang with -Wtypedef-redefinition produced warnings:
src/include/storage/latch.h:122:3: error: redefinition of typedef 'Latch' is a C11 feature [-Werror,-Wtypedef-redefinition]
Per buildfarm
This commit adds two improvements related to the monitoring of WAL
writes for the WAL receiver.
First, write counts and timings are now counted in pg_stat_io for the
WAL receiver. These have been discarded from pg_stat_wal in
ff99918c62 due to performance concerns, related to the fact that we
still relied on an on-disk file for the stats back then, even with
track_wal_io_timing to avoid the overhead of the timestamp calculations.
This implementation is simpler than the original proposal as it is
possible to rely on the APIs of pgstat_io.c to do the job. Like the
fsync and read data, track_wal_io_timing needs to be enabled to track
the timings.
Second, a wait event is added around the pg_pwrite() call in charge of
the writes, using the exiting WAIT_EVENT_WAL_WRITE. This is useful as
the WAL receiver data is tracked in pg_stat_activity.
Reviewed-by: Bertrand Drouvot <bertranddrouvot.pg@gmail.com>
Discussion: https://postgr.es/m/Z8gFnH4o3jBm5BRz@ip-10-97-1-34.eu-west-3.compute.internal
This is in preparation for splitting WaitEventSet related functions to
a separate source file. That will hide the details of WaitEventSet
from WaitLatch, so it must use an exposed function instead of
modifying WaitEventSet->exit_on_postmaster_death directly.
Reviewed-by: Andres Freund <andres@anarazel.de>
Discussion: https://www.postgresql.org/message-id/8a507fb6-df28-49d3-81a5-ede180d7f0fb@iki.fi
Previously, Meson could produce a warning about the use of 'deps' in ecpg:
WARNING: Project targets '>=0.54' but uses a feature introduced in '0.60.0': list.<plus>. The right-hand operand was not a list.
The right-hand operand of 'deps' should be a list. This commit fixes
the warning by wrapping it with square brackets.
This issue was introduced in commit 28f04984f0.
Author: Jacob Champion <jacob.champion@enterprisedb.com>
Discussion: https://postgr.es/m/CAOYmi+ks8wO06Ymxduw2h_eQJ_D4_jHGeyMK0P=p5Q3psnEdMA@mail.gmail.com
The CI images come with libcurl pre-installed since commit a119426
in the pg-vm-images repository so remove the installation commands
from the Cirrus tasks. Installation of libcurl packages was added
in the OAuth patchset which introduced the dependency, a backpatch
is thus not applicable.
Author: Daniel Gustafsson <daniel@yesql.se>
Reviewed-by: Andres Freund <andres@anarazel.de>
Discussion: https://postgr.es/m/8745B9D8-D897-4302-BD4C-FC18F291ECB7@yesql.se
To increase coverage without drastically increasing CI resource usage, we have
different CI tasks test different things (e.g. the linux tasks use
sanitizers). Unfortunately that can create confusing situations where CI
fails on some OS, but not others, without the problem appearing to be platform
dependent.
To, partially, address that, add a comment, prefixed with SPECIAL, to each
task that we use to test in some non-default way.
Reviewed-by: Tom Lane <tgl@sss.pgh.pa.us>
Discussion: https://postgr.es/m/321570.1741195755@sss.pgh.pa.us
A lot of buildfarm animals run with debug_parallel_query=regress, while CI
didn't test that. That lead to the annoying situation of only noticing related
test instabilities after merging changes upstream.
FreeBSD was chosen because it's a relatively fast task. It also tests
debug_write_read_parse_plan_trees etc, which probably is exercised a bit more
heavily with debug_parallel_query=regress.
Discussion: https://postgr.es/m/zbuk4mlov22yfoktf5ub3lwjw2b7ezwphwolbplthepda42int@h6wpvq7orc44
Upgrade to the current stable version. To avoid needing commits like this in
the future, the CI image name now doesn't contain the OS version number
anymore.
Backpatch to all versions with CI support, we don't want to generate CI images
for multiple FreeBSD versions.
Author: Nazir Bilal Yavuz <byavuz81@gmail.com>
Discussion: https://postgr.es/m/CAN55FZ3_P4JJ6tWZafjf-_XbHgG6DQGXhH-y6Yp78_bwBJjcww@mail.gmail.com
Backpatch-through: 15
This reverts commit 5ead85fbc8.
This commit shows test failures with debug_parallel_query=regress. The
underlying issue needs to be debugged, so revert for now.
An additional paramater ("strip_in_arrays") is added to these functions.
It defaults to false. If true, then null array elements are removed as
well as null valued object fields. JSON that just consists of a single
null is not affected.
Author: Florents Tselai <florents.tselai@gmail.com>
Discussion: https://postgr.es/m/4BCECCD5-4F40-4313-9E98-9E16BEB0B01D@gmail.com
Expose the count of index searches/index descents in EXPLAIN ANALYZE's
output for index scan nodes. This information is particularly useful
with scans that use ScalarArrayOp quals, where the number of index scans
isn't predictable in advance (at least not with optimizations like the
one added to nbtree by Postgres 17 commit 5bf748b8). It will also be
useful when EXPLAIN ANALYZE shows details of an nbtree index scan that
uses skip scan optimizations set to be introduced by an upcoming patch.
The instrumentation works by teaching index AMs to increment a new
nsearches counter whenever a new index search begins. The counter is
incremented at exactly the same point that index AMs must already
increment the index's pg_stat_*_indexes.idx_scan counter (we're counting
the same event, but at the scan level rather than the relation level).
The new counter is stored in the scan descriptor (IndexScanDescData),
which explain.c reaches by going through the scan node's PlanState.
This approach doesn't match the approach used when tracking other index
scan specific costs (e.g., "Rows Removed by Filter:"). It is similar to
the approach used in other cases where we must track costs that are only
readily accessible inside an access method, and not from the executor
(e.g., "Heap Blocks:" output for a Bitmap Heap Scan). It is inherently
necessary to maintain a counter that can be incremented multiple times
during a single amgettuple call (or amgetbitmap call), and directly
exposing PlanState.instrument to index access methods seems unappealing.
Author: Peter Geoghegan <pg@bowt.ie>
Reviewed-By: Tomas Vondra <tomas@vondra.me>
Reviewed-By: Robert Haas <robertmhaas@gmail.com>
Reviewed-By: Masahiro Ikeda <ikedamsh@oss.nttdata.com>
Reviewed-By: Matthias van de Meent <boekewurm+postgres@gmail.com>
Discussion: https://postgr.es/m/CAH2-Wz=PKR6rB7qbx+Vnd7eqeB5VTcrW=iJvAsTsKbdG+kW_UA@mail.gmail.com
Discussion: https://postgr.es/m/CAH2-WzkRqvaqR2CTNqTZP0z6FuL4-3ED6eQB0yx38XBNj1v-4Q@mail.gmail.com
The usual pattern for handling a signal is that the signal handler
sets a flag and calls SetLatch(MyLatch), and CHECK_FOR_INTERRUPTS() or
other code that is part of a wait loop calls another function to deal
with it. The naming of the functions involved was a bit inconsistent,
however. CHECK_FOR_INTERRUPTS() calls ProcessInterrupts() to do the
heavy-lifting, but the analogous functions in aux processes were
called HandleMainLoopInterrupts(), HandleStartupProcInterrupts(),
etc. Similarly, most subroutines of ProcessInterrupts() were called
Process*(), but some were called Handle*().
To make things less confusing, rename all the functions that are part
of the overall signal/interrupt handling system but are not executed
in a signal handler to e.g. ProcessSomething(), rather than
HandleSomething(). The "Process" prefix is now consistently used in
the non-signal-handler functions, and the "Handle" prefix in functions
that are part of signal handlers, except for some completely unrelated
functions that clearly have nothing to do with signal or interrupt
handling.
Reviewed-by: Nathan Bossart <nathandbossart@gmail.com>
Discussion: https://www.postgresql.org/message-id/8a384b26-1499-41f6-be33-64b801fb98b8@iki.fi
