Maxim Orlov reported that the shutdown of standby server could result in
the following assertion failure. The cause of this issue was that,
when the shutdown caused the startup process to exit, recovery-time
transaction tracking was not shut down even if it's already initialized,
and some locks the tracked transactions were holding could not be released.
At this situation, if other process was invoked and the PGPROC entry that
the startup process used was assigned to it, it found such unreleased locks
and caused the assertion failure, during the initialization of it.
TRAP: FailedAssertion("SHMQueueEmpty(&(MyProc->myProcLocks[i]))"
This commit fixes this issue by making the startup process shut down
transaction tracking and release all locks, at the exit of it.
Back-patch to all supported branches.
Reported-by: Maxim Orlov
Author: Fujii Masao
Reviewed-by: Maxim Orlov
Discussion: https://postgr.es/m/ad4ce692cc1d89a093b471ab1d969b0b@postgrespro.ru
pgstat_report_wait_start() and pgstat_report_wait_end() required two
conditional branches so far. One to check if MyProc is NULL, the other to
check if pgstat_track_activities is set. As wait events are used around
comparatively lightweight operations, and are inlined (reducing branch
predictor effectiveness), that's not great.
The dependency on MyProc has a second disadvantage: Low-level subsystems, like
storage/file/fd.c, report wait events, but architecturally it is preferable
for them to not depend on inter-process subsystems like proc.h (defining
PGPROC). After this change including pgstat.h (nor obviously its
sub-components like backend_status.h, wait_event.h, ...) does not pull in IPC
related headers anymore.
These goals, efficiency and abstraction, are achieved by having
pgstat_report_wait_start/end() not interact with MyProc, but instead a new
my_wait_event_info variable. At backend startup it points to a local variable,
removing the need to check for MyProc being NULL. During process
initialization my_wait_event_info is redirected to MyProc->wait_event_info. At
shutdown this is reversed. Because wait event reporting now does not need to
know about where the wait event is stored, it does not need to know about
PGPROC anymore.
The removal of the branch for checking pgstat_track_activities is simpler:
Don't check anymore. The cost due to the branch are often higher than the
store - and even if not, pgstat_track_activities is rarely disabled.
The main motivator to commit this work now is that removing the (indirect)
pgproc.h include from pgstat.h simplifies a patch to move statistics reporting
to shared memory (which still has a chance to get into 14).
Author: Andres Freund <andres@anarazel.de>
Discussion: https://postgr.es/m/20210402194458.2vu324hkk2djq6ce@alap3.anarazel.de
This implements asynchronous execution, which runs multiple parts of a
non-parallel-aware Append concurrently rather than serially to improve
performance when possible. Currently, the only node type that can be
run concurrently is a ForeignScan that is an immediate child of such an
Append. In the case where such ForeignScans access data on different
remote servers, this would run those ForeignScans concurrently, and
overlap the remote operations to be performed simultaneously, so it'll
improve the performance especially when the operations involve
time-consuming ones such as remote join and remote aggregation.
We may extend this to other node types such as joins or aggregates over
ForeignScans in the future.
This also adds the support for postgres_fdw, which is enabled by the
table-level/server-level option "async_capable". The default is false.
Robert Haas, Kyotaro Horiguchi, Thomas Munro, and myself. This commit
is mostly based on the patch proposed by Robert Haas, but also uses
stuff from the patch proposed by Kyotaro Horiguchi and from the patch
proposed by Thomas Munro. Reviewed by Kyotaro Horiguchi, Konstantin
Knizhnik, Andrey Lepikhov, Movead Li, Thomas Munro, Justin Pryzby, and
others.
Discussion: https://postgr.es/m/CA%2BTgmoaXQEt4tZ03FtQhnzeDEMzBck%2BLrni0UWHVVgOTnA6C1w%40mail.gmail.com
Discussion: https://postgr.es/m/CA%2BhUKGLBRyu0rHrDCMC4%3DRn3252gogyp1SjOgG8SEKKZv%3DFwfQ%40mail.gmail.com
Discussion: https://postgr.es/m/20200228.170650.667613673625155850.horikyota.ntt%40gmail.com
This commit is mostly a revert of aaa3aed, that switched the routine
doing the internal renaming of recycled WAL segments to use on Windows a
combination of CreateHardLinkA() plus unlink() instead of rename(). As
reported by several users of Postgres 13, this is causing concurrency
issues when manipulating WAL segments, mostly in the shape of the
following error:
LOG: could not rename file "pg_wal/000000XX000000YY000000ZZ":
Permission denied
This moves back to a logic where a single rename() (well, pgrename() for
Windows) is used. This issue has proved to be hard to hit when I tested
it, facing it only once with an archive_command that was not able to do
its work, so it is environment-sensitive. The reporters of this issue
have been able to confirm that the situation improved once we switched
back to a single rename(). In order to check things, I have provided to
the reporters a patched build based on 13.2 with aaa3aed reverted, to
test if the error goes away, and an unpatched build of 13.2 to test if
the error still showed up (just to make sure that I did not mess up my
build process).
Extra thanks to Fujii Masao for pointing out what looked like the
culprit commit, and to all the reporters for taking the time to test
what I have sent them.
Reported-by: Andrus, Guy Burgess, Yaroslav Pashinsky, Thomas Trenz
Reviewed-by: Tom Lane, Andres Freund
Discussion: https://postgr.es/m/3861ff1e-0923-7838-e826-094cc9bef737@hot.ee
Discussion: https://postgr.es/m/16874-c3eecd319e36a2bf@postgresql.org
Discussion: https://postgr.es/m/095ccf8d-7f58-d928-427c-b17ace23cae6@burgess.co.nz
Discussion: https://postgr.es/m/16927-67c570d968c99567%40postgresql.org
Discussion: https://postgr.es/m/YFBcRbnBiPdGZvfW@paquier.xyz
Backpatch-through: 13
Since commit 2ce439f3 we have opened every file in the data directory
and called fsync() at the start of crash recovery. This can be very
slow if there are many files, leading to field complaints of systems
taking minutes or even hours to begin crash recovery.
Provide an alternative method, for Linux only, where we call syncfs() on
every possibly different filesystem under the data directory. This is
equivalent, but avoids faulting in potentially many inodes from
potentially slow storage.
The new mode comes with some caveats, described in the documentation, so
the default value for the new setting is "fsync", preserving the older
behavior.
Reported-by: Michael Brown <michael.brown@discourse.org>
Reviewed-by: Fujii Masao <masao.fujii@oss.nttdata.com>
Reviewed-by: Paul Guo <guopa@vmware.com>
Reviewed-by: Bruce Momjian <bruce@momjian.us>
Reviewed-by: Justin Pryzby <pryzby@telsasoft.com>
Reviewed-by: David Steele <david@pgmasters.net>
Discussion: https://postgr.es/m/11bc2bb7-ecb5-3ad0-b39f-df632734cd81%40discourse.org
Discussion: https://postgr.es/m/CAEET0ZHGnbXmi8yF3ywsDZvb3m9CbdsGZgfTXscQ6agcbzcZAw%40mail.gmail.com
After a crash of a backend using temporary files, the files used to be
left behind, on the basis that it might be useful for debugging. But we
don't have any reports of anyone actually doing that, and it means the
disk usage may grow over time due to repeated backend failures (possibly
even hitting ENOSPC). So this behavior is a bit unfortunate, and fixing
it required either manual cleanup (deleting files, which is error-prone)
or restart of the instance (i.e. service disruption).
This implements automatic cleanup of temporary files, controled by a new
GUC remove_temp_files_after_crash. By default the files are removed, but
it can be disabled to restore the old behavior if needed.
Author: Euler Taveira
Reviewed-by: Tomas Vondra, Michael Paquier, Anastasia Lubennikova, Thomas Munro
Discussion: https://postgr.es/m/CAH503wDKdYzyq7U-QJqGn%3DGm6XmoK%2B6_6xTJ-Yn5WSvoHLY1Ww%40mail.gmail.com
This commit changes WAL archiver process so that it's treated as
an auxiliary process and can use shared memory. This is an infrastructure
patch required for upcoming shared-memory based stats collector patch
series. These patch series basically need any processes including archiver
that can report the statistics to access to shared memory. Since this patch
itself is useful to simplify the code and when users monitor the status of
archiver, it's committed separately in advance.
This commit simplifies the code for WAL archiving. For example, previously
backends need to signal to archiver via postmaster when they notify
archiver that there are some WAL files to archive. On the other hand,
this commit removes that signal to postmaster and enables backends to
notify archier directly using shared latch.
Also, as the side of this change, the information about archiver process
becomes viewable at pg_stat_activity view.
Author: Kyotaro Horiguchi
Reviewed-by: Andres Freund, Álvaro Herrera, Julien Rouhaud, Tomas Vondra, Arthur Zakirov, Fujii Masao
Discussion: https://postgr.es/m/20180629.173418.190173462.horiguchi.kyotaro@lab.ntt.co.jp
1. Backends waiting for buffer I/O are now interruptible.
2. If something goes wrong in a backend that is currently performing
I/O, waiting backends no longer wake up until that backend reaches
AbortBufferIO() and broadcasts on the CV. Previously, any waiters would
wake up (because the I/O lock was automatically released) and then
busy-loop until AbortBufferIO() cleared BM_IO_IN_PROGRESS.
3. LWLockMinimallyPadded is removed, as it would now be unused.
Author: Robert Haas <robertmhaas@gmail.com>
Reviewed-by: Thomas Munro <thomas.munro@gmail.com>
Reviewed-by: Julien Rouhaud <rjuju123@gmail.com>
Reviewed-by: Tom Lane <tgl@sss.pgh.pa.us> (earlier version, 2016)
Discussion: https://postgr.es/m/CA%2BhUKGJ8nBFrjLuCTuqKN0pd2PQOwj9b_jnsiGFFMDvUxahj_A%40mail.gmail.com
Discussion: https://postgr.es/m/CA+Tgmoaj2aPti0yho7FeEf2qt-JgQPRWb0gci_o1Hfr=C56Xng@mail.gmail.com
Cut down on system calls and other overheads by waiting for SIGURG
explicitly with kqueue instead of using a signal handler and self-pipe.
Affects *BSD and macOS systems.
This leaves only the poll implementation with a signal handler and the
traditional self-pipe trick.
Discussion: https://postgr.es/m/CA+hUKGJjxPDpzBE0a3hyUywBvaZuC89yx3jK9RFZgfv_KHU7gg@mail.gmail.com
Cut down on system calls and other overheads by reading from a signalfd
instead of using a signal handler and self-pipe. Affects Linux sytems,
and possibly others including illumos that implement the Linux epoll and
signalfd interfaces.
Reviewed-by: Andres Freund <andres@anarazel.de>
Discussion: https://postgr.es/m/CA+hUKGJjxPDpzBE0a3hyUywBvaZuC89yx3jK9RFZgfv_KHU7gg@mail.gmail.com
Traditionally, SIGUSR1 has been overloaded for ad-hoc signals,
procsignal.c signals and latch.c wakeups. Move that last use over to a
new dedicated signal. SIGURG is normally used to report out-of-band
socket data, but PostgreSQL doesn't use that facility.
The signal handler is now installed in all postmaster children by
InitializeLatchSupport(). Those wishing to disconnect from it should
call ShutdownLatchSupport().
Future patches will use this separation of signals to avoid the need for
a signal handler on some operating systems.
Discussion: https://postgr.es/m/CA+hUKGJjxPDpzBE0a3hyUywBvaZuC89yx3jK9RFZgfv_KHU7gg@mail.gmail.com
Commit 82ebbeb0 added a workaround for systems with no epoll_create1()
and EPOLL_CLOEXEC. Linux < 2.6.27 and glibc < 2.9 are long gone. Now
seems like a good time to drop the extra code, because otherwise we'd
have to add similar already-dead workaround code to new patches using
XXX_CLOEXEC flags that arrived in the same kernel release.
Reviewed-by: Tom Lane <tgl@sss.pgh.pa.us>
Discussion: https://postgr.es/m/CA%2BhUKGKL_%3DaO%3Dr30N%3Ds9VoDgTqHpRSzePRbA9dkYO7snc7HsxA%40mail.gmail.com
This adds a new executor node named TID Range Scan. The query planner
will generate paths for TID Range scans when quals are discovered on base
relations which search for ranges on the table's ctid column. These
ranges may be open at either end. For example, WHERE ctid >= '(10,0)';
will return all tuples on page 10 and over.
To support this, two new optional callback functions have been added to
table AM. scan_set_tidrange is used to set the scan range to just the
given range of TIDs. scan_getnextslot_tidrange fetches the next tuple
in the given range.
For AMs were scanning ranges of TIDs would not make sense, these functions
can be set to NULL in the TableAmRoutine. The query planner won't
generate TID Range Scan Paths in that case.
Author: Edmund Horner, David Rowley
Reviewed-by: David Rowley, Tomas Vondra, Tom Lane, Andres Freund, Zhihong Yu
Discussion: https://postgr.es/m/CAMyN-kB-nFTkF=VA_JPwFNo08S0d-Yk0F741S2B7LDmYAi8eyA@mail.gmail.com
Otherwise we risk "leaking" deleted pages by making them non-recyclable
indefinitely. Commit 6655a729 did the same thing for deleted pages in
GiST indexes. That work was used as a starting point here.
Stop storing an XID indicating the oldest bpto.xact across all deleted
though unrecycled pages in nbtree metapages. There is no longer any
reason to care about that condition/the oldest XID. It only ever made
sense when wraparound was something _bt_vacuum_needs_cleanup() had to
consider.
The btm_oldest_btpo_xact metapage field has been repurposed and renamed.
It is now btm_last_cleanup_num_delpages, which is used to remember how
many non-recycled deleted pages remain from the last VACUUM (in practice
its value is usually the precise number of pages that were _newly
deleted_ during the specific VACUUM operation that last set the field).
The general idea behind storing btm_last_cleanup_num_delpages is to use
it to give _some_ consideration to non-recycled deleted pages inside
_bt_vacuum_needs_cleanup() -- though never too much. We only really
need to avoid leaving a truly excessive number of deleted pages in an
unrecycled state forever. We only do this to cover certain narrow cases
where no other factor makes VACUUM do a full scan, and yet the index
continues to grow (and so actually misses out on recycling existing
deleted pages).
These metapage changes result in a clear user-visible benefit: We no
longer trigger full index scans during VACUUM operations solely due to
the presence of only 1 or 2 known deleted (though unrecycled) blocks
from a very large index. All that matters now is keeping the costs and
benefits in balance over time.
Fix an issue that has been around since commit 857f9c36, which added the
"skip full scan of index" mechanism (i.e. the _bt_vacuum_needs_cleanup()
logic). The accuracy of btm_last_cleanup_num_heap_tuples accidentally
hinged upon _when_ the source value gets stored. We now always store
btm_last_cleanup_num_heap_tuples in btvacuumcleanup(). This fixes the
issue because IndexVacuumInfo.num_heap_tuples (the source field) is
expected to accurately indicate the state of the table _after_ the
VACUUM completes inside btvacuumcleanup().
A backpatchable fix cannot easily be extracted from this commit. A
targeted fix for the issue will follow in a later commit, though that
won't happen today.
I (pgeoghegan) have chosen to remove any mention of deleted pages in the
documentation of the vacuum_cleanup_index_scale_factor GUC/param, since
the presence of deleted (though unrecycled) pages is no longer of much
concern to users. The vacuum_cleanup_index_scale_factor description in
the docs now seems rather unclear in any case, and it should probably be
rewritten in the near future. Perhaps some passing mention of page
deletion will be added back at the same time.
Bump XLOG_PAGE_MAGIC due to nbtree WAL records using full XIDs now.
Author: Peter Geoghegan <pg@bowt.ie>
Reviewed-By: Masahiko Sawada <sawada.mshk@gmail.com>
Discussion: https://postgr.es/m/CAH2-WznpdHvujGUwYZ8sihX=d5u-tRYhi-F4wnV2uN2zHpMUXw@mail.gmail.com
As envisioned in commit c98763bf51, it is possible for VACUUM to
ignore certain transactions that are executing CREATE INDEX CONCURRENTLY
and REINDEX CONCURRENTLY for the purposes of computing Xmin; that's
because we know those transactions are not going to examine any other
tables, and are not going to execute anything else in the same
transaction. (Only operations on "safe" indexes can be ignored: those
on indexes that are neither partial nor expressional).
This is extremely useful in cases where CIC/RC can run for a very long
time, because that used to be a significant headache for concurrent
vacuuming of other tables.
Reviewed-by: Matthias van de Meent <boekewurm+postgres@gmail.com>
Reviewed-by: Masahiko Sawada <sawada.mshk@gmail.com>
Discussion: https://postgr.es/m/20210115133858.GA18931@alvherre.pgsql
Commit 46d6e5f567 added the atomic variable "waitStart" into PGPROC struct,
to store the time at which wait for lock acquisition started. Previously
this variable was initialized every time each backend started. Instead,
this commit makes postmaster initialize it at the startup, to ensure that
the variable should be initialized before any use of it.
This commit also moves the code to initialize "waitStart" variable for
prepare transaction, from TwoPhaseGetDummyProc() to MarkAsPreparingGuts().
Because MarkAsPreparingGuts() is more proper place to do that since
it initializes other PGPROC variables.
Author: Fujii Masao
Reviewed-by: Atsushi Torikoshi
Discussion: https://postgr.es/m/1df88660-6f08-cc6e-b7e2-f85296a2bdab@oss.nttdata.com
This commit adds new column "waitstart" into pg_locks view. This column
reports the time when the server process started waiting for the lock
if the lock is not held. This information is useful, for example, when
examining the amount of time to wait on a lock by subtracting
"waitstart" in pg_locks from the current time, and identify the lock
that the processes are waiting for very long.
This feature uses the current time obtained for the deadlock timeout
timer as "waitstart" (i.e., the time when this process started waiting
for the lock). Since getting the current time newly can cause overhead,
we reuse the already-obtained time to avoid that overhead.
Note that "waitstart" is updated without holding the lock table's
partition lock, to avoid the overhead by additional lock acquisition.
This can cause "waitstart" in pg_locks to become NULL for a very short
period of time after the wait started even though "granted" is false.
This is OK in practice because we can assume that users are likely to
look at "waitstart" when waiting for the lock for a long time.
The first attempt of this patch (commit 3b733fcd04) caused the buildfarm
member "rorqual" (built with --disable-atomics --disable-spinlocks) to report
the failure of the regression test. It was reverted by commit 890d2182a2.
The cause of this failure was that the atomic variable for "waitstart"
in the dummy process entry created at the end of prepare transaction was
not initialized. This second attempt fixes that issue.
Bump catalog version.
Author: Atsushi Torikoshi
Reviewed-by: Ian Lawrence Barwick, Robert Haas, Justin Pryzby, Fujii Masao
Discussion: https://postgr.es/m/a96013dc51cdc56b2a2b84fa8a16a993@oss.nttdata.com
While cleaning up after a parallel query or parallel index creation that
created temporary files, we could be interrupted by a statement timeout.
The error handling path would then fail to clean up the files when it
ran dsm_detach() again, because the callback was already popped off the
list. Prevent this hazard by holding interrupts while the cleanup code
runs.
Thanks to Heikki Linnakangas for this suggestion, and also to Kyotaro
Horiguchi, Masahiko Sawada, Justin Pryzby and Tom Lane for discussion of
this and earlier ideas on how to fix the problem.
Back-patch to all supported releases.
Reported-by: Justin Pryzby <pryzby@telsasoft.com>
Discussion: https://postgr.es/m/20191212180506.GR2082@telsasoft.com
This commit adds new column "waitstart" into pg_locks view. This column
reports the time when the server process started waiting for the lock
if the lock is not held. This information is useful, for example, when
examining the amount of time to wait on a lock by subtracting
"waitstart" in pg_locks from the current time, and identify the lock
that the processes are waiting for very long.
This feature uses the current time obtained for the deadlock timeout
timer as "waitstart" (i.e., the time when this process started waiting
for the lock). Since getting the current time newly can cause overhead,
we reuse the already-obtained time to avoid that overhead.
Note that "waitstart" is updated without holding the lock table's
partition lock, to avoid the overhead by additional lock acquisition.
This can cause "waitstart" in pg_locks to become NULL for a very short
period of time after the wait started even though "granted" is false.
This is OK in practice because we can assume that users are likely to
look at "waitstart" when waiting for the lock for a long time.
Bump catalog version.
Author: Atsushi Torikoshi
Reviewed-by: Ian Lawrence Barwick, Robert Haas, Justin Pryzby, Fujii Masao
Discussion: https://postgr.es/m/a96013dc51cdc56b2a2b84fa8a16a993@oss.nttdata.com
GlobalVisIsRemovableFullXid() is now GlobalVisCheckRemovableFullXid().
This is consistent with the general convention for FullTransactionId
equivalents of functions that deal with TransactionId values. It now
matches the nearby GlobalVisCheckRemovableXid() function, which performs
the same check for callers that use TransactionId values.
Oversight in commit dc7420c2c9.
Discussion: https://postgr.es/m/CAH2-Wzmes12jFNDcVgpU89Vp=r6uLFrE-MT0fjSWGsE70UiNaA@mail.gmail.com
In a cluster having used CREATE INDEX CONCURRENTLY while having enabled
prepared transactions, queries that use the resulting index can silently
fail to find rows. Fix this for future CREATE INDEX CONCURRENTLY by
making it wait for prepared transactions like it waits for ordinary
transactions. This expands the VirtualTransactionId structure domain to
admit prepared transactions. It may be necessary to reindex to recover
from past occurrences. Back-patch to 9.5 (all supported versions).
Andrey Borodin, reviewed (in earlier versions) by Tom Lane and Michael
Paquier.
Discussion: https://postgr.es/m/2E712143-97F7-4890-B470-4A35142ABC82@yandex-team.ru
Up until now, we've held this lock when performing a checkpoint or
restartpoint, but commit 076a055acf back
in 2004 and commit 7e48b77b1c from 2009,
taken together, have removed all need for this. In the present code,
there's only ever one process entitled to attempt a checkpoint: either
the checkpointer, during normal operation, or the postmaster, during
single-user operation. So, we don't need the lock.
One possible concern in making this change is that it means that
a substantial amount of code where HOLD_INTERRUPTS() was previously
in effect due to the preceding LWLockAcquire() will now be
running without that. This could mean that ProcessInterrupts()
gets called in places from which it didn't before. However, this
seems unlikely to do very much, because the checkpointer doesn't
have any signal mapped to die(), so it's not clear how,
for example, ProcDiePending = true could happen in the first
place. Similarly with ClientConnectionLost and recovery conflicts.
Also, if there are any such problems, we might want to fix them
rather than reverting this, since running lots of code with
interrupt handling suspended is generally bad.
Patch by me, per an inquiry by Amul Sul. Review by Tom Lane
and Michael Paquier.
Discussion: http://postgr.es/m/CAAJ_b97XnBBfYeSREDJorFsyoD1sHgqnNuCi=02mNQBUMnA=FA@mail.gmail.com
Previously, the per-barrier-type functions tasked with absorbing
them were expected to always succeed and never throw an error.
However, that's a bit inconvenient. Further study has revealed that
there are realistic cases where it might not be possible to absorb
a ProcSignalBarrier without terminating the transaction, or even
the whole backend. Similarly, for some barrier types, there might
be other reasons where it's not reasonably possible to absorb the
barrier at certain points in the code, so provide a way for a
per-barrier-type function to reject absorbing the barrier.
Unfortunately, there's still no committed code making use of this
infrastructure; hopefully, we'll get there. :-(
Patch by me, reviewed by Andres Freund and Amul Sul.
Discussion: http://postgr.es/m/20200908182005.xya7wetdh3pndzim@alap3.anarazel.de
Discussion: http://postgr.es/m/CA+Tgmob56Pk1-5aTJdVPCWFHon7me4M96ENpGe9n_R4JUjjhZA@mail.gmail.com
Every core SLRU wraps around. With the exception of pg_notify, the wrap
point can fall in the middle of a page. Account for this in the
PagePrecedes callback specification and in SimpleLruTruncate()'s use of
said callback. Update each callback implementation to fit the new
specification. This changes SerialPagePrecedesLogically() from the
style of asyncQueuePagePrecedes() to the style of CLOGPagePrecedes().
(Whereas pg_clog and pg_serial share a key space, pg_serial is nothing
like pg_notify.) The bug fixed here has the same symptoms and user
followup steps as 592a589a04. Back-patch
to 9.5 (all supported versions).
Reviewed by Andrey Borodin and (in earlier versions) by Tom Lane.
Discussion: https://postgr.es/m/20190202083822.GC32531@gust.leadboat.com
macOS's ranlib issued a warning about an empty pread.o file with the
previous arrangement, on systems new enough to require no replacement
functions. Let's go back to using configure's AC_REPLACE_FUNCS system
to build and include each .o in the library only if it's needed, which
requires moving the *v() functions to their own files.
Also move the _with_retry() wrapper to a more permanent home.
Reported-by: Tom Lane <tgl@sss.pgh.pa.us>
Discussion: https://postgr.es/m/1283127.1610554395%40sss.pgh.pa.us
This is a follow-up of the work done in commit 0650ff2303. This commit
extends log_recovery_conflict_waits so that a log message is produced
also when recovery conflict has already been resolved after deadlock_timeout
passes, i.e., when the startup process finishes waiting for recovery
conflict after deadlock_timeout. This is useful in investigating how long
recovery conflicts prevented the recovery from applying WAL.
Author: Fujii Masao
Reviewed-by: Kyotaro Horiguchi, Bertrand Drouvot
Discussion: https://postgr.es/m/9a60178c-a853-1440-2cdc-c3af916cff59@amazon.com
Similar to commit d6ad34f341, this patch optimizes
DropRelFileNodesAllBuffers() by avoiding the complete buffer pool scan and
instead find the buffers to be invalidated by doing lookups in the
BufMapping table.
This optimization helps operations where the relation files need to be
removed like Truncate, Drop, Abort of Create Table, etc.
Author: Kirk Jamison
Reviewed-by: Kyotaro Horiguchi, Takayuki Tsunakawa, and Amit Kapila
Tested-By: Haiying Tang
Discussion: https://postgr.es/m/OSBPR01MB3207DCA7EC725FDD661B3EDAEF660@OSBPR01MB3207.jpnprd01.prod.outlook.com
The recovery path of DropRelFileNodeBuffers() is optimized so that
scanning of the whole buffer pool can be avoided when the number of
blocks to be truncated in a relation is below a certain threshold. For
such cases, we find the buffers by doing lookups in BufMapping table.
This improves the performance by more than 100 times in many cases
when several small tables (tested with 1000 relations) are truncated
and where the server is configured with a large value of shared
buffers (greater than equal to 100GB).
This optimization helps cases (a) when vacuum or autovacuum truncated off
any of the empty pages at the end of a relation, or (b) when the relation is
truncated in the same transaction in which it was created.
This commit introduces a new API smgrnblocks_cached which returns a cached
value for the number of blocks in a relation fork. This helps us to determine
the exact size of relation which is required to apply this optimization. The
exact size is required to ensure that we don't leave any buffer for the
relation being dropped as otherwise the background writer or checkpointer
can lead to a PANIC error while flushing buffers corresponding to files that
don't exist.
Author: Kirk Jamison based on ideas by Amit Kapila
Reviewed-by: Kyotaro Horiguchi, Takayuki Tsunakawa, and Amit Kapila
Tested-By: Haiying Tang
Discussion: https://postgr.es/m/OSBPR01MB3207DCA7EC725FDD661B3EDAEF660@OSBPR01MB3207.jpnprd01.prod.outlook.com
This commit adds GUC log_recovery_conflict_waits that controls whether
a log message is produced when the startup process is waiting longer than
deadlock_timeout for recovery conflicts. This is useful in determining
if recovery conflicts prevent the recovery from applying WAL.
Note that currently a log message is produced only when recovery conflict
has not been resolved yet even after deadlock_timeout passes, i.e.,
only when the startup process is still waiting for recovery conflict
even after deadlock_timeout.
Author: Bertrand Drouvot, Masahiko Sawada
Reviewed-by: Alvaro Herrera, Kyotaro Horiguchi, Fujii Masao
Discussion: https://postgr.es/m/9a60178c-a853-1440-2cdc-c3af916cff59@amazon.com
This GUC variable works much like idle_in_transaction_session_timeout,
in that it kills sessions that have waited too long for a new client
query. But it applies when we're not in a transaction, rather than
when we are.
Li Japin, reviewed by David Johnston and Hayato Kuroda, some
fixes by me
Discussion: https://postgr.es/m/763A0689-F189-459E-946F-F0EC4458980B@hotmail.com
The deadlocks that the recovery conflict on lock is involved in can
happen between hot-standby backends and the startup process.
If a backend takes an access exclusive lock on the table and which
finally triggers the deadlock, that deadlock can be detected
as expected. On the other hand, previously, if the startup process
took an access exclusive lock and which finally triggered the deadlock,
that deadlock could not be detected and could remain even after
deadlock_timeout passed. This is a bug.
The cause of this bug was that the code for handling the recovery
conflict on lock didn't take care of deadlock case at all. It assumed
that deadlocks involving the startup process and backends were able
to be detected by the deadlock detector invoked within backends.
But this assumption was incorrect. The startup process also should
have invoked the deadlock detector if necessary.
To fix this bug, this commit makes the startup process invoke
the deadlock detector if deadlock_timeout is reached while handling
the recovery conflict on lock. Specifically, in that case, the startup
process requests all the backends holding the conflicting locks to
check themselves for deadlocks.
Back-patch to v9.6. v9.5 has also this bug, but per discussion we decided
not to back-patch the fix to v9.5. Because v9.5 doesn't have some
infrastructure codes (e.g., 37c54863cf) that this bug fix patch depends on.
We can apply those codes for the back-patch, but since the next minor
version release is the final one for v9.5, it's risky to do that. If we
unexpectedly introduce new bug to v9.5 by the back-patch, there is no
chance to fix that. We determined that the back-patch to v9.5 would give
more risk than gain.
Author: Fujii Masao
Reviewed-by: Bertrand Drouvot, Masahiko Sawada, Kyotaro Horiguchi
Discussion: https://postgr.es/m/4041d6b6-cf24-a120-36fa-1294220f8243@oss.nttdata.com
The logic for determining the latest removed XID for the purposes of
generating recovery conflicts in REDO routines was subtly broken. It
failed to follow links from HOT chains, and so failed to consider all
relevant heap tuple headers in some cases.
To fix, expand the loop that deals with LP_REDIRECT line pointers to
also deal with HOT chains. The new version of the loop is loosely based
on a similar loop from heap_prune_chain().
The impact of this bug is probably quite limited, since the horizon code
necessarily deals with heap tuples that are pointed to by LP_DEAD-set
index tuples. The process of setting LP_DEAD index tuples (e.g. within
the kill_prior_tuple mechanism) is highly correlated with opportunistic
pruning of pointed-to heap tuples. Plus the question of generating a
recovery conflict usually comes up some time after index tuple LP_DEAD
bits were initially set, unlike heap pruning, where a latestRemovedXid
is generated at the point of the pruning operation (heap pruning has no
deferred "would-be page split" style processing that produces conflicts
lazily).
Only backpatch to Postgres 12, the first version where this logic runs
during original execution (following commit 558a9165e0). The index
latestRemovedXid mechanism has had the same bug since it first appeared
over 10 years ago (in commit a760893d), but backpatching to all
supported versions now seems like a bad idea on balance. Running the
new improved code during recovery seems risky, especially given the lack
of complaints from the field.
Author: Peter Geoghegan <pg@bowt.ie>
Discussion: https://postgr.es/m/CAH2-Wz=Eib393+HHcERK_9MtgNS7Ew1HY=RDC_g6GL46zM5C6Q@mail.gmail.com
Backpatch: 12-
The patch needs test cases, reorganization, and cfbot testing.
Technically reverts commits 5c31afc49d..e35b2bad1a (exclusive/inclusive)
and 08db7c63f3..ccbe34139b.
Reported-by: Tom Lane, Michael Paquier
Discussion: https://postgr.es/m/E1ktAAG-0002V2-VB@gemulon.postgresql.org
This adds a key management system that stores (currently) two data
encryption keys of length 128, 192, or 256 bits. The data keys are
AES256 encrypted using a key encryption key, and validated via GCM
cipher mode. A command to obtain the key encryption key must be
specified at initdb time, and will be run at every database server
start. New parameters allow a file descriptor open to the terminal to
be passed. pg_upgrade support has also been added.
Discussion: https://postgr.es/m/CA+fd4k7q5o6Nc_AaX6BcYM9yqTbC6_pnH-6nSD=54Zp6NBQTCQ@mail.gmail.com
Discussion: https://postgr.es/m/20201202213814.GG20285@momjian.us
Author: Masahiko Sawada, me, Stephen Frost
Up to now, if the DBA issued "pg_ctl stop -m immediate", the message
sent to clients was the same as for a crash-and-restart situation.
This is confusing, not least because the message claims that the
database will soon be up again, something we have no business
predicting.
Improve things so that we can generate distinct messages for the two
cases (and also recognize an ad-hoc SIGQUIT, should somebody try that).
To do that, add a field to pmsignal.c's shared memory data structure
that the postmaster sets just before broadcasting SIGQUIT to its
children. No interlocking seems to be necessary; the intervening
signal-sending and signal-receipt should sufficiently serialize accesses
to the field. Hence, this isn't any riskier than the existing usages
of pmsignal.c.
We might in future extend this idea to improve other
postmaster-to-children signal scenarios, although none of them
currently seem to be as badly overloaded as SIGQUIT.
Discussion: https://postgr.es/m/559291.1608587013@sss.pgh.pa.us
Revert ac22929a26, as well as the followup fix 113d3591b8. Because it broke
the assumption that the startup process waiting for the recovery conflict
on buffer pin should be waken up only by buffer unpin or the timeout enabled
in ResolveRecoveryConflictWithBufferPin(). It caused, for example,
SIGHUP signal handler or walreceiver process to wake that startup process
up unnecessarily frequently.
Additionally, add the comments about why that dedicated latch that
the reverted patch tried to get rid of should not be removed.
Thanks to Kyotaro Horiguchi for the discussion.
Author: Fujii Masao
Discussion: https://postgr.es/m/d8c0c608-021b-3c73-fffd-3240829ee986@oss.nttdata.com
Invent a new flag bit HASH_STRINGS to specify C-string hashing, which
was formerly the default; and add assertions insisting that exactly
one of the bits HASH_STRINGS, HASH_BLOBS, and HASH_FUNCTION be set.
This is in hopes of preventing recurrences of the type of oversight
fixed in commit a1b8aa1e4 (i.e., mistakenly omitting HASH_BLOBS).
Also, when HASH_STRINGS is specified, insist that the keysize be
more than 8 bytes. This is a heuristic, but it should catch
accidental use of HASH_STRINGS for integer or pointer keys.
(Nearly all existing use-cases set the keysize to NAMEDATALEN or
more, so there's little reason to think this restriction should
be problematic.)
Tweak hash_create() to insist that the HASH_ELEM flag be set, and
remove the defaults it had for keysize and entrysize. Since those
defaults were undocumented and basically useless, no callers
omitted HASH_ELEM anyway.
Also, remove memset's zeroing the HASHCTL parameter struct from
those callers that had one. This has never been really necessary,
and while it wasn't a bad coding convention it was confusing that
some callers did it and some did not. We might as well save a few
cycles by standardizing on "not".
Also improve the documentation for hash_create().
In passing, improve reinit.c's usage of a hash table by storing
the key as a binary Oid rather than a string; and, since that's
a temporary hash table, allocate it in CurrentMemoryContext for
neatness.
Discussion: https://postgr.es/m/590625.1607878171@sss.pgh.pa.us
User-visible log messages should go through ereport(), so they are
subject to translation. Many remaining elog(LOG) calls are really
debugging calls.
Reviewed-by: Alvaro Herrera <alvherre@alvh.no-ip.org>
Reviewed-by: Michael Paquier <michael@paquier.xyz>
Reviewed-by: Noah Misch <noah@leadboat.com>
Discussion: https://www.postgresql.org/message-id/flat/92d6f545-5102-65d8-3c87-489f71ea0a37%40enterprisedb.com
When committing a transaction that dropped a relation, we previously
truncated only the first segment file to free up disk space (the one
that won't be unlinked until the next checkpoint).
Truncate higher numbered segments too, even though we unlink them on
commit. This frees the disk space immediately, even if other backends
have open file descriptors and might take a long time to get around to
handling shared invalidation events and closing them. Also extend the
same behavior to the first segment, in recovery.
Back-patch to all supported releases.
Bug: #16663
Reported-by: Denis Patron <denis.patron@previnet.it>
Reviewed-by: Pavel Borisov <pashkin.elfe@gmail.com>
Reviewed-by: Neil Chen <carpenter.nail.cz@gmail.com>
Reviewed-by: David Zhang <david.zhang@highgo.ca>
Discussion: https://postgr.es/m/16663-fe97ccf9932fc800%40postgresql.org
Reverts 27838981be (some comments are kept). Per discussion, it does
not seem safe to relax the lock level used for this; in order for it to
be safe, there would have to be memory barriers between the point we set
the flag and the point we set the trasaction Xid, which perhaps would
not be so bad; but there would also have to be barriers at the readers'
side, which from a performance perspective might be bad.
Now maybe this analysis is wrong and it *is* safe for some reason, but
proof of that is not trivial.
Discussion: https://postgr.es/m/20201118190928.vnztes7c2sldu43a@alap3.anarazel.de
While ereport() and elog() themselves are quite cheap when the
error message level is too low to be printed, some places need to do
substantial work before they can call those macros at all. To allow
optimizing away such setup work when nothing is to be printed, make
elog.c export a new function message_level_is_interesting(elevel)
that reports whether ereport/elog will do anything. Make use of that
in various places that had ad-hoc direct tests of log_min_messages etc.
Also teach ProcSleep to use it to avoid some work. (There may well
be other places that could usefully use this; I didn't search hard.)
Within elog.c, refactor a little bit to avoid having duplicate copies
of the policy-setting logic. When that code was written, we weren't
relying on the availability of inline functions; so it had some
duplications in the name of efficiency, which I got rid of.
Alvaro Herrera and Tom Lane
Discussion: https://postgr.es/m/129515.1606166429@sss.pgh.pa.us
While cancelling an autovacuum worker, we hold ProcArrayLock while
formatting a debugging log string. We can make this shorter by saving
the data we need to produce the message and doing the formatting outside
the locked region.
This isn't terribly critical, as it only occurs pretty rarely: when a
backend runs deadlock detection and it happens to be blocked by a
autovacuum running autovacuum. Still, there's no need to cause a hiccup
in ProcArrayLock processing, which can be very high-traffic in some
cases.
While at it, rework code so that we only print the string when it is
really going to be used, as suggested by Michael Paquier.
Discussion: https://postgr.es/m/20201118214127.GA3179@alvherre.pgsql
Reviewed-by: Michael Paquier <michael@paquier.xyz>
We don't actually need a lock to set PGPROC->statusFlags itself; what we
do need is a shared lock on either XidGenLock or ProcArrayLock in order to
ensure MyProc->pgxactoff keeps still while we modify the mirror array in
ProcGlobal->statusFlags. Some places were using an exclusive lock for
that, which is excessive. Relax those to use shared lock only.
procarray.c has a couple of places with somewhat brittle assumptions
about PGPROC changes: ProcArrayEndTransaction uses only shared lock, so
it's permissible to change MyProc only. On the other hand,
ProcArrayEndTransactionInternal also changes other procs, so it must
hold exclusive lock. Add asserts to ensure those assumptions continue
to hold.
Author: Álvaro Herrera <alvherre@alvh.no-ip.org>
Reviewed-by: Michael Paquier <michael@paquier.xyz>
Discussion: https://postgr.es/m/20201117155501.GA13805@alvherre.pgsql
Otherwise, if FDDEBUG is enabled, the debugging output fails because
it tries to read the fileName, which isn't set up yet (and should in
fact always be NULL).
AFAICT, this has been wrong since Berkeley. Before 96bf88d52,
it would accidentally fail to crash on platforms where snprintf()
is forgiving about being passed a NULL pointer for %s; but the
file name intended to be included in the debug output wouldn't
ever have shown up.
Report and fix by Greg Nancarrow. Although this is only visibly
broken in custom-made builds, it still seems worth back-patching
to all supported branches, as the FDDEBUG code is pretty useless
as it stands.
Discussion: https://postgr.es/m/CAJcOf-cUDgm9qYtC_B6XrC6MktMPNRby2p61EtSGZKnfotMArw@mail.gmail.com
When ComputeXidHorizons() was called before MyDatabaseOid is set,
e.g. because a dead row in a shared relation is encountered during
InitPostgres(), the horizon for normal tables was computed too
aggressively, ignoring all backends connected to a database.
During subsequent pruning in a data table the too aggressive horizon
could end up still being used, possibly leading to still needed tuples
being removed. Not good.
This is a bug in dc7420c2c9, which the test added in 94bc27b576 made
visible, if run with force_parallel_mode set to regress. In that case
the bug is reliably triggered, because "pruning_query" is run in a
parallel worker and the start of that parallel worker is likely to
encounter a dead row in pg_database.
The fix is trivial: Compute a more pessimistic data table horizon if
MyDatabaseId is not yet known.
Author: Andres Freund
Discussion: https://postgr.es/m/20201029040030.p4osrmaywhqaesd4@alap3.anarazel.de
This fixes a bug in the edge case where, for a temp table, heap_page_prune()
can end up with a different horizon than heap_vacuum_rel(). Which can trigger
errors like "ERROR: cannot freeze committed xmax ...".
The bug was introduced due to interaction of a7212be8b9 "Set cutoff xmin more
aggressively when vacuuming a temporary table." with dc7420c2c9 "snapshot
scalability: Don't compute global horizons while building snapshots.".
The problem is caused by lazy_scan_heap() assuming that the only reason its
HeapTupleSatisfiesVacuum() call would return HEAPTUPLE_DEAD is if the tuple is
a HOT tuple, or if the tuple's inserting transaction has aborted since the
heap_page_prune() call. But after a7212be8b9 that was also possible in other
cases for temp tables, because heap_page_prune() uses a different visibility
test after dc7420c2c9.
The fix is fairly simple: Move the special case logic for temp tables from
vacuum_set_xid_limits() to the infrastructure introduced in dc7420c2c9. That
ensures that the horizon used for pruning is at least as aggressive as the one
used by lazy_scan_heap(). The concrete horizon used for temp tables is
slightly different than the logic in dc7420c2c9, but should always be as
aggressive as before (see comments).
A significant benefit to centralizing the logic procarray.c is that now the
more aggressive horizons for temp tables does not just apply to VACUUM but
also to e.g. HOT pruning and the nbtree killtuples logic.
Because isTopLevel is not needed by vacuum_set_xid_limits() anymore, I
undid the the related changes from a7212be8b9.
This commit also adds an isolation test ensuring that the more aggressive
vacuuming and pruning of temp tables keeps working.
Debugged-By: Amit Kapila <amit.kapila16@gmail.com>
Debugged-By: Tom Lane <tgl@sss.pgh.pa.us>
Debugged-By: Ashutosh Sharma <ashu.coek88@gmail.com>
Author: Andres Freund <andres@anarazel.de>
Discussion: https://postgr.es/m/20201014203103.72oke6hqywcyhx7s@alap3.anarazel.de
Discussion: https://postgr.es/m/20201015083735.derdzysdtqdvxshp@alap3.anarazel.de
CheckBuffer() is designed to be a concurrent-safe function able to run
sanity checks on a relation page without loading it into the shared
buffers. The operation is done using a lock on the partition involved
in the shared buffer mapping hashtable and an I/O lock for the buffer
itself, preventing the risk of false positives due to any concurrent
activity.
The primary use of this function is the detection of on-disk corruptions
for relation pages. If a page is found in shared buffers, the on-disk
page is checked if not dirty (a follow-up checkpoint would flush a valid
version of the page if dirty anyway), as it could be possible that a
page was present for a long time in shared buffers with its on-disk
version corrupted. Such a scenario could lead to a corrupted cluster if
a host is plugged off for example. If the page is not found in shared
buffers, its on-disk state is checked. PageIsVerifiedExtended() is used
to apply the same sanity checks as when a page gets loaded into shared
buffers.
This function will be used by an upcoming patch able to check the state
of on-disk relation pages using a SQL function.
Author: Julien Rouhaud, Michael Paquier
Reviewed-by: Masahiko Sawada
Discussion: https://postgr.es/m/CAOBaU_aVvMjQn=ge5qPiJOPMmOj5=ii3st5Q0Y+WuLML5sR17w@mail.gmail.com
This is useful for checks of relation pages without having to load the
pages into the shared buffers, and two cases can make use of that: page
verification in base backups and the online, lock-safe, flavor.
Compatibility is kept with past versions using a macro that calls the
new extended routine with the set of options compatible with the
original version.
Extracted from a larger patch by the same author.
Author: Anastasia Lubennikova
Reviewed-by: Michael Paquier, Julien Rouhaud
Discussion: https://postgr.es/m/608f3476-0598-2514-2c03-e05c7d2b0cbd@postgrespro.ru
In shm_mq_receive(), a huge payload could trigger an unjustified
"invalid memory alloc request size" error due to the way the buffer
size is increased.
Add error checks (documenting the upper limit) and avoid the error by
limiting the allocation size to MaxAllocSize.
Author: Markus Wanner <markus.wanner@2ndquadrant.com>
Discussion: https://www.postgresql.org/message-id/flat/3bb363e7-ac04-0ac4-9fe8-db1148755bfa%402ndquadrant.com
While registering for postmaster exit events, we have to handle a couple
of edge cases where the postmaster is already gone. Commit 815c2f09
missed one: EACCES must surely imply that PostmasterPid no longer
belongs to our postmaster process (or alternatively an unexpected
permissions model has been imposed on us). Like ESRCH, this should be
treated as a WL_POSTMASTER_DEATH event, rather than being raised with
ereport().
No known problems reported in the wild. Per code review from Tom Lane.
Back-patch to 13.
Reported-by: Tom Lane <tgl@sss.pgh.pa.us>
Discussion: https://postgr.es/m/3624029.1602701929%40sss.pgh.pa.us
If WaitEventSetWait() reports that the postmaster has gone away, later
calls to WaitEventSetWait() should continue to report that. Otherwise
further waits that occur in the proc_exit() path after we already
noticed the postmaster's demise could block forever.
Back-patch to 13, where the kqueue support landed.
Reported-by: Tom Lane <tgl@sss.pgh.pa.us>
Discussion: https://postgr.es/m/3624029.1602701929%40sss.pgh.pa.us
I (Andres) broke this in 623a9CA79bx, because I didn't think about the
way snapshots are built on standbys sufficiently. Unfortunately our
existing tests did not catch this, as they are all just querying with
psql (therefore ending up with fresh snapshots).
The fix is trivial, we just need to increment the transaction
completion counter in ExpireTreeKnownAssignedTransactionIds(), which
is the equivalent of ProcArrayEndTransaction() during recovery.
This commit also adds a new test doing some basic testing of the
correctness of snapshots built on standbys. To avoid the
aforementioned issue of one-shot psql's not exercising the snapshot
caching, the test uses a long lived psqls, similar to
013_crash_restart.pl. It'd be good to extend the test further.
Reported-By: Ian Barwick <ian.barwick@2ndquadrant.com>
Author: Andres Freund <andres@anarazel.de>
Author: Ian Barwick <ian.barwick@2ndquadrant.com>
Discussion: https://postgr.es/m/61291ffe-d611-f889-68b5-c298da9fb18f@2ndquadrant.com
Previously, we called fsync() after writing out individual pg_xact,
pg_multixact and pg_commit_ts pages due to cache pressure, leading to
regular I/O stalls in user backends and recovery. Collapse requests for
the same file into a single system call as part of the next checkpoint,
as we already did for relation files, using the infrastructure developed
by commit 3eb77eba. This can cause a significant improvement to
recovery performance, especially when it's otherwise CPU-bound.
Hoist ProcessSyncRequests() up into CheckPointGuts() to make it clearer
that it applies to all the SLRU mini-buffer-pools as well as the main
buffer pool. Rearrange things so that data collected in CheckpointStats
includes SLRU activity.
Also remove the Shutdown{CLOG,CommitTS,SUBTRANS,MultiXact}() functions,
because they were redundant after the shutdown checkpoint that
immediately precedes them. (I'm not sure if they were ever needed, but
they aren't now.)
Reviewed-by: Tom Lane <tgl@sss.pgh.pa.us> (parts)
Tested-by: Jakub Wartak <Jakub.Wartak@tomtom.com>
Discussion: https://postgr.es/m/CA+hUKGLJ=84YT+NvhkEEDAuUtVHMfQ9i-N7k_o50JmQ6Rpj_OQ@mail.gmail.com
This function could often be seen in profiles of vacuum and could often
be a significant bottleneck during recovery. The problem was that a qsort
was performed in order to sort an array of item pointers in reverse offset
order so that we could use that to safely move tuples up to the end of the
page without overwriting the memory of yet-to-be-moved tuples. i.e. we
used to compact the page starting at the back of the page and move towards
the front. The qsort that this required could be expensive for pages with
a large number of tuples.
In this commit, we take another approach to tuple compactification.
Now, instead of sorting the remaining item pointers array we first check
if the array is presorted and only memmove() the tuples that need to be
moved. This presorted check can be done very cheaply in the calling
functions when the array is being populated. This presorted case is very
fast.
When the item pointer array is not presorted we must copy tuples that need
to be moved into a temp buffer before copying them back into the page
again. This differs from what we used to do here as we're now copying the
tuples back into the page in reverse line pointer order. Previously we
left the existing order alone. Reordering the tuples results in an
increased likelihood of hitting the pre-sorted case the next time around.
Any newly added tuple which consumes a new line pointer will also maintain
the correct sort order of tuples in the page which will also result in the
presorted case being hit the next time. Only consuming an unused line
pointer can cause the order of tuples to go out again, but that will be
corrected next time the function is called for the page.
Benchmarks have shown that the non-presorted case is at least equally as
fast as the original qsort method even when the page just has a few
tuples. As the number of tuples becomes larger the new method maintains
its performance whereas the original qsort method became much slower when
the number of tuples on the page became large.
Author: David Rowley
Reviewed-by: Thomas Munro
Tested-by: Jakub Wartak
Discussion: https://postgr.es/m/CA+hUKGKMQFVpjr106gRhwk6R-nXv0qOcTreZuQzxgpHESAL6dw@mail.gmail.com
Although 58c6feccf fixed the case for SIGQUIT, we were still calling
proc_exit() from signal handlers for SIGTERM and timeout failures in
ProcessStartupPacket. Fortunately, at the point where that code runs,
we haven't yet connected to shared memory in any meaningful way, so
there is nothing we need to undo in shared memory. This means it
should be safe to use _exit(1) here, ie, not run any atexit handlers
but also inform the postmaster that it's not a crash exit.
To make sure nobody breaks the "nothing to undo" expectation, add
a cross-check that no on-shmem-exit or before-shmem-exit handlers
have been registered yet when we finish using these signal handlers.
This change is simple enough that maybe it could be back-patched,
but I won't risk that right now.
Discussion: https://postgr.es/m/1850884.1599601164@sss.pgh.pa.us
Historically, cancel_before_shmem_exit() just silently did nothing
if the specified callback wasn't the top-of-stack. The folly of
ignoring this case was exposed by the bugs fixed in 303640199 and
bab150045, so let's make it throw elog(ERROR) instead.
There is a decent argument to be made that PG_ENSURE_ERROR_CLEANUP
should use some separate infrastructure, so it wouldn't break if
something inside the guarded code decides to register a new
before_shmem_exit callback. However, a survey of the surviving
uses of before_shmem_exit() and PG_ENSURE_ERROR_CLEANUP doesn't
show any plausible conflicts of that sort today, so for now we'll
forgo the extra complexity. (It will almost certainly become
necessary if anyone ever wants to wrap PG_ENSURE_ERROR_CLEANUP
around arbitrary user-defined actions, though.)
No backpatch, since this is developer support not a production issue.
Bharath Rupireddy, per advice from Andres Freund, Robert Haas, and myself
Discussion: https://postgr.es/m/CALj2ACWk7j4F2v2fxxYfrroOF=AdFNPr1WsV+AGtHAFQOqm_pw@mail.gmail.com
The problem is caused by me (Andres) having ProcSleep() look at the
wrong PGPROC entry in 5788e258bb.
Unfortunately it seems hard to write a reliable test for autovacuum
cancellations. Perhaps somebody will come up with a good approach, but
it seems worth fixing the issue even without a test.
Reported-By: Jeff Janes <jeff.janes@gmail.com>
Author: Jeff Janes <jeff.janes@gmail.com>
Discussion: https://postgr.es/m/CAMkU=1wH2aUy+wDRDz+5RZALdcUnEofV1t9PzXS_gBJO9vZZ0Q@mail.gmail.com
Some kernels can tell us the type of a "dirent", so we can avoid a call
to stat() or lstat() in many cases. Define a new function
get_dirent_type() to contain that logic, for use by the backend and
frontend versions of walkdir(), and perhaps other callers in future.
Reviewed-by: Tom Lane <tgl@sss.pgh.pa.us>
Reviewed-by: Juan José Santamaría Flecha <juanjo.santamaria@gmail.com>
Discussion: https://postgr.es/m/CA%2BhUKG%2BFzxupGGN4GpUdbzZN%2Btn6FQPHo8w0Q%2BAPH5Wz8RG%2Bww%40mail.gmail.com
Replace CFLAGS_VECTOR with CFLAGS_UNROLL_LOOPS and CFLAGS_VECTORIZE,
allowing us to distinguish whether we want to apply -funroll-loops,
-ftree-vectorize, or both to a particular source file. Up to now
the only consumer of the symbol has been checksum.c which wants
both, so that there was no need to distinguish; but that's about
to change.
Amit Khandekar, reviewed and edited a little by me
Discussion: https://postgr.es/m/CAJ3gD9evtA_vBo+WMYMyT-u=keHX7-r8p2w7OSRfXf42LTwCZQ@mail.gmail.com
Commit 808e13b282 introduced a few APIs to extend the existing Buffile
interface. In SharedFileSetDeleteOnProcExit, it tries to delete the list
element while traversing the list with 'foreach' construct which makes the
behavior of list traversal unpredictable.
Author: Amit Kapila
Reviewed-by: Dilip Kumar
Tested-by: Dilip Kumar and Neha Sharma
Discussion: https://postgr.es/m/CAA4eK1JhLatVcQ2OvwA_3s0ih6Hx9+kZbq107cXVsSWWukH7vA@mail.gmail.com
Allow BufFile to support temporary files that can be used by the single
backend when the corresponding files need to be survived across the
transaction and need to be opened and closed multiple times. Such files
need to be created as a member of a SharedFileSet.
Additionally, this commit implements the interface for BufFileTruncate to
allow files to be truncated up to a particular offset and extends the
BufFileSeek API to support the SEEK_END case. This also adds an option to
provide a mode while opening the shared BufFiles instead of always opening
in read-only mode.
These enhancements in BufFile interface are required for the upcoming
patch to allow the replication apply worker, to handle streamed
in-progress transactions.
Author: Dilip Kumar, Amit Kapila
Reviewed-by: Amit Kapila
Tested-by: Neha Sharma
Discussion: https://postgr.es/m/688b0b7f-2f6c-d827-c27b-216a8e3ea700@2ndquadrant.com
This corrects an oversight by me in 2072932407, which made
ProcArrayClearTransaction() increment xactCompletionCount. That requires an
exclusive lock, obviously.
There's other approaches that avoid the exclusive acquisition, but given that a
2PC commit is fairly heavyweight, it doesn't seem worth doing so. I've not been
able to measure a performance difference, unsurprisingly. I did add a
comment documenting that we could do so, should it ever become a bottleneck.
Reported-By: Tom Lane <tgl@sss.pgh.pa.us>
Author: Andres Freund <andres@anarazel.de>
Discussion: https://postgr.es/m/1355915.1597794204@sss.pgh.pa.us
When preparing a transaction xactCompletionCount needs to be
incremented, even though the transaction has not committed
yet. Otherwise the snapshot used within the transaction otherwise can
get reused outside of the prepared transaction. As GetSnapshotData()
does not include the current xid when building a snapshot, reuse would
not be correct.
Somewhat surprisingly the regression tests only rarely show incorrect
results without the fix. The reason for that is that often the
snapshot's xmax will be >= the backend xid, yielding a snapshot that
is correct, despite the bug.
I'm working on a reliable test for the bug, but it seems worth seeing
whether this fixes all the BF failures while I do.
Author: Andres Freund <andres@anarazel.de>
Discussion: https://postgr.es/m/E1k7tGP-0005V0-5k@gemulon.postgresql.org
Previous commits made it faster/more scalable to compute snapshots. But not
building a snapshot is still faster. Now that GetSnapshotData() does not
maintain RecentGlobal* anymore, that is actually not too hard:
This commit introduces xactCompletionCount, which tracks the number of
top-level transactions with xids (i.e. which may have modified the database)
that completed in some form since the start of the server.
We can avoid rebuilding the snapshot's contents whenever the current
xactCompletionCount is the same as it was when the snapshot was
originally built. Currently this check happens while holding
ProcArrayLock. While it's likely possible to perform the check without
acquiring ProcArrayLock, it seems better to do that separately /
later, some careful analysis is required. Even with the lock this is a
significant win on its own.
On a smaller two socket machine this gains another ~1.03x, on a larger
machine the effect is roughly double (earlier patch version tested
though). If we were able to safely avoid the lock there'd be another
significant gain on top of that.
Author: Andres Freund <andres@anarazel.de>
Reviewed-By: Robert Haas <robertmhaas@gmail.com>
Reviewed-By: Thomas Munro <thomas.munro@gmail.com>
Reviewed-By: David Rowley <dgrowleyml@gmail.com>
Discussion: https://postgr.es/m/20200301083601.ews6hz5dduc3w2se@alap3.anarazel.de
This bug, recently introduced in 941697c3c1, at least lead to vacuum
failing because it found tuples inserted by a running transaction, but
below the freeze limit. The freeze limit in turn is directly affected
by the aforementioned bug.
Thanks to Tom Lane figuring how to make the bug reproducible.
We should add a few more assertions to make sure this type of bug
isn't as hard to notice, but it's not yet clear how to best do so.
Co-Diagnosed-By: Tom Lane <tgl@sss.pgh.pa.us>
Author: Andres Freund <andres@anarazel.de>
Discussion: https://postgr.es/m/1013484.1597609043@sss.pgh.pa.us
Reuse cautionary language from src/test/ssl/README in
src/test/kerberos/README. SLRUs have had access to six-character
segments names since commit 73c986adde,
and recovery stopped calling HeapTupleHeaderAdvanceLatestRemovedXid() in
commit 558a9165e0. The other corrections
are more self-evident.
The SimpleLruTruncate() header comment states the new coding rule. To
achieve this, add locktype "frozenid" and two LWLocks. This closes a
rare opportunity for data loss, which manifested as "apparent
wraparound" or "could not access status of transaction" errors. Data
loss is more likely in pg_multixact, due to released branches' thin
margin between multiStopLimit and multiWrapLimit. If a user's physical
replication primary logged ": apparent wraparound" messages, the user
should rebuild standbys of that primary regardless of symptoms. At less
risk is a cluster having emitted "not accepting commands" errors or
"must be vacuumed" warnings at some point. One can test a cluster for
this data loss by running VACUUM FREEZE in every database. Back-patch
to 9.5 (all supported versions).
Discussion: https://postgr.es/m/20190218073103.GA1434723@rfd.leadboat.com
Similar to the previous changes this increases the chance that data
frequently needed by GetSnapshotData() stays in l2 cache. In many
workloads subtransactions are very rare, and this makes the check for
that considerably cheaper.
As this removes the last member of PGXACT, there is no need to keep it
around anymore.
On a larger 2 socket machine this and the two preceding commits result
in a ~1.07x performance increase in read-only pgbench. For read-heavy
mixed r/w workloads without row level contention, I see about 1.1x.
Author: Andres Freund <andres@anarazel.de>
Reviewed-By: Robert Haas <robertmhaas@gmail.com>
Reviewed-By: Thomas Munro <thomas.munro@gmail.com>
Reviewed-By: David Rowley <dgrowleyml@gmail.com>
Discussion: https://postgr.es/m/20200301083601.ews6hz5dduc3w2se@alap3.anarazel.de
Similar to the previous commit this increases the chance that data
frequently needed by GetSnapshotData() stays in l2 cache. As we now
take care to not unnecessarily write to ProcGlobal->vacuumFlags, there
should be very few modifications to the ProcGlobal->vacuumFlags array.
Author: Andres Freund <andres@anarazel.de>
Reviewed-By: Robert Haas <robertmhaas@gmail.com>
Reviewed-By: Thomas Munro <thomas.munro@gmail.com>
Reviewed-By: David Rowley <dgrowleyml@gmail.com>
Discussion: https://postgr.es/m/20200301083601.ews6hz5dduc3w2se@alap3.anarazel.de
The new array contains the xids for all connected backends / in-use
PGPROC entries in a dense manner (in contrast to the PGPROC/PGXACT
arrays which can have unused entries interspersed).
This improves performance because GetSnapshotData() always needs to
scan the xids of all live procarray entries and now there's no need to
go through the procArray->pgprocnos indirection anymore.
As the set of running top-level xids changes rarely, compared to the
number of snapshots taken, this substantially increases the likelihood
of most data required for a snapshot being in l2 cache. In
read-mostly workloads scanning the xids[] array will sufficient to
build a snapshot, as most backends will not have an xid assigned.
To keep the xid array dense ProcArrayRemove() needs to move entries
behind the to-be-removed proc's one further up in the array. Obviously
moving array entries cannot happen while a backend sets it
xid. I.e. locking needs to prevent that array entries are moved while
a backend modifies its xid.
To avoid locking ProcArrayLock in GetNewTransactionId() - a fairly hot
spot already - ProcArrayAdd() / ProcArrayRemove() now needs to hold
XidGenLock in addition to ProcArrayLock. Adding / Removing a procarray
entry is not a very frequent operation, even taking 2PC into account.
Due to the above, the dense array entries can only be read or modified
while holding ProcArrayLock and/or XidGenLock. This prevents a
concurrent ProcArrayRemove() from shifting the dense array while it is
accessed concurrently.
While the new dense array is very good when needing to look at all
xids it is less suitable when accessing a single backend's xid. In
particular it would be problematic to have to acquire a lock to access
a backend's own xid. Therefore a backend's xid is not just stored in
the dense array, but also in PGPROC. This also allows a backend to
only access the shared xid value when the backend had acquired an
xid.
The infrastructure added in this commit will be used for the remaining
PGXACT fields in subsequent commits. They are kept separate to make
review easier.
Author: Andres Freund <andres@anarazel.de>
Reviewed-By: Robert Haas <robertmhaas@gmail.com>
Reviewed-By: Thomas Munro <thomas.munro@gmail.com>
Reviewed-By: David Rowley <dgrowleyml@gmail.com>
Discussion: https://postgr.es/m/20200301083601.ews6hz5dduc3w2se@alap3.anarazel.de
Now that xmin isn't needed for GetSnapshotData() anymore, it leads to
unnecessary cacheline ping-pong to have it in PGXACT, as it is updated
considerably more frequently than the other PGXACT members.
After the changes in dc7420c2c9, this is a very straight-forward change.
For highly concurrent, snapshot acquisition heavy, workloads this change alone
can significantly increase scalability. E.g. plain pgbench on a smaller 2
socket machine gains 1.07x for read-only pgbench, 1.22x for read-only pgbench
when submitting queries in batches of 100, and 2.85x for batches of 100
'SELECT';. The latter numbers are obviously not to be expected in the
real-world, but micro-benchmark the snapshot computation
scalability (previously spending ~80% of the time in GetSnapshotData()).
Author: Andres Freund <andres@anarazel.de>
Reviewed-By: Robert Haas <robertmhaas@gmail.com>
Reviewed-By: Thomas Munro <thomas.munro@gmail.com>
Reviewed-By: David Rowley <dgrowleyml@gmail.com>
Discussion: https://postgr.es/m/20200301083601.ews6hz5dduc3w2se@alap3.anarazel.de
To make GetSnapshotData() more scalable, it cannot not look at at each proc's
xmin: While snapshot contents do not need to change whenever a read-only
transaction commits or a snapshot is released, a proc's xmin is modified in
those cases. The frequency of xmin modifications leads to, particularly on
higher core count systems, many cache misses inside GetSnapshotData(), despite
the data underlying a snapshot not changing. That is the most
significant source of GetSnapshotData() scaling poorly on larger systems.
Without accessing xmins, GetSnapshotData() cannot calculate accurate horizons /
thresholds as it has so far. But we don't really have to: The horizons don't
actually change that much between GetSnapshotData() calls. Nor are the horizons
actually used every time a snapshot is built.
The trick this commit introduces is to delay computation of accurate horizons
until there use and using horizon boundaries to determine whether accurate
horizons need to be computed.
The use of RecentGlobal[Data]Xmin to decide whether a row version could be
removed has been replaces with new GlobalVisTest* functions. These use two
thresholds to determine whether a row can be pruned:
1) definitely_needed, indicating that rows deleted by XIDs >= definitely_needed
are definitely still visible.
2) maybe_needed, indicating that rows deleted by XIDs < maybe_needed can
definitely be removed
GetSnapshotData() updates definitely_needed to be the xmin of the computed
snapshot.
When testing whether a row can be removed (with GlobalVisTestIsRemovableXid())
and the tested XID falls in between the two (i.e. XID >= maybe_needed && XID <
definitely_needed) the boundaries can be recomputed to be more accurate. As it
is not cheap to compute accurate boundaries, we limit the number of times that
happens in short succession. As the boundaries used by
GlobalVisTestIsRemovableXid() are never reset (with maybe_needed updated by
GetSnapshotData()), it is likely that further test can benefit from an earlier
computation of accurate horizons.
To avoid regressing performance when old_snapshot_threshold is set (as that
requires an accurate horizon to be computed), heap_page_prune_opt() doesn't
unconditionally call TransactionIdLimitedForOldSnapshots() anymore. Both the
computation of the limited horizon, and the triggering of errors (with
SetOldSnapshotThresholdTimestamp()) is now only done when necessary to remove
tuples.
This commit just removes the accesses to PGXACT->xmin from
GetSnapshotData(), but other members of PGXACT residing in the same
cache line are accessed. Therefore this in itself does not result in a
significant improvement. Subsequent commits will take advantage of the
fact that GetSnapshotData() now does not need to access xmins anymore.
Note: This contains a workaround in heap_page_prune_opt() to keep the
snapshot_too_old tests working. While that workaround is ugly, the tests
currently are not meaningful, and it seems best to address them separately.
Author: Andres Freund <andres@anarazel.de>
Reviewed-By: Robert Haas <robertmhaas@gmail.com>
Reviewed-By: Thomas Munro <thomas.munro@gmail.com>
Reviewed-By: David Rowley <dgrowleyml@gmail.com>
Discussion: https://postgr.es/m/20200301083601.ews6hz5dduc3w2se@alap3.anarazel.de
The reason for doing so is that a subsequent commit will need that to
avoid wraparound issues. As the subsequent change is large this was
split out for easier review.
The reason this is not a perfect straight-forward change is that we do
not want track 64bit xids in the procarray or the WAL. Therefore we
need to advance lastestCompletedXid in relation to 32 bit xids. The
code for that is now centralized in MaintainLatestCompletedXid*.
Author: Andres Freund
Reviewed-By: Thomas Munro, Robert Haas, David Rowley
Discussion: https://postgr.es/m/20200301083601.ews6hz5dduc3w2se@alap3.anarazel.de
Including Full in variable names duplicates the type information and
leads to overly long names. As FullTransactionId cannot accidentally
be casted to TransactionId that does not seem necessary.
Author: Andres Freund
Discussion: https://postgr.es/m/20200724011143.jccsyvsvymuiqfxu@alap3.anarazel.de
Create an optional region in the main shared memory segment that can be
used to acquire and release "fast" DSM segments, and can benefit from
huge pages allocated at cluster startup time, if configured. Fall back
to the existing mechanisms when that space is full. The size is
controlled by a new GUC min_dynamic_shared_memory, defaulting to 0.
Main region DSM segments initially contain whatever garbage the memory
held last time they were used, rather than zeroes. That change revealed
that DSA areas failed to initialize themselves correctly in memory that
wasn't zeroed first, so fix that problem.
Discussion: https://postgr.es/m/CA%2BhUKGLAE2QBv-WgGp%2BD9P_J-%3Dyne3zof9nfMaqq1h3EGHFXYQ%40mail.gmail.com
Avoid repeatedly calling lseek(SEEK_END) during recovery by caching
the size of each fork. For now, we can't use the same technique in
other processes, because we lack a shared invalidation mechanism.
Do this by generalizing the pre-existing caching used by FSM and VM
to support all forks.
Discussion: https://postgr.es/m/CAEepm%3D3SSw-Ty1DFcK%3D1rU-K6GSzYzfdD4d%2BZwapdN7dTa6%3DnQ%40mail.gmail.com
Create LatchWaitSet at backend startup time, and use it to implement
WaitLatch(). This avoids repeated epoll/kqueue setup and teardown
system calls.
Reorder SubPostmasterMain() slightly so that we restore the postmaster
pipe and Windows signal emulation before we reach InitPostmasterChild(),
to make this work in EXEC_BACKEND builds.
Reviewed-by: Kyotaro Horiguchi <horikyota.ntt@gmail.com>
Discussion: https://postgr.es/m/CA%2BhUKGJAC4Oqao%3DqforhNey20J8CiG2R%3DoBPqvfR0vOJrFysGw%40mail.gmail.com
Holding just a buffer pin (with no buffer lock) on an nbtree buffer/page
provides very weak guarantees, especially compared to heapam, where it's
often safe to read a page while only holding a buffer pin. This commit
has Valgrind enforce the following rule: it is never okay to access an
nbtree buffer without holding both a pin and a lock on the buffer.
A draft version of this patch detected questionable code that was
cleaned up by commits fa7ff642 and 7154aa16. The code in question used
to access an nbtree buffer page's special/opaque area with no buffer
lock (only a buffer pin). This practice (which isn't obviously unsafe)
is hereby formally disallowed in nbtree. There doesn't seem to be any
reason to allow it, and banning it keeps things simple for Valgrind.
The new checks are implemented by adding custom nbtree client requests
(located in LockBuffer() wrapper functions); these requests are
"superimposed" on top of the generic bufmgr.c Valgrind client requests
added by commit 1e0dfd16. No custom resource management cleanup code is
needed to undo the effects of marking buffers as non-accessible under
this scheme.
Author: Peter Geoghegan
Reviewed-By: Anastasia Lubennikova, Georgios Kokolatos
Discussion: https://postgr.es/m/CAH2-WzkLgyN3zBvRZ1pkNJThC=xi_0gpWRUb_45eexLH1+k2_Q@mail.gmail.com
Strengthen the LockBuffer() assertion that verifies BufferIsValid() by
making it verify BufferIsPinned() instead. Do the same in nearby
related functions.
There is probably not much chance that anybody will try to lock a buffer
that is not already pinned, but we might as well make sure of that.
Make PinBuffer() mark buffers as defined to Valgrind unconditionally,
including when the buffer header spinlock must be acquired. Failure to
handle that case could lead to false positive reports from Valgrind.
This theoretically creates a risk that we'll mark buffers defined even
when external callers don't end up with a buffer pin. That seems
perfectly acceptable, though, since in general we make no guarantees
about buffers that are unsafe to access being reliably marked as unsafe.
Oversight in commit 1e0dfd16, which added valgrind buffer access
instrumentation.
Teach Valgrind memcheck to maintain the "defined-ness" of each shared
buffer based on whether the backend holds at least one pin at the point
it is accessed by access method code. Bugs like the one fixed by commit
b0229f26 can be detected using this new instrumentation.
Note that backends running with Valgrind naturally have their own
independent ideas about whether any given byte in shared memory is safe
or unsafe to access. There is no risk that concurrent access by
multiple backends to the same shared memory will confuse Valgrind's
instrumentation, because everything already works at the process level
(or at the memory mapping level, if you prefer).
Author: Álvaro Herrera, Peter Geoghegan
Reviewed-By: Anastasia Lubennikova
Discussion: https://postgr.es/m/20150723195349.GW5596@postgresql.org
Discussion: https://postgr.es/m/CAH2-WzkLgyN3zBvRZ1pkNJThC=xi_0gpWRUb_45eexLH1+k2_Q@mail.gmail.com
Commit ecd9e9f0b fixed the problem in the wrong place, causing unwanted
side-effects on the behavior of GetNextTempTableSpace(). Instead,
let's make SharedFileSetInit() responsible for subbing in the value
of MyDatabaseTableSpace when the default tablespace is called for.
The convention about what is in the tempTableSpaces[] array is
evidently insufficiently documented, so try to improve that.
It also looks like SharedFileSetInit() is doing the wrong thing in the
case where temp_tablespaces is empty. It was hard-wiring use of the
pg_default tablespace, but it seems like using MyDatabaseTableSpace
is more consistent with what happens for other temp files.
Back-patch the reversion of PrepareTempTablespaces()'s behavior to
9.5, as ecd9e9f0b was. The changes in SharedFileSetInit() go back
to v11 where that was introduced. (Note there is net zero code change
before v11 from these two patch sets, so nothing to release-note.)
Magnus Hagander and Tom Lane
Discussion: https://postgr.es/m/CABUevExg5YEsOvqMxrjoNvb3ApVyH+9jggWGKwTDFyFCVWczGQ@mail.gmail.com
This was a danger only for --disable-spinlocks in combination with
atomic operations unsupported by the current platform.
While atomics.c was careful to signal that a separate semaphore ought
to be used when spinlock emulation is active, spin.c didn't actually
implement that mechanism. That's my (Andres') fault, it seems to have
gotten lost during the development of the atomic operations support.
Fix that issue and add test for nesting atomic operations inside a
spinlock.
Author: Andres Freund
Discussion: https://postgr.es/m/20200605023302.g6v3ydozy5txifji@alap3.anarazel.de
Backpatch: 9.5-
On platforms without support for 64bit atomic operations where we also
cannot rely on 64bit reads to have single copy atomicity, such atomics
are implemented using a spinlock based fallback. That means it's not
safe to even read such atomics from within a signal handler (since the
signal handler might run when the spinlock already is held).
To avoid this issue defer global barrier processing out of the signal
handler. Instead of checking local / shared barrier generation to
determine whether to set ProcSignalBarrierPending, introduce
PROCSIGNAL_BARRIER and always set ProcSignalBarrierPending when
receiving such a signal. Additionally avoid redundant work in
ProcessProcSignalBarrier if ProcSignalBarrierPending is unnecessarily.
Also do a small amount of other polishing.
Author: Andres Freund
Reviewed-By: Robert Haas
Discussion: https://postgr.es/m/20200609193723.eu5ilsjxwdpyxhgz@alap3.anarazel.de
Backpatch: 13-, where the code was introduced.
Convert buffile.c error handling to use ereport. This fixes cases where
I/O errors were indistinguishable from EOF or not reported. Also remove
"%m" from error messages where errno would be bogus. While we're
modifying those strings, add block numbers and short read byte counts
where appropriate.
Back-patch to all supported releases.
Reported-by: Amit Khandekar <amitdkhan.pg@gmail.com>
Reviewed-by: Melanie Plageman <melanieplageman@gmail.com>
Reviewed-by: Alvaro Herrera <alvherre@2ndquadrant.com>
Reviewed-by: Robert Haas <robertmhaas@gmail.com>
Reviewed-by: Ibrar Ahmed <ibrar.ahmad@gmail.com>
Reviewed-by: Michael Paquier <michael@paquier.xyz>
Discussion: https://postgr.es/m/CA%2BhUKGJE04G%3D8TLK0DLypT_27D9dR8F1RQgNp0jK6qR0tZGWOw%40mail.gmail.com
Rewrite the documentation of these functions, in light of recent bug fix
commit 5940ffb2.
Back-patch to 13 where the check-for-conflict-out code was split up into
AM-specific and generic parts, and new documentation was added that now
looked wrong.
Reviewed-by: Peter Geoghegan <pg@bowt.ie>
Discussion: https://postgr.es/m/db7b729d-0226-d162-a126-8a8ab2dc4443%40jepsen.io
This was mostly confusing, especially since some wait events in
this class had the suffix and some did not.
While at it, stop exposing MainLWLockNames[] as a globally visible
name; any code using that directly is almost certainly wrong, as
its name has been misleading for some time.
(GetLWLockIdentifier() is what to use instead.)
Discussion: https://postgr.es/m/28683.1589405363@sss.pgh.pa.us
Choose names that fit into the conventions for wait event names
(particularly, that multi-word names are in the style MultiWordName)
and hopefully convey more information to non-hacker users than the
previous names did.
Also rename SerializablePredicateLockListLock to
SerializablePredicateListLock; the old name was long enough to cause
table formatting problems, plus the double occurrence of "Lock" seems
confusing/error-prone.
Also change a couple of particularly opaque LWLock field names.
Discussion: https://postgr.es/m/28683.1589405363@sss.pgh.pa.us
Originally, the names assigned to SLRUs had no purpose other than
being shmem lookup keys, so not a lot of thought went into them.
As of v13, though, we're exposing them in the pg_stat_slru view and
the pg_stat_reset_slru function, so it seems advisable to take a bit
more care. Rename them to names based on the associated on-disk
storage directories (which fortunately we *did* think about, to some
extent; since those are also visible to DBAs, consistency seems like
a good thing). Also rename the associated LWLocks, since those names
are likewise user-exposed now as wait event names.
For the most part I only touched symbols used in the respective modules'
SimpleLruInit() calls, not the names of other related objects. This
renaming could have been taken further, and maybe someday we will do so.
But for now it seems undesirable to change the names of any globally
visible functions or structs, so some inconsistency is unavoidable.
(But I *did* terminate "oldserxid" with prejudice, as I found that
name both unreadable and not descriptive of the SLRU's contents.)
Table 27.12 needs re-alphabetization now, but I'll leave that till
after the other LWLock renamings I have in mind.
Discussion: https://postgr.es/m/28683.1589405363@sss.pgh.pa.us
Includes some manual cleanup of places that pgindent messed up,
most of which weren't per project style anyway.
Notably, it seems some people didn't absorb the style rules of
commit c9d297751, because there were a bunch of new occurrences
of function calls with a newline just after the left paren, all
with faulty expectations about how the rest of the call would get
indented.
There is little point in using the LWLockRegisterTranche mechanism for
built-in tranche names. It wastes cycles, it creates opportunities for
bugs (since failing to register a tranche name is a very hard-to-detect
problem), and the lack of any centralized list of names encourages
sloppy nonconformity in name choices. Moreover, since we have a
centralized list of the tranches anyway in enum BuiltinTrancheIds, we're
certainly not buying any flexibility in return for these disadvantages.
Hence, nuke all the backend-internal LWLockRegisterTranche calls,
and instead provide a const array of the builtin tranche names.
(I have in mind to change a bunch of these names shortly, but this
patch is just about getting them into one place.)
Discussion: https://postgr.es/m/9056.1589419765@sss.pgh.pa.us
Commit 3eb77eba5a moved the loop and refactored it, and inadvertently
changed the effect of fsync=off so that it also skipped removing entries
from the pendingOps table. That was not intentional, and leads to an
assertion failure if you turn fsync on while the server is running and
reload the config.
Backpatch-through: 12-
Reviewed-By: Thomas Munro
Discussion: https://www.postgresql.org/message-id/3cbc7f4b-a5fa-56e9-9591-c886deb07513%40iki.fi
0f5ca02f53 introduces 3 new keywords. It appears to be too much for relatively
small feature. Given now we past feature freeze, it's already late for
discussion of the new syntax. So, revert.
Discussion: https://postgr.es/m/28209.1586294824%40sss.pgh.pa.us
Provide PrefetchSharedBuffer(), a variant that takes SMgrRelation, for
use in recovery. Rename LocalPrefetchBuffer() to PrefetchLocalBuffer()
for consistency.
Add a return value to all of these. In recovery, tolerate and report
missing files, so we can handle relations unlinked before crash recovery
began. Also report cache hits and misses, so that callers can do faster
buffer lookups and better I/O accounting.
Reviewed-by: Alvaro Herrera <alvherre@2ndquadrant.com>
Reviewed-by: Andres Freund <andres@anarazel.de>
Discussion: https://postgr.es/m/CA%2BhUKGJ4VJN8ttxScUFM8dOKX0BrBiboo5uz1cq%3DAovOddfHpA%40mail.gmail.com
The goal of separating hotly accessed per-backend data from PGPROC
into PGXACT is to make accesses fast (GetSnapshotData() in
particular). But delayChkpt is not actually accessed frequently; only
when starting a checkpoint. As it is frequently modified (multiple
times in the course of a single transaction), storing it in the same
cacheline as hotly accessed data unnecessarily dirties a contended
cacheline.
Therefore move delayChkpt to PGPROC.
This is part of a larger series of patches intending to improve
GetSnapshotData() scalability. It is committed and pushed separately,
as it is independently beneficial (small but measurable win, limited
by the other frequent modifications of PGXACT).
Author: Andres Freund
Reviewed-By: Robert Haas, Thomas Munro, David Rowley
Discussion: https://postgr.es/m/20200301083601.ews6hz5dduc3w2se@alap3.anarazel.de
This commit adds following optional clause to BEGIN and START TRANSACTION
commands.
WAIT FOR LSN lsn [ TIMEOUT timeout ]
New clause pospones transaction start till given lsn is applied on standby.
This clause allows user be sure, that changes previously made on primary would
be visible on standby.
New shared memory struct is used to track awaited lsn per backend. Recovery
process wakes up backend once required lsn is applied.
Author: Ivan Kartyshov, Anna Akenteva
Reviewed-by: Craig Ringer, Thomas Munro, Robert Haas, Kyotaro Horiguchi
Reviewed-by: Masahiko Sawada, Ants Aasma, Dmitry Ivanov, Simon Riggs
Reviewed-by: Amit Kapila, Alexander Korotkov
Discussion: https://postgr.es/m/0240c26c-9f84-30ea-fca9-93ab2df5f305%40postgrespro.ru
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.
Bump XLOG_PAGE_MAGIC, since this introduces XLOG_GIST_ASSIGN_LSN.
Future servers accept older WAL, so this bump is discretionary.
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
This commit introduces new wait events RecoveryConflictSnapshot and
RecoveryConflictTablespace. The former is reported while waiting for
recovery conflict resolution on a vacuum cleanup. The latter is reported
while waiting for recovery conflict resolution on dropping tablespace.
Also this commit changes the code so that the wait event Lock is reported
while waiting in ResolveRecoveryConflictWithVirtualXIDs() for recovery
conflict resolution on a lock. Basically the wait event Lock is reported
during that wait, but previously was not reported only when that wait
happened in ResolveRecoveryConflictWithVirtualXIDs().
Author: Masahiko Sawada
Reviewed-by: Fujii Masao
Discussion: https://postgr.es/m/CA+fd4k4mXWTwfQLS3RPwGr4xnfAEs1ysFfgYHvmmoUgv6Zxvmg@mail.gmail.com
Previously while the startup process was waiting for the recovery conflict
with snapshot, tablespace or lock to be resolved, waiting was reported in
PS display, but not in the case of recovery conflict with buffer pin.
This commit makes the startup process in hot standby report waiting via PS
while waiting for the conflicts with other backends holding buffer pins to
be resolved.
Author: Masahiko Sawada
Reviewed-by: Fujii Masao
Discussion: https://postgr.es/m/CA+fd4k4mXWTwfQLS3RPwGr4xnfAEs1ysFfgYHvmmoUgv6Zxvmg@mail.gmail.com
When certain parameters are changed on a physical replication primary,
this is communicated to standbys using the XLOG_PARAMETER_CHANGE WAL
record. The standby then checks whether its own settings are at least
as big as the ones on the primary. If not, the standby shuts down
with a fatal error.
The correspondence of settings between primary and standby is required
because those settings influence certain shared memory sizings that
are required for processing WAL records that the primary might send.
For example, if the primary sends a prepared transaction, the standby
must have had max_prepared_transaction set appropriately or it won't
be able to process those WAL records.
However, fatally shutting down the standby immediately upon receipt of
the parameter change record might be a bit of an overreaction. The
resources related to those settings are not required immediately at
that point, and might never be required if the activity on the primary
does not exhaust all those resources. If we just let the standby roll
on with recovery, it will eventually produce an appropriate error when
those resources are used.
So this patch relaxes this a bit. Upon receipt of
XLOG_PARAMETER_CHANGE, we still check the settings but only issue a
warning and set a global flag if there is a problem. Then when we
actually hit the resource issue and the flag was set, we issue another
warning message with relevant information. At that point we pause
recovery, so a hot standby remains usable. We also repeat the last
warning message once a minute so it is harder to miss or ignore.
Reviewed-by: Sergei Kornilov <sk@zsrv.org>
Reviewed-by: Masahiko Sawada <masahiko.sawada@2ndquadrant.com>
Reviewed-by: Kyotaro Horiguchi <horikyota.ntt@gmail.com>
Discussion: https://www.postgresql.org/message-id/flat/4ad69a4c-cc9b-0dfe-0352-8b1b0cd36c7b@2ndquadrant.com
This reverts the parts of commit 17a28b0364
that changed ereport's auxiliary functions from returning dummy integer
values to returning void. It turns out that a minority of compilers
complain (not entirely unreasonably) about constructs such as
(condition) ? errdetail(...) : 0
if errdetail() returns void rather than int. We could update those
call sites to say "(void) 0" perhaps, but the expectation for this
patch set was that ereport callers would not have to change anything.
And this aspect of the patch set was already the most invasive and
least compelling part of it, so let's just drop it.
Per buildfarm.
Discussion: https://postgr.es/m/CA+fd4k6N8EjNvZpM8nme+y+05mz-SM8Z_BgkixzkA34R+ej0Kw@mail.gmail.com
Change all the auxiliary error-reporting routines to return void,
now that we no longer need to pretend they are passing something
useful to errfinish(). While this probably doesn't save anything
significant at the machine-code level, it allows detection of some
additional types of mistakes.
Pass the error location details (__FILE__, __LINE__, PG_FUNCNAME_MACRO)
to errfinish not errstart. This shaves a few cycles off the case where
errstart decides we're not going to emit anything.
Re-implement elog() as a trivial wrapper around ereport(), removing
the separate support infrastructure it used to have. Aside from
getting rid of some now-surplus code, this means that elog() now
really does have exactly the same semantics as ereport(), in particular
that it can skip evaluation work if the message is not to be emitted.
Andres Freund and Tom Lane
Discussion: https://postgr.es/m/CA+fd4k6N8EjNvZpM8nme+y+05mz-SM8Z_BgkixzkA34R+ej0Kw@mail.gmail.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
This is required as it is no safer for two related processes to perform
clean up in gin indexes at a time than for unrelated processes to do the
same. After acquiring page locks, we can acquire relation extension lock
but reverse never happens which means these will also not participate in
deadlock. So, avoid checking wait edges from this lock.
Currently, the parallel mode is strictly read-only, but after this patch
we have the infrastructure to allow parallel inserts and parallel copy.
Author: Dilip Kumar, Amit Kapila
Reviewed-by: Amit Kapila, Kuntal Ghosh and Sawada Masahiko
Discussion: https://postgr.es/m/CAD21AoCmT3cFQUN4aVvzy5chw7DuzXrJCbrjTU05B+Ss=Gn1LA@mail.gmail.com
This is required as it is no safer for two related processes to extend the
same relation at a time than for unrelated processes to do the same. We
don't acquire a heavyweight lock on any other object after relation
extension lock which means such a lock can never participate in the
deadlock cycle. So, avoid checking wait edges from this lock.
This provides an infrastructure to allow parallel operations like insert,
copy, etc. which were earlier not possible as parallel group members won't
conflict for relation extension lock.
Author: Dilip Kumar, Amit Kapila
Reviewed-by: Amit Kapila, Kuntal Ghosh and Sawada Masahiko
Discussion: https://postgr.es/m/CAD21AoCmT3cFQUN4aVvzy5chw7DuzXrJCbrjTU05B+Ss=Gn1LA@mail.gmail.com
Assert that we don't acquire any other heavyweight lock while holding the
page lock except for relation extension. However, these locks are never
taken in reverse order which implies that page locks will never
participate in the deadlock cycle.
Similar to relation extension, page locks are also held for a short
duration, so imposing such a restriction won't hurt.
Author: Dilip Kumar, with few changes by Amit Kapila
Reviewed-by: Amit Kapila, Kuntal Ghosh and Sawada Masahiko
Discussion: https://postgr.es/m/CAD21AoCmT3cFQUN4aVvzy5chw7DuzXrJCbrjTU05B+Ss=Gn1LA@mail.gmail.com
relation extension lock.
The only exception to the rule is that we can try to acquire the same
relation extension lock more than once. This is allowed as we are not
creating any new lock for this case. This restriction implies that the
relation extension lock won't ever participate in the deadlock cycle
because we can never wait for any other heavyweight lock after acquiring
this lock.
Such a restriction is okay for relation extension locks as unlike other
heavyweight locks these are not held till the transaction end. These are
taken for a short duration to extend a particular relation and then
released.
Author: Dilip Kumar, with few changes by Amit Kapila
Reviewed-by: Amit Kapila, Kuntal Ghosh and Sawada Masahiko
Discussion: https://postgr.es/m/CAD21AoCmT3cFQUN4aVvzy5chw7DuzXrJCbrjTU05B+Ss=Gn1LA@mail.gmail.com
For the semantics to match the epoll implementation, we need a socket to
continue to appear readable/writable if you wait multiple times without
doing I/O in between (in Linux terminology: level-triggered rather than
edge-triggered). This distinction will be important for later commits.
Similar to commit 3b790d256f for Windows.
Reviewed-by: Kyotaro Horiguchi <horikyota.ntt@gmail.com>
Discussion: https://postgr.es/m/CA%2BhUKGJAC4Oqao%3DqforhNey20J8CiG2R%3DoBPqvfR0vOJrFysGw%40mail.gmail.com
While running under a debugger, macOS's getppid() can return the
debugger's PID. That could cause a backend to exit because it falsely
believed that the postmaster had died, since commit 815c2f09.
Continue to use getppid() as a fast postmaster check after adding the
postmaster's PID to a kqueue, to close a PID-reuse race, but double
check that it actually exited by trying to read the pipe. The new check
isn't reached in the common case.
Reported-by: Alexander Korotkov <a.korotkov@postgrespro.ru>
Discussion: https://postgr.es/m/CA%2BhUKGKhAxJ8V8RVwCo6zJaeVrdOG1kFBHGZOOjf6DzW_omeMA%40mail.gmail.com
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 effective_io_concurrency GUC and equivalent tablespace option were
previously passed through a formula based on a theory about RAID
spindles and probabilities, to arrive at the number of pages to prefetch
in bitmap heap scans. Tomas Vondra, Andres Freund and others argued
that it was anachronistic and hard to justify, and commit 558a9165e0
already started down the path of bypassing it in new code. We agreed to
drop that logic and use the value directly.
For the default setting of 1, there is no change in effect. Higher
settings can be converted from the old meaning to the new with:
select round(sum(OLD / n::float)) from generate_series(1, OLD) s(n);
We might want to consider renaming the GUC before the next release given
the change in meaning, but it's not clear that many users had set it
very carefully anyway. That decision is deferred for now.
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
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
Previously "waiting" could appear twice via PS in case of lock conflict
in hot standby mode. Specifically this issue happend when the delay
in WAL application determined by max_standby_archive_delay and
max_standby_streaming_delay had passed but it took more than 500 msec
to cancel all the conflicting transactions. Especially we can observe this
easily by setting those delay parameters to -1.
The cause of this issue was that WaitOnLock() and
ResolveRecoveryConflictWithVirtualXIDs() added "waiting" to
the process title in that case. This commit prevents
ResolveRecoveryConflictWithVirtualXIDs() from reporting waiting
in case of lock conflict, to fix the bug.
Back-patch to all back branches.
Author: Masahiko Sawada
Reviewed-by: Fujii Masao
Discussion: https://postgr.es/m/CA+fd4k4mXWTwfQLS3RPwGr4xnfAEs1ysFfgYHvmmoUgv6Zxvmg@mail.gmail.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
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
Previously PostgreSQL built with -DLWLOCK_STATS could report
more than one LWLock statistics entries for the same backend
process and the same LWLock. This is strange and only one
statistics should be output in that case, instead.
The cause of this issue is that the key variable used for
LWLock stats hash table was not fully initialized. The key
consists of two fields and they were initialized. But
the following 4 bytes allocated in the key variable for
the alignment was not initialized. So even if the same key
was specified, hash_search(HASH_ENTER) could not find
the existing entry for that key and created new one.
This commit fixes this issue by initializing the key
variable with zero. As the side effect of this commit,
the volume of LWLock statistics output would be reduced
very much.
Back-patch to v10, where commit 3761fe3c20 introduced the issue.
Author: Fujii Masao
Reviewed-by: Julien Rouhaud, Kyotaro Horiguchi
Discussion: https://postgr.es/m/26359edb-798a-568f-d93a-6aafac49752d@oss.nttdata.com
Use kevent(2) to wait for events on the BSD family of operating
systems and macOS. This is similar to the epoll(2) support added
for Linux by commit 98a64d0bd.
Author: Thomas Munro
Reviewed-by: Andres Freund, Marko Tiikkaja, Tom Lane
Tested-by: Mateusz Guzik, Matteo Beccati, Keith Fiske, Heikki Linnakangas, Michael Paquier, Peter Eisentraut, Rui DeSousa, Tom Lane, Mark Wong
Discussion: https://postgr.es/m/CAEepm%3D37oF84-iXDTQ9MrGjENwVGds%2B5zTr38ca73kWR7ez_tA%40mail.gmail.com
Those new assertions can be used at file scope, outside of any function
for compilation checks. This commit provides implementations for C and
C++, and fallback implementations.
Author: Peter Smith
Reviewed-by: Andres Freund, Kyotaro Horiguchi, Dagfinn Ilmari Mannsåker,
Michael Paquier
Discussion: https://postgr.es/m/201DD0641B056142AC8C6645EC1B5F62014B8E8030@SYD1217
If we attempt to create a DSM segment when no slots are available,
we should return the memory to the operating system. Previously
we did that if the DSM_CREATE_NULL_IF_MAXSEGMENTS flag was
passed in, but we didn't do it if an error was raised. Repair.
Back-patch to 9.4, where DSM segments arrived.
Author: Thomas Munro
Reviewed-by: Robert Haas
Reported-by: Julian Backes
Discussion: https://postgr.es/m/CA%2BhUKGKAAoEw-R4om0d2YM4eqT1eGEi6%3DQot-3ceDR-SLiWVDw%40mail.gmail.com
When running a lot of large parallel queries concurrently, or a plan with
a lot of separate Gather nodes, it is possible to run out of DSM slots.
There are better solutions to these problems requiring architectural
redesign work, but for now, let's adjust the constants so that it's more
difficult to hit the limit.
1. Previously, a DSA area would create up to four segments at each size
before doubling the size. After this commit, it will create only two at
each size, so it ramps up faster and therefore needs fewer slots.
2. Previously, the total limit on DSM slots allowed for 2 per connection.
Switch to 5 per connection.
Also remove an obsolete nearby comment.
Author: Thomas Munro
Reviewed-by: Robert Haas, Andres Freund
Discussion: https://postre.es/m/CA%2BhUKGL6H2BpGbiF7Lj6QiTjTGyTLW_vLR%3DSn2tEBeTcYXiMKw%40mail.gmail.com
It's not OK to do that without calling CHECK_FOR_INTERRUPTS().
Let the next wait loop deal with it, following the usual pattern.
One consequence of this bug was that a SIGTERM delivered in a very
narrow timing window could leave a parallel worker process waiting
forever for a condition variable that will never be signaled, after
an error was raised in other process.
The code is a bit different in the stable branches due to commit
1321509f, making problems less likely there. No back-patch for now,
but we may finish up deciding to make a similar change after more
discussion.
Author: Thomas Munro
Reviewed-by: Shawn Debnath
Reported-by: Tomas Vondra
Discussion: https://postgr.es/m/CA%2BhUKGJOm8zZHjVA8svoNT3tHY0XdqmaC_kHitmgXDQM49m1dA%40mail.gmail.com
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
This function allows us to fsync the specified file or directory.
It's useful, for example, when we want to sync the file that
pg_file_write() writes out or that COPY TO exports the data into,
for durability.
Author: Fujii Masao
Reviewed-By: Julien Rouhaud, Arthur Zakirov, Michael Paquier, Atsushi Torikoshi
Discussion: https://www.postgresql.org/message-id/CAHGQGwGY8uzZ_k8dHRoW1zDcy1Z7=5GQ+So4ZkVy2u=nLsk=hA@mail.gmail.com
FileClose() failure ordinarily causes a PANIC. Suppose the user
disables that PANIC via data_sync_retry=on. After mdclose() issued a
FileClose() that failed, calls into md.c raised SIGSEGV. This fix adds
repalloc() calls during mdclose(); update a comment about ignoring
repalloc() cost. The rate of relation segment count change is a minor
factor; more relevant to overall performance is the rate of mdclose()
and subsequent re-opening of segments. Back-patch to v10, where commit
45e191e3aa introduced the bug.
Reviewed by Kyotaro Horiguchi.
Discussion: https://postgr.es/m/20191222091930.GA1280238@rfd.leadboat.com
This tells you about allocations that have been made from the main
shared memory segment. The original patch also tried to show information
about dynamic shared memory allocation as well, but I decided to
leave that problem for another time.
Andres Freund and Robert Haas, reviewed by Michael Paquier, Marti
Raudsepp, Tom Lane, Álvaro Herrera, and Kyotaro Horiguchi.
Discussion: http://postgr.es/m/20140504114417.GM12715@awork2.anarazel.de
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 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
A new function EmitProcSignalBarrier() can be used to emit a global
barrier which all backends that participate in the ProcSignal
mechanism must absorb, and a new function WaitForProcSignalBarrier()
can be used to wait until all relevant backends have in fact
absorbed the barrier.
This can be used to coordinate global state changes, such as turning
checksums on while the system is running.
There's no real client of this mechanism yet, although two are
proposed, but an enum has to have at least one element, so this
includes a placeholder type (PROCSIGNAL_BARRIER_PLACEHOLDER) which
should be replaced by the first real client of this mechanism to
get committed.
Andres Freund and Robert Haas, reviewed by Daniel Gustafsson and,
in earlier versions, by Magnus Hagander.
Discussion: http://postgr.es/m/CA+TgmoZwDk=BguVDVa+qdA6SBKef=PKbaKDQALTC_9qoz1mJqg@mail.gmail.com
The previous commit failed to consider that FileGetRawDesc() might
not return a valid fd, as discovered on the build farm. Switch to
using the File interface only.
Back-patch to 12, like the previous commit.
_mdfd_getseg() opens all segments up to the requested one. That
causes problems for mdsyncfiletag(), if mdunlinkfork() has
already unlinked other segment files. Open the file we want
directly by name instead, if it's not already open.
The consequence of this bug was a rare panic in the checkpointer,
made more likely if you saturated the sync request queue so that
the SYNC_FORGET_REQUEST messages for a given relation were more
likely to be absorbed in separate cycles by the checkpointer.
Back-patch to 12. Defect in commit 3eb77eba.
Author: Thomas Munro
Reported-by: Justin Pryzby
Discussion: https://postgr.es/m/20191119115759.GI30362%40telsasoft.com
On some platforms, fsync() returns EBADFD when opening a file descriptor
with O_RDONLY (read-only), leading ultimately now to a PANIC to prevent
data corruption.
This commit adds a new sanity check in pg_fsync() based on fcntl() to
make sure that we don't repeat again mistakes with incorrectly-set file
descriptors so as problems are detected at an early stage. Without
that, such errors could only be detected after running Postgres on a
specific supported platform for the culprit code path, which could take
some time before being found. b8e19b93 was a fix for such a problem,
which got undetected for more than 5 years, and a586cc4b fixed another
similar issue.
Note that the new check added works as well when fsync=off is
configured, so as all regression tests would detect problems as long as
assertions are enabled. fcntl() being not available on Windows, the
new checks do not happen there.
Author: Michael Paquier
Reviewed-by: Mark Dilger
Discussion: https://postgr.es/m/20191009062640.GB21379@paquier.xyz
This new option terminates the other sessions connected to the target
database and then drop it. To terminate other sessions, the current user
must have desired permissions (same as pg_terminate_backend()). We don't
allow to terminate the sessions if prepared transactions, active logical
replication slots or subscriptions are present in the target database.
Author: Pavel Stehule with changes by me
Reviewed-by: Dilip Kumar, Vignesh C, Ibrar Ahmed, Anthony Nowocien,
Ryan Lambert and Amit Kapila
Discussion: https://postgr.es/m/CAP_rwwmLJJbn70vLOZFpxGw3XD7nLB_7+NKz46H5EOO2k5H7OQ@mail.gmail.com