e85662df44 made GetRunningTransactionData() calculate the oldest running
transaction id within the current database. However, because of the typo,
the new code uses oldestRunningXid instead of oldestDatabaseRunningXid
in comparison before updating oldestDatabaseRunningXid. This commit fixes
that issue.
Reported-by: Noah Misch
Discussion: https://postgr.es/m/20240630231816.bf.nmisch%40google.com
Backpatch-through: 17
We did not recover the subtransaction IDs of prepared transactions
when starting a hot standby from a shutdown checkpoint. As a result,
such subtransactions were considered as aborted, rather than
in-progress. That would lead to hint bits being set incorrectly, and
the subtransactions suddenly becoming visible to old snapshots when
the prepared transaction was committed.
To fix, update pg_subtrans with prepared transactions's subxids when
starting hot standby from a shutdown checkpoint. The snapshots taken
from that state need to be marked as "suboverflowed", so that we also
check the pg_subtrans.
Backport to all supported versions.
Discussion: https://www.postgresql.org/message-id/6b852e98-2d49-4ca1-9e95-db419a2696e0@iki.fi
Specifically, it terminates a background worker even if the caller
couldn't terminate the background worker with pg_terminate_backend().
Commit 3a9b18b309 neglected to update
this. Back-patch to v13, which introduced DROP DATABASE FORCE.
Reviewed by Amit Kapila. Reported by Kirill Reshke.
Discussion: https://postgr.es/m/20240429212756.60.nmisch@google.com
GlobalVisTestNonRemovableHorizon() was only used for the implementation of
snapshot_too_old, which was removed in f691f5b80a. As using
GlobalVisTestNonRemovableHorizon() is not particularly efficient, no new uses
for it should be added. Therefore remove.
Discussion: https://postgr.es/m/20240415185720.q4dg4dlcyvvrabz4@awork3.anarazel.de
Execute both freezing and pruning of tuples in the same
heap_page_prune() function, now called heap_page_prune_and_freeze(),
and emit a single WAL record containing all changes. That reduces the
overall amount of WAL generated.
This moves the freezing logic from vacuumlazy.c to the
heap_page_prune_and_freeze() function. The main difference in the
coding is that in vacuumlazy.c, we looked at the tuples after the
pruning had already happened, but in heap_page_prune_and_freeze() we
operate on the tuples before pruning. The heap_prepare_freeze_tuple()
function is now invoked after we have determined that a tuple is not
going to be pruned away.
VACUUM no longer needs to loop through the items on the page after
pruning. heap_page_prune_and_freeze() does all the work. It now
returns the list of dead offsets, including existing LP_DEAD items, to
the caller. Similarly it's now responsible for tracking 'all_visible',
'all_frozen', and 'hastup' on the caller's behalf.
Author: Melanie Plageman <melanieplageman@gmail.com>
Discussion: https://www.postgresql.org/message-id/20240330055710.kqg6ii2cdojsxgje@liskov
Currently, pg_visibility computes its xid horizon using the
GetOldestNonRemovableTransactionId(). The problem is that this horizon can
sometimes go backward. That can lead to reporting false errors.
In order to fix that, this commit implements a new function
GetStrictOldestNonRemovableTransactionId(). This function computes the xid
horizon, which would be guaranteed to be newer or equal to any xid horizon
computed before.
We have to do the following to achieve this.
1. Ignore processes xmin's, because they consider connection to other databases
that were ignored before.
2. Ignore KnownAssignedXids, because they are not database-aware. At the same
time, the primary could compute its horizons database-aware.
3. Ignore walsender xmin, because it could go backward if some replication
connections don't use replication slots.
As a result, we're using only currently running xids to compute the horizon.
Surely these would significantly sacrifice accuracy. But we have to do so to
avoid reporting false errors.
Inspired by earlier patch by Daniel Shelepanov and the following discussion
with Robert Haas and Tom Lane.
Discussion: https://postgr.es/m/1649062270.289865713%40f403.i.mail.ru
Reviewed-by: Alexander Lakhin, Dmitry Koval
as determined by include-what-you-use (IWYU)
While IWYU also suggests to *add* a bunch of #include's (which is its
main purpose), this patch does not do that. In some cases, a more
specific #include replaces another less specific one.
Some manual adjustments of the automatic result:
- IWYU currently doesn't know about includes that provide global
variable declarations (like -Wmissing-variable-declarations), so
those includes are being kept manually.
- All includes for port(ability) headers are being kept for now, to
play it safe.
- No changes of catalog/pg_foo.h to catalog/pg_foo_d.h, to keep the
patch from exploding in size.
Note that this patch touches just *.c files, so nothing declared in
header files changes in hidden ways.
As a small example, in src/backend/access/transam/rmgr.c, some IWYU
pragma annotations are added to handle a special case there.
Discussion: https://www.postgresql.org/message-id/flat/af837490-6b2f-46df-ba05-37ea6a6653fc%40eisentraut.org
Now that BackendId was just another index into the proc array, it was
redundant with the 0-based proc numbers used in other places. Replace
all usage of backend IDs with proc numbers.
The only place where the term "backend id" remains is in a few pgstat
functions that expose backend IDs at the SQL level. Those IDs are now
in fact 0-based ProcNumbers too, but the documentation still calls
them "backend ids". That term still seems appropriate to describe what
the numbers are, so I let it be.
One user-visible effect is that pg_temp_0 is now a valid temp schema
name, for backend with ProcNumber 0.
Reviewed-by: Andres Freund
Discussion: https://www.postgresql.org/message-id/8171f1aa-496f-46a6-afc3-c46fe7a9b407@iki.fi
Previously, backend ID was an index into the ProcState array, in the
shared cache invalidation manager (sinvaladt.c). The entry in the
ProcState array was reserved at backend startup by scanning the array
for a free entry, and that was also when the backend got its backend
ID. Things become slightly simpler if we redefine backend ID to be the
index into the PGPROC array, and directly use it also as an index to
the ProcState array. This uses a little more memory, as we reserve a
few extra slots in the ProcState array for aux processes that don't
need them, but the simplicity is worth it.
Aux processes now also have a backend ID. This simplifies the
reservation of BackendStatusArray and ProcSignal slots.
You can now convert a backend ID into an index into the PGPROC array
simply by subtracting 1. We still use 0-based "pgprocnos" in various
places, for indexes into the PGPROC array, but the only difference now
is that backend IDs start at 1 while pgprocnos start at 0. (The next
commmit will get rid of the term "backend ID" altogether and make
everything 0-based.)
There is still a 'backendId' field in PGPROC, now part of 'vxid' which
encapsulates the backend ID and local transaction ID together. It's
needed for prepared xacts. For regular backends, the backendId is
always equal to pgprocno + 1, but for prepared xact PGPROC entries,
it's the ID of the original backend that processed the transaction.
Reviewed-by: Andres Freund, Reid Thompson
Discussion: https://www.postgresql.org/message-id/8171f1aa-496f-46a6-afc3-c46fe7a9b407@iki.fi
Since 5a1dfde833, 2PC filenames use FullTransactionId. Thus, it needs to
convert TransactionId to FullTransactionId in StandbyTransactionIdIsPrepared()
using TransamVariables->nextXid. However, ProcArrayApplyRecoveryInfo()
first releases locks with usage StandbyTransactionIdIsPrepared(), then advances
TransamVariables->nextXid. This sequence of actions could cause errors.
This commit makes ProcArrayApplyRecoveryInfo() advance
TransamVariables->nextXid before releasing locks.
Reported-by: Thomas Munro, Michael Paquier
Discussion: https://postgr.es/m/CA%2BhUKGLj_ve1_pNAnxwYU9rDcv7GOhsYXJt7jMKSA%3D5-6ss-Cw%40mail.gmail.com
Discussion: https://postgr.es/m/Zadp9f4E1MYvMJqe%40paquier.xyz
This GUC was intended as a debugging help in the 9.0 area when hot
standby and streaming replication were being developped, able to offer
more information at LOG level rather than DEBUGn. There are more tools
available these days that are able to offer rather equivalent
information, like pg_waldump introduced in 9.3. It is not obvious how
this facility is useful these days, so let's remove it.
Author: Bharath Rupireddy
Discussion: https://postgr.es/m/ZXEXEAUVFrvpquSd@paquier.xyz
Commit b7eda3e0e moved XidInMVCCSnapshot() from tqual.c into snapmgr.c,
but follow-up commit c91560def incorrectly updated this reference. We
could fix it, but as pointed out by Daniel Gustafsson, 1) the reader can
easily find the file that contains the definition of that function, e.g.
by grepping, and 2) this kind of reference is prone to going stale; so
let's just remove it.
Back-patch to all supported branches.
Reviewed by Daniel Gustafsson.
Discussion: https://postgr.es/m/CAPmGK145VdKkPBLWS2urwhgsfidbSexwY-9zCL6xSUJH%2BBTUUg%40mail.gmail.com
Since C99, there can be a trailing comma after the last value in an
enum definition. A lot of new code has been introducing this style on
the fly. Some new patches are now taking an inconsistent approach to
this. Some add the last comma on the fly if they add a new last
value, some are trying to preserve the existing style in each place,
some are even dropping the last comma if there was one. We could
nudge this all in a consistent direction if we just add the trailing
commas everywhere once.
I omitted a few places where there was a fixed "last" value that will
always stay last. I also skipped the header files of libpq and ecpg,
in case people want to use those with older compilers. There were
also a small number of cases where the enum type wasn't used anywhere
(but the enum values were), which ended up confusing pgindent a bit,
so I left those alone.
Discussion: https://www.postgresql.org/message-id/flat/386f8c45-c8ac-4681-8add-e3b0852c1620%40eisentraut.org
This lock was introduced before memory barrier support was added,
and it is only used to guarantee proper memory ordering when
KnownAssignedXidsAdd() appends to the array without a lock. Now
that such memory barrier support exists, we can remove the lock and
use barriers instead.
Suggested-by: Tom Lane
Author: Michail Nikolaev
Reviewed-by: Robert Haas
Discussion: https://postgr.es/m/CANtu0oh0si%3DjG5z_fLeFtmYcETssQ08kLEa8b6TQqDm_cinroA%40mail.gmail.com
vacuum_defer_cleanup_age was introduced before hot_standby_feedback and
replication slots existed. It is hard to use reasonably - commonly it will
either be set too low (not preventing recovery conflicts, while still causing
some bloat), or too high (causing a lot of bloat). The alternatives do not
have that issue.
That on its own might not be sufficient reason to remove
vacuum_defer_cleanup_age, but it also complicates computation of xid
horizons. See e.g. the bug fixed in be504a3e97. It also is untested.
This commit removes TransactionIdRetreatSafely(), as there are no users
anymore. There might be potential future users, hence noting that here.
Reviewed-by: Daniel Gustafsson <daniel@yesql.se>
Reviewed-by: Justin Pryzby <pryzby@telsasoft.com>
Reviewed-by: Alvaro Herrera <alvherre@alvh.no-ip.org>
Discussion: https://postgr.es/m/20230317230930.nhsgk3qfk7f4axls@awork3.anarazel.de
When vacuum_defer_cleanup_age is bigger than the current xid, including the
epoch, the subtraction of vacuum_defer_cleanup_age would lead to a wrapped
around xid. While that normally is not a problem, the subsequent conversion to
a 64bit xid results in a 64bit-xid very far into the future. As that xid is
used as a horizon to detect whether rows versions are old enough to be
removed, that allows removal of rows that are still visible (i.e. corruption).
If vacuum_defer_cleanup_age was never changed from the default, there is no
chance of this bug occurring.
This bug was introduced in dc7420c2c9. A lesser version of it exists in
12-13, introduced by fb5344c969, affecting only GiST.
The 12-13 version of the issue can, in rare cases, lead to pages in a gist
index getting recycled too early, potentially causing index entries to be
found multiple times.
The fix is fairly simple - don't allow vacuum_defer_cleanup_age to retreat
further than FirstNormalTransactionId.
Patches to make similar bugs easier to find, by adding asserts to the 64bit
xid infrastructure, have been proposed, but are not suitable for backpatching.
Currently there are no tests for vacuum_defer_cleanup_age. A patch introducing
infrastructure to make writing a test easier has been posted to the list.
Reported-by: Michail Nikolaev <michail.nikolaev@gmail.com>
Reviewed-by: Matthias van de Meent <boekewurm+postgres@gmail.com>
Author: Andres Freund <andres@anarazel.de>
Discussion: https://postgr.es/m/20230108002923.cyoser3ttmt63bfn@awork3.anarazel.de
Backpatch: 12-, but impact/fix is smaller for 12-13
Previously, we'd compress only when the active range of array entries
reached Max(4 * PROCARRAY_MAXPROCS, 2 * pArray->numKnownAssignedXids).
If max_connections is large, the first term could result in not
compressing for a long time, resulting in much wastage of cycles in
hot-standby backends scanning the array to take snapshots. Get rid
of that term, and just bound it to 2 * pArray->numKnownAssignedXids.
That however creates the opposite risk, that we might spend too much
effort compressing. Hence, consider compressing only once every 128
commit records. (This frequency was chosen by benchmarking. While
we only tried one benchmark scenario, the results seem stable over
a fairly wide range of frequencies.)
Also, force compression when processing RecoveryInfo WAL records
(which should be infrequent); the old code could perform compression
then, but would do so only after the same array-range check as for
the transaction-commit path.
Also, opportunistically run compression if the startup process is about
to wait for WAL, though not oftener than once a second. This should
prevent cases where we waste lots of time by leaving the array
not-compressed for long intervals due to low WAL traffic.
Lastly, add a simple check to keep us from uselessly compressing
when the array storage is already compact.
Back-patch, as the performance problem is worse in pre-v14 branches
than in HEAD.
Simon Riggs and Michail Nikolaev, with help from Tom Lane and
Andres Freund.
Discussion: https://postgr.es/m/CALdSSPgahNUD_=pB_j=1zSnDBaiOtqVfzo8Ejt5J_k7qZiU1Tw@mail.gmail.com
Once a logical slot has acquired a catalog_xmin, it doesn't let go of
it, even when invalidated by exceeding the max_slot_wal_keep_size, which
means that dead catalog tuples are not removed by vacuum anymore since
the point is invalidated, until the slot is dropped. This could be
catastrophic if catalog churn is high.
Change the computation of Xmin to ignore invalidated slots,
to prevent dead rows from accumulating.
Backpatch to 13, where slot invalidation appeared.
Author: Sirisha Chamarthi <sirichamarthi22@gmail.com>
Reviewed-by: Ashutosh Bapat <ashutosh.bapat.oss@gmail.com>
Discussion: https://postgr.es/m/CAKrAKeUEDeqquN9vwzNeG-CN8wuVsfRYbeOUV9qKO_RHok=j+g@mail.gmail.com
Standardize on the name snapshotConflictHorizon for all XID fields from
WAL records that generate recovery conflicts when in hot standby mode.
This supersedes the previous latestRemovedXid naming convention.
The new naming convention places emphasis on how the values are actually
used by REDO routines. How the values are generated during original
execution (details of which vary by record type) is deemphasized. Users
of tools like pg_waldump can now grep for snapshotConflictHorizon to see
all potential sources of recovery conflicts in a standardized way,
without necessarily having to consider which specific record types might
be involved.
Also bring a couple of WAL record types that didn't follow any kind of
naming convention into line. These are heapam's VISIBLE record type and
SP-GiST's VACUUM_REDIRECT record type. Now every WAL record whose REDO
routine calls ResolveRecoveryConflictWithSnapshot() passes through the
snapshotConflictHorizon field from its WAL record. This is follow-up
work to the refactoring from commit 9e540599 that made FREEZE_PAGE WAL
records use a standard snapshotConflictHorizon style XID cutoff.
No bump in XLOG_PAGE_MAGIC, since the underlying format of affected WAL
records doesn't change.
Author: Peter Geoghegan <pg@bowt.ie>
Reviewed-By: Andres Freund <andres@anarazel.de>
Discussion: https://postgr.es/m/CAH2-Wzm2CQUmViUq7Opgk=McVREHSOorYaAjR1ZpLYkRN7_dPw@mail.gmail.com
This commit updates two code paths to use pg_lfind32() introduced by
b6ef167 with TransactionId arrays:
- At the end of TransactionIdIsInProgress(), when checking for the case
of still running but overflowed subxids.
- XidIsConcurrent(), when checking for a serializable conflict.
These cases are less impactful than 37a6e5d, but a bit of
micro-benchmarking of this API shows that linear search speeds up by
~20% depending on the number of items involved (x86_64 and amd64 looked
at here).
Author: Nathan Bossart
Reviewed-by: Richard Guo, Michael Paquier
Discussion: https://postgr.es/m/20220901185153.GA783106@nathanxps13
Make sure that function declarations use names that exactly match the
corresponding names from function definitions in storage, catalog,
access method, executor, and logical replication code, as well as in
miscellaneous utility/library code.
Like other recent commits that cleaned up function parameter names, this
commit was written with help from clang-tidy. Later commits will do the
same for other parts of the codebase.
Author: Peter Geoghegan <pg@bowt.ie>
Reviewed-By: David Rowley <dgrowleyml@gmail.com>
Discussion: https://postgr.es/m/CAH2-WznJt9CMM9KJTMjJh_zbL5hD9oX44qdJ4aqZtjFi-zA3Tg@mail.gmail.com
TransactionIdIsInProgress had a fast path to return 'false' if the
single-item CLOG cache said that the transaction was known to be
committed. However, that was wrong, because a transaction is first
marked as committed in the CLOG but doesn't become visible to others
until it has removed its XID from the proc array. That could lead to an
error:
ERROR: t_xmin is uncommitted in tuple to be updated
or for an UPDATE to go ahead without blocking, before the previous
UPDATE on the same row was made visible.
The window is usually very short, but synchronous replication makes it
much wider, because the wait for synchronous replica happens in that
window.
Another thing that makes it hard to hit is that it's hard to get such
a commit-in-progress transaction into the single item CLOG cache.
Normally, if you call TransactionIdIsInProgress on such a transaction,
it determines that the XID is in progress without checking the CLOG
and without populating the cache. One way to prime the cache is to
explicitly call pg_xact_status() on the XID. Another way is to use a
lot of subtransactions, so that the subxid cache in the proc array is
overflown, making TransactionIdIsInProgress rely on pg_subtrans and
CLOG checks.
This has been broken ever since it was introduced in 2008, but the race
condition is very hard to hit, especially without synchronous
replication. There were a couple of reports of the error starting from
summer 2021, but no one was able to find the root cause then.
TransactionIdIsKnownCompleted() is now unused. In 'master', remove it,
but I left it in place in backbranches in case it's used by extensions.
Also change pg_xact_status() to check TransactionIdIsInProgress().
Previously, it only checked the CLOG, and returned "committed" before
the transaction was actually made visible to other queries. Note that
this also means that you cannot use pg_xact_status() to reproduce the
bug anymore, even if the code wasn't fixed.
Report and analysis by Konstantin Knizhnik. Patch by Simon Riggs, with
the pg_xact_status() change added by me.
Author: Simon Riggs
Reviewed-by: Andres Freund
Discussion: https://www.postgresql.org/message-id/flat/4da7913d-398c-e2ad-d777-f752cf7f0bbb%40garret.ru
This reverts commit d9d076222f "VACUUM: ignore indexing operations
with CONCURRENTLY".
These changes caused indexes created with the CONCURRENTLY option to
miss heap tuples that were HOT-updated and HOT-pruned during the index
creation. Before these changes, HOT pruning would have been prevented
by the Xmin of the transaction creating the index, but because this
change was precisely to allow the Xmin to move forward ignoring that
backend, now other backends scanning the table can prune them. This is
not a problem for VACUUM (which requires a lock that conflicts with a
CREATE INDEX CONCURRENTLY operation), but HOT-prune can definitely
occur. In other words, Xmin advancement was sped up, but at the cost of
corrupting the resulting index.
Regrettably, this means that the new feature in PG14 that RIC/CIC on
very large tables no longer force VACUUM to retain very old tuples goes
away. We might try to implement it again in a later release, but for
now the risk of indexes missing tuples is too high and there's no easy
fix.
Backpatch to 14, where this change appeared.
Reported-by: Peter Slavov <pet.slavov@gmail.com>
Diagnosys-by: Andrey Borodin <x4mmm@yandex-team.ru>
Diagnosys-by: Michael Paquier <michael@paquier.xyz>
Diagnosys-by: Andres Freund <andres@anarazel.de>
Discussion: https://postgr.es/m/17485-396609c6925b982d%40postgresql.org
This is a slight, convenient semantics change from what commit
0f0cfb4940 ("Fix parallel operations that prevent oldest xmin from
advancing") introduced that lets us simplify the coding in the one place
where it is used.
Backpatch to 13. This is related to commit 6fea65508a ("Tighten
ComputeXidHorizons' handling of walsenders") rewriting the code site
where this is used, which has not yet been backpatched, but it may well
be in the future.
Reviewed-by: Masahiko Sawada <sawada.mshk@gmail.com>
Discussion: https://postgr.es/m/202204191637.eldwa2exvguw@alvherre.pgsql
ComputeXidHorizons (nee GetOldestXmin) thought that it could identify
walsenders by checking for proc->databaseId == 0. Perhaps that was
safe when the code was written, but it's been wrong at least since
autovacuum was invented. Background processes that aren't connected
to any particular database, such as the autovacuum launcher and
logical replication launcher, look like that too.
This imprecision is harmful because when such a process advertises an
xmin, the result is to hold back dead-tuple cleanup in all databases,
though it'd be sufficient to hold it back in shared catalogs (which
are the only relations such a process can access). Aside from being
generally inefficient, this has recently been seen to cause regression
test failures in the buildfarm, as a consequence of the logical
replication launcher's startup transaction preventing VACUUM from
marking pages of a user table as all-visible.
We only want that global hold-back effect for the case where a
walsender is advertising a hot standby feedback xmin. Therefore,
invent a new PGPROC flag that says that a process' xmin should be
considered globally, and check that instead of using the incorrect
databaseId == 0 test. Currently only a walsender sets that flag,
and only if it is not connected to any particular database. (This is
for bug-compatibility with the undocumented behavior of the existing
code, namely that feedback sent by a client who has connected to a
particular database would not be applied globally. I'm not sure this
is a great definition; however, such a client is capable of issuing
plain SQL commands, and I don't think we want xmins advertised for
such commands to be applied globally. Perhaps this could do with
refinement later.)
While at it, I rewrote the comment in ComputeXidHorizons, and
re-ordered the commented-upon if-tests, to make them match up
for intelligibility's sake.
This is arguably a back-patchable bug fix, but given the lack of
complaints I think it prudent to let it age awhile in HEAD first.
Discussion: https://postgr.es/m/1346227.1649887693@sss.pgh.pa.us
This reverts commits 0147fc7, 4567596, aa64f23, and 5ecd018.
There is no longer agreement that introducing this function
was the right way to address the problem. The consensus now
seems to favor trying to make a correct value for MaxBackends
available to mdules executing their _PG_init() functions.
Nathan Bossart
Discussion: http://postgr.es/m/20220323045229.i23skfscdbvrsuxa@jrouhaud
Before commit 412ad7a556, delayChkpt
was a Boolean. Now it's an integer. Extensions using it need to be
appropriately updated, so let's rename the field to make sure that
a hard compilation failure occurs.
Replacing delayChkpt with delayChkptFlags made a few comments extend
past 80 characters, so I reflowed them and changed some wording very
slightly.
The back-branches will need a different change to restore compatibility
with existing minor releases; this is just for master.
Per suggestion from Tom Lane.
Discussion: http://postgr.es/m/a7880f4d-1d74-582a-ada7-dad168d046d1@enterprisedb.com
If TRUNCATE causes some buffers to be invalidated and thus the
checkpoint does not flush them, TRUNCATE must also ensure that the
corresponding files are truncated on disk. Otherwise, a replay
from the checkpoint might find that the buffers exist but have
the wrong contents, which may cause replay to fail.
Report by Teja Mupparti. Patch by Kyotaro Horiguchi, per a design
suggestion from Heikki Linnakangas, with some changes to the
comments by me. Review of this and a prior patch that approached
the issue differently by Heikki Linnakangas, Andres Freund, Álvaro
Herrera, Masahiko Sawada, and Tom Lane.
Discussion: http://postgr.es/m/BYAPR06MB6373BF50B469CA393C614257ABF00@BYAPR06MB6373.namprd06.prod.outlook.com
Previously, it was really easy to write code that accessed MaxBackends
before we'd actually initialized it, especially when coding up an
extension. To make this less error-prune, introduce a new function
GetMaxBackends() which should be used to obtain the correct value.
This will ERROR if called too early. Demote the global variable to
a file-level static, so that nobody can peak at it directly.
Nathan Bossart. Idea by Andres Freund. Review by Greg Sabino Mullane,
by Michael Paquier (who had doubts about the approach), and by me.
Discussion: http://postgr.es/m/20210802224204.bckcikl45uezv5e4@alap3.anarazel.de
Commit 8431e296ea reworked ProcArrayApplyRecoveryInfo to sort XIDs
before adding them to KnownAssignedXids. But the XIDs are sorted using
xidComparator, which compares the XIDs simply as uint32 values, not
logically. KnownAssignedXidsAdd() however expects XIDs in logical order,
and calls TransactionIdFollowsOrEquals() to enforce that. If there are
XIDs for which the two orderings disagree, an error is raised and the
recovery fails/restarts.
Hitting this issue is fairly easy - you just need two transactions, one
started before the 4B limit (e.g. XID 4294967290), the other sometime
after it (e.g. XID 1000). Logically (4294967290 <= 1000) but when
compared using xidComparator we try to add them in the opposite order.
Which makes KnownAssignedXidsAdd() fail with an error like this:
ERROR: out-of-order XID insertion in KnownAssignedXids
This only happens during replica startup, while processing RUNNING_XACTS
records to build the snapshot. Once we reach STANDBY_SNAPSHOT_READY, we
skip these records. So this does not affect already running replicas,
but if you restart (or create) a replica while there are transactions
with XIDs for which the two orderings disagree, you may hit this.
Long-running transactions and frequent replica restarts increase the
likelihood of hitting this issue. Once the replica gets into this state,
it can't be started (even if the old transactions are terminated).
Fixed by sorting the XIDs logically - this is fine because we're dealing
with normal XIDs (because it's XIDs assigned to backends) and from the
same wraparound epoch (otherwise the backends could not be running at
the same time on the primary node). So there are no problems with the
triangle inequality, which is why xidComparator compares raw values.
Investigation and root cause analysis by Abhijit Menon-Sen. Patch by me.
This issue is present in all releases since 9.4, however releases up to
9.6 are EOL already so backpatch to 10 only.
Reviewed-by: Abhijit Menon-Sen
Reviewed-by: Alvaro Herrera
Backpatch-through: 10
Discussion: https://postgr.es/m/36b8a501-5d73-277c-4972-f58a4dce088a%40enterprisedb.com
While determining xid horizons, we skip over backends that are running
Vacuum. We also ignore Create Index Concurrently, or Reindex Concurrently
for the purposes of computing Xmin for Vacuum. But we were not setting the
flags corresponding to these operations when they are performed in
parallel which was preventing Xid horizon from advancing.
The optimization related to skipping Create Index Concurrently, or Reindex
Concurrently operations was implemented in PG-14 but the fix is the same
for the Parallel Vacuum as well so back-patched till PG-13.
Author: Masahiko Sawada
Reviewed-by: Amit Kapila
Backpatch-through: 13
Discussion: https://postgr.es/m/CAD21AoCLQqgM1sXh9BrDFq0uzd3RBFKi=Vfo6cjjKODm0Onr5w@mail.gmail.com
