Teach xlogreader.c to decode the WAL into a circular buffer. This will
support optimizations based on looking ahead, to follow in a later
commit.
* XLogReadRecord() works as before, decoding records one by one, and
allowing them to be examined via the traditional XLogRecGetXXX()
macros and certain traditional members like xlogreader->ReadRecPtr.
* An alternative new interface XLogReadAhead()/XLogNextRecord() is
added that returns pointers to DecodedXLogRecord objects so that it's
now possible to look ahead in the WAL stream while replaying.
* In order to be able to use the new interface effectively while
streaming data, support is added for the page_read() callback to
respond to a new nonblocking mode with XLREAD_WOULDBLOCK instead of
waiting for more data to arrive.
No direct user of the new interface is included in this commit, though
XLogReadRecord() uses it internally. Existing code doesn't need to
change, except in a few places where it was accessing reader internals
directly and now needs to go through accessor macros.
Reviewed-by: Julien Rouhaud <rjuju123@gmail.com>
Reviewed-by: Tomas Vondra <tomas.vondra@enterprisedb.com>
Reviewed-by: Andres Freund <andres@anarazel.de> (earlier versions)
Discussion: https://postgr.es/m/CA+hUKGJ4VJN8ttxScUFM8dOKX0BrBiboo5uz1cq=AovOddfHpA@mail.gmail.com
This paragraph has been added in the documentation by f6f0db4, but after
more discussion we found that this just makes things more confusing,
adding some cross-references between a general feature and something
only aimed at being used by developers. The original documentation is
not wrong either, and this commit brings back this part of the docs to
the same state as before f6f0db4.
Per discussion with Kyotaro Horiguchi and Thomas Munro.
Discussion: https://postgr.es/m/CA+hUKGL2uaRKu=3+bMBpejHh4k7wqzWC05aiasTsSsHGRCWa8g@mail.gmail.com
lz4frame.h was getting declared after the headers specific to Postgres,
but it needs to be included between postgres_fe.h and the internal
headers.
Issue introduced by babbbb5.
Reported-by: Justin Prysby
Discussion: https://postgr.es/m/20220317111220.GI28503@telsasoft.com
If a RowExpr is marked as returning a named composite type, we aren't
going to consult its colnames list; we'll use the attribute names
shown for the type in pg_attribute. Hence, skip storing that list,
to save a few nanoseconds when copying the expression tree around.
Discussion: https://postgr.es/m/2950001.1638729947@sss.pgh.pa.us
In commit bf7ca1587, I had the bright idea that we could make the
result of a whole-row Var (that is, foo.*) track any column aliases
that had been applied to the FROM entry the Var refers to. However,
that's not terribly logically consistent, because now the output of
the Var is no longer of the named composite type that the Var claims
to emit. bf7ca1587 tried to handle that by changing the output
tuple values to be labeled with a blessed RECORD type, but that's
really pretty disastrous: we can wind up storing such tuples onto
disk, whereupon they're not readable by other sessions.
The only practical fix I can see is to give up on what bf7ca1587
tried to do, and say that the column names of tuples produced by
a whole-row Var are always those of the underlying named composite
type, query aliases or no. While this introduces some inconsistencies,
it removes others, so it's not that awful in the abstract. What *is*
kind of awful is to make such a behavioral change in a back-patched
bug fix. But corrupt data is worse, so back-patched it will be.
(A workaround available to anyone who's unhappy about this is to
introduce an extra level of sub-SELECT, so that the whole-row Var is
referring to the sub-SELECT's output and not to a named table type.
Then the Var is of type RECORD to begin with and there's no issue.)
Per report from Miles Delahunty. The faulty commit dates to 9.5,
so back-patch to all supported branches.
Discussion: https://postgr.es/m/2950001.1638729947@sss.pgh.pa.us
Restructure things so that the functions which update the global
variables shared_map and local_map are separate from the functions
which just read and write relation map files without touching any
global variables.
In the new structure of things, write_relmap_file() writes a relmap
file but no longer performs global variable updates. A symmetric
function read_relmap_file() that just reads a file without changing
any global variables is added, and load_relmap_file(), which does
change the global variables, uses it as a subroutine.
Because write_relmap_file() no longer updates shared_map and
local_map, that logic is moved to perform_relmap_update(). However,
no similar logic is added to relmap_redo() even though it also calls
write_relmap_file(). That's because recovery must not rely on the
contents of the relation map, and therefore there is no need to
initialize it. In fact, doing so seems like a mistake, because we
might then manage to rely on the in-memory map where we shouldn't.
Patch by me, based on earlier work by Dilip Kumar. Reviewed by
Ashutosh Sharma.
Discussion: http://postgr.es/m/CA+TgmobQLgrt4AXsc0ru7aFFkzv=9fS-Q_yO69=k9WY67RCctg@mail.gmail.com
When publishing changes through a artition root, we should use the row
filter for the top-most ancestor. The relation may be added to multiple
publications, using different ancestors, and 52e4f0cd47 handled this
incorrectly. With c91f71b9dc we find the correct top-most ancestor, but
the code tried to fetch the row filter from all publications, including
those using a different ancestor etc. No row filter can be found for
such publications, which was treated as replicating all rows.
Similarly to c91f71b9dc, this seems to be a rare issue in practice. It
requires multiple publications including the same partitioned relation,
through different ancestors.
Fixed by only passing publications containing the top-most ancestor to
pgoutput_row_filter_init(), so that treating a missing row filter as
replicating all rows is correct.
Report and fix by me, test case by Hou zj. Reviews and improvements by
Amit Kapila.
Author: Tomas Vondra, Hou zj, Amit Kapila
Reviewed-by: Amit Kapila, Hou zj
Discussion: https://postgr.es/m/d26d24dd-2fab-3c48-0162-2b7f84a9c893%40enterprisedb.com
Create subroutines ExecUpdatePrologue / ExecUpdateAct /
ExecUpdateEpilogue, and similar for ExecDelete.
Introduce a new struct to be used internally in nodeModifyTable.c,
dubbed ModifyTableContext, which contains all context information needed
to perform these operations, as well as ExecInsert and others.
This allows using a different schedule and a different way of evaluating
the results of these operations, which can be exploited by a later
commit introducing support for MERGE. It also makes ExecUpdate and
ExecDelete proper shorter and (hopefully) simpler.
Author: Álvaro Herrera <alvherre@alvh.no-ip.org>
Reviewed-by: Amit Langote <amitlangote09@gmail.com>
Reviewed-by: Japin Li <japinli@hotmail.com>
Reviewed-by: Zhihong Yu <zyu@yugabyte.com>
Discussion: https://postgr.es/m/202202271724.4z7xv3cf46kv@alvherre.pgsql
This adds the option to use ICU as the default locale provider for
either the whole cluster or a database. New options for initdb,
createdb, and CREATE DATABASE are used to select this.
Since some (legacy) code still uses the libc locale facilities
directly, we still need to set the libc global locale settings even if
ICU is otherwise selected. So pg_database now has three
locale-related fields: the existing datcollate and datctype, which are
always set, and a new daticulocale, which is only set if ICU is
selected. A similar change is made in pg_collation for consistency,
but in that case, only the libc-related fields or the ICU-related
field is set, never both.
Reviewed-by: Julien Rouhaud <rjuju123@gmail.com>
Discussion: https://www.postgresql.org/message-id/flat/5e756dd6-0e91-d778-96fd-b1bcb06c161a%402ndquadrant.com
Using this system function with an in-place tablespace (created when
allow_in_place_tablespaces is enabled by specifying an empty string as
location) caused a failure when using readlink(), as the tablespace is,
in this case, not a symbolic link in pg_tblspc/ but a directory.
Rather than getting a failure, the commit changes
pg_tablespace_location() so as a relative path to the data directory is
returned for in-place tablespaces, to make a difference between
tablespaces created when allow_in_place_tablespaces is enabled or not.
Getting a path rather than an empty string that would match the CREATE
TABLESPACE command in this case is more useful for tests that would like
to rely on this function.
While on it, a regression test is added for this case. This is simple
to add in the main regression test suite thanks to regexp_replace() to
mask the part of the tablespace location dependent on its OID.
Author: Michael Paquier
Reviewed-by: Kyotaro Horiguchi, Thomas Munro
Discussion: https://postgr.es/m/YiG1RleON1WBcLnX@paquier.xyz
Commit 83fd4532a7 allowed publishing of changes via ancestors, for
publications defined with publish_via_partition_root. But the way
the ancestor was determined in get_rel_sync_entry() was incorrect,
simply updating the same variable. So with multiple publications,
replicating different ancestors, the outcome depended on the order
of publications in the list - the value from the last loop was used,
even if it wasn't the top-most ancestor.
This is a probably rare situation, as in most cases publications do
not overlap, so each partition has exactly one candidate ancestor
to replicate as and there's no ambiguity.
Fixed by tracking the "ancestor level" for each publication, and
picking the top-most ancestor. Adds a test case, verifying the
correct ancestor is used for publishing the changes and that this
does not depend on order of publications in the list.
Older releases have another bug in this loop - once all actions are
replicated, the loop is terminated, on the assumption that inspecting
additional publications is unecessary. But that misses the fact that
those additional applications may replicate different ancestors.
Fixed by removal of this break condition. We might still terminate the
loop in some cases (e.g. when replicating all actions and the ancestor
is the partition root).
Backpatch to 13, where publish_via_partition_root was introduced.
Initial report and fix by me, test added by Hou zj. Reviews and
improvements by Amit Kapila.
Author: Tomas Vondra, Hou zj, Amit Kapila
Reviewed-by: Amit Kapila, Hou zj
Discussion: https://postgr.es/m/d26d24dd-2fab-3c48-0162-2b7f84a9c893%40enterprisedb.com
The mkdir command in the Installation from Source Short Version docs
didn't use the -p intermediate path creation parameter which likely
would cause the command to fail. At the time of writing, -p wasn't
universally available but it can now be relied upon existing. The -p
parameter is defined by POSIX, at least since posix-2004.
Reported-by: Daniel Westermann <daniel.westermann@dbi-services.com>
Reviewed-by: Tom Lane <tgl@sss.pgh.pa.us>
Discussion: https://postgr.es/m/ZR0P278MB0920263E7F2D546A33E50079D20E9@ZR0P278MB0920.CHEP278.PROD.OUTLOOK.COM
911e702077 implemented operator class parameters including the signature length
in ltree. Previously, the signature length for gist_ltree_ops was 8. Because
of bug 911e702077 the default signature length for gist_ltree_ops became 28 for
ltree 1.1 (where options method is NOT provided) and 8 for ltree 1.2 (where
options method is provided). This commit changes the default signature length
for ltree 1.1 to 8.
Existing gist_ltree_ops indexes might be corrupted in various scenarios.
Thus, we have to recommend reindexing all the gist_ltree_ops indexes after
the upgrade.
Reported-by: Victor Yegorov
Reviewed-by: Tomas Vondra, Tom Lane, Andres Freund, Nikita Glukhov
Reviewed-by: Andrew Dunstan
Author: Tomas Vondra, Alexander Korotkov
Discussion: https://postgr.es/m/17406-71e02820ae79bb40%40postgresql.org
Discussion: https://postgr.es/m/d80e0a55-6c3e-5b26-53e3-3c4f973f737c%40enterprisedb.com
Commands like ALTER TABLE SET TABLESPACE may leave files for the next
checkpoint to clean up. If such files are not removed by the time DROP
TABLESPACE is called, we request a checkpoint so that they are deleted.
However, there is presently a window before checkpoint start where new
unlink requests won't be scheduled until the following checkpoint. This
means that the checkpoint forced by DROP TABLESPACE might not remove the
files we expect it to remove, and the following ERROR will be emitted:
ERROR: tablespace "mytblspc" is not empty
To fix, add a call to AbsorbSyncRequests() just before advancing the
unlink cycle counter. This ensures that any unlink requests forwarded
prior to checkpoint start (i.e., when ckpt_started is incremented) will
be processed by the current checkpoint. Since AbsorbSyncRequests()
performs memory allocations, it cannot be called within a critical
section, so we also need to move SyncPreCheckpoint() to before
CreateCheckPoint()'s critical section.
This is an old bug, so back-patch to all supported versions.
Author: Nathan Bossart <nathandbossart@gmail.com>
Reported-by: Nathan Bossart <nathandbossart@gmail.com>
Reviewed-by: Thomas Munro <thomas.munro@gmail.com>
Reviewed-by: Andres Freund <andres@anarazel.de>
Discussion: https://postgr.es/m/20220215235845.GA2665318%40nathanxps13
This caused the function to fail, as the aligned copy of the raw page
given by the function caller was not saved in the correct memory
context, which needs to be multi_call_memory_ctx in this case.
Issue introduced by 076f4d9.
Per buildfarm members sifika, mylodon and longfin. I have reproduced
that locally with macos.
Discussion: https://postgr.es/m/YjFPOtfCW6yLXUeM@paquier.xyz
Backpatch-through: 10
If we run out of space in the checkpointer sync request queue (which is
hopefully rare on real systems, but common with very small buffer pool),
we wait for it to drain. While waiting, we should report that as a wait
event so that users know what is going on, and also handle postmaster
death, since otherwise the loop might never terminate if the
checkpointer has exited.
Back-patch to 12. Although the problem exists in earlier releases too,
the code is structured differently before 12 so I haven't gone any
further for now, in the absence of field complaints.
Reported-by: Andres Freund <andres@anarazel.de>
Reviewed-by: Andres Freund <andres@anarazel.de>
Discussion: https://postgr.es/m/20220226213942.nb7uvb2pamyu26dj%40alap3.anarazel.de
This commit fixes a set of issues related to the use of the SQL
functions in this module when the caller is able to pass down raw page
data as input argument:
- The page size check was fuzzy in a couple of places, sometimes
looking after only a sub-range, but what we are looking for is an exact
match on BLCKSZ. After considering a few options here, I have settled
down to do a generalization of get_page_from_raw(). Most of the SQL
functions already used that, and this is not strictly required if not
accessing an 8-byte-wide value from a raw page, but this feels safer in
the long run for alignment-picky environment, particularly if a code
path begins to access such values. This also reduces the number of
strings that need to be translated.
- The BRIN function brin_page_items() uses a Relation but it did not
check the access method of the opened index, potentially leading to
crashes. All the other functions in need of a Relation already did
that.
- Some code paths could fail on elog(), but we should to use ereport()
for failures that can be triggered by the user.
Tests are added to stress all the cases that are fixed as of this
commit, with some junk raw pages (\set VERBOSITY ensures that this works
across all page sizes) and unexpected index types when functions open
relations.
Author: Michael Paquier, Justin Prysby
Discussion: https://postgr.es/m/20220218030020.GA1137@telsasoft.com
Backpatch-through: 10
The checkpointer shouldn't ignore its latch. Other backends may be
waiting for it to drain the request queue. Hopefully real systems don't
have a full queue often, but the condition is reached easily when
shared_buffers is small.
This involves defining a new wait event, which will appear in the
pg_stat_activity view often due to spread checkpoints.
Back-patch only to 14. Even though the problem exists in earlier
branches too, it's hard to hit there. In 14 we stopped using signal
handlers for latches on Linux, *BSD and macOS, which were previously
hiding this problem by interrupting the sleep (though not reliably, as
the signal could arrive before the sleep begins; precisely the problem
latches address).
Reported-by: Andres Freund <andres@anarazel.de>
Reviewed-by: Andres Freund <andres@anarazel.de>
Discussion: https://postgr.es/m/20220226213942.nb7uvb2pamyu26dj%40alap3.anarazel.de
As a demonstration of the sort of thing that can be done by adding a
custom backup target, this defines a 'shell' target which executes a
command defined by the system administrator. The command is executed
once for each tar archive generate by the backup and once for the
backup manifest, if any. Each time the command is executed, it
receives the contents of th file for which it is executed via standard
input.
The configured command can use %f to refer to the name of the archive
(e.g. base.tar, $TABLESPACE_OID.tar, backup_manifest) and %d to refer
to the target detail (pg_basebackup --target shell:DETAIL). A target
detail is required if %d appears in the configured command and
forbidden if it does not.
Patch by me, reviewed by Abhijit Menon-Sen.
Discussion: http://postgr.es/m/CA+TgmoaqvdT-u3nt+_kkZ7bgDAyqDB0i-+XOMmr5JN2Rd37hxw@mail.gmail.com
Commit 3500ccc39b allowed for base backup
targets, meaning that we could do something with the backup other than
send it to the client, but all of those targets had to be baked in to
the core code. This commit makes it possible for extensions to define
additional backup targets.
Patch by me, reviewed by Abhijit Menon-Sen.
Discussion: http://postgr.es/m/CA+TgmoaqvdT-u3nt+_kkZ7bgDAyqDB0i-+XOMmr5JN2Rd37hxw@mail.gmail.com
These tests were added recently, but older code tests USE_LZ4 rathr
than HAVE_LIBLZ4, so let's follow the established precedent. It
also seems more consistent with the intent of the configure tests,
since I think that the USE_* symbols are intended to correspond to
what the user requested, and the HAVE_* symbols to what configure
found while probing.
Discussion: http://postgr.es/m/CA+Tgmoap+hTD2-QNPJLH4tffeFE8MX5+xkbFKMU3FKBy=ZSNKA@mail.gmail.com
This system function was being triggered once in the main regression
test suite to check its SRF configuration, and more in other test
modules but nothing checked the behavior of the options missing_ok and
include_dot_dirs. This commit adds some tests for both options, to
avoid mistakes if this code is manipulated in the future.
Extracted from a larger patch by the same author, with a few tweaks by
me.
Author: Justin Pryzby
Discussion: https://postgr.es/m/20191227170220.GE12890@telsasoft.com
Previously, pg_basebackup from a cluster that contained an 'in-place'
tablespace, as introduced by commit 7170f215, would produce a harmless
warning on Unix and fail completely on Windows.
Reported-by: Kyotaro Horiguchi <horikyota.ntt@gmail.com>
Reviewed-by: Kyotaro Horiguchi <horikyota.ntt@gmail.com>
Reviewed-by: Michael Paquier <michael@paquier.xyz>
Discussion: https://postgr.es/m/20220304.165449.1200020258723305904.horikyota.ntt%40gmail.com
The upper case versions "OF", "TZH", and "TZM" are already supported,
and all other format codes that are supported in upper case are also
supported in lower case, so we should support these as well for
consistency.
Nitin Jadhav, with a tiny cosmetic change by me. Reviewed by Suraj
Kharage and David Zhang.
Discussion: http://postgr.es/m/CAMm1aWZ-oZyKd75+8D=VJ0sAoSwtdXWLP-MAWD4D8R1Dgandzw@mail.gmail.com
Logical replication apply workers for a subscription can easily get stuck
in an infinite loop of attempting to apply a change, triggering an error
(such as a constraint violation), exiting with the error written to the
subscription server log, and restarting.
To partially remedy the situation, this patch adds a new subscription
option named 'disable_on_error'. To be consistent with old behavior, this
option defaults to false. When true, both the tablesync worker and apply
worker catch any errors thrown and disable the subscription in order to
break the loop. The error is still also written in the logs.
Once the subscription is disabled, users can either manually resolve the
conflict/error or skip the conflicting transaction by using
pg_replication_origin_advance() function. After resolving the conflict,
users need to enable the subscription to allow apply process to proceed.
Author: Osumi Takamichi and Mark Dilger
Reviewed-by: Greg Nancarrow, Vignesh C, Amit Kapila, Wang wei, Tang Haiying, Peter Smith, Masahiko Sawada, Shi Yu
Discussion : https://postgr.es/m/DB35438F-9356-4841-89A0-412709EBD3AB%40enterprisedb.com
Commit 872770fd6c taught VACUUM VERBOSE and autovacuum logging to
display the total number of pages scanned by VACUUM. This information
was also displayed as a percentage of rel_pages in parenthesis, which
makes it easy to spot trends over time and across tables.
The instrumentation displayed "0 scanned (0.00% of total)" for totally
empty tables. Tweak the instrumentation: have it show "0 scanned
(100.00% of total)" for empty tables instead. This approach is clearer
and more consistent.
Starting in cc50080a82 create_index test fails when run with
synchronous_commit=off. synchronous_commit=off delays when hint bits may be
set. Some plans change depending on the number of all-visible pages, which in
turn can be influenced by the delayed hint bits.
Force synchronous_commit to `on` in test_setup.sql. Not very satisfying, but
there's no obvious alternative.
Reported-By: Aleksander Alekseev <aleksander@timescale.com>
Author: Andres Freund <andres@anarazel.de>
Author: Aleksander Alekseev <aleksander@timescale.com>
Discussion: https://www.postgresql.org/message-id/flat/CAJ7c6TPJNof1Q+vJsy3QebgbPgXdu2ErPvYkBdhD6_Ckv5EZRg@mail.gmail.com
VACUUM's rel_pages field indicates the size of the target heap rel just
after the table_relation_vacuum() operation began. There are specific
expectations around how rel_pages can be related to other nearby state.
In particular, the range of rel_pages must contain every tuple in the
relation whose tuple headers might contain an XID < OldestXmin.
Consistently refer to the field as rel_pages to make this clearer and
more discoverable.
This is follow-up work to commit 73f6ec3d from earlier today.
Author: Peter Geoghegan <pg@bowt.ie>
Reviewed-By: Andres Freund <andres@anarazel.de>
Discussion: https://postgr.es/m/20220311031351.sbge5m2bpvy2ttxg@alap3.anarazel.de
Explain the relationship between vacuumlazy.c's vistest and OldestXmin
cutoffs. These closely related cutoffs are different in subtle but
important ways. Also document a closely related rule: we must establish
rel_pages _after_ OldestXmin to ensure that no XID < OldestXmin can be
missed by lazy_scan_heap().
It's easier to explain these issues by initializing everything together,
so consolidate initialization of vacrel state. Now almost every vacrel
field is initialized by heap_vacuum_rel(). The only remaining exception
is the dead_items array, which is still managed by lazy_scan_heap() due
to interactions with how we initialize parallel VACUUM.
Also move the process that updates pg_class entries for each index into
heap_vacuum_rel(), and adjust related assertions. All pg_class updates
now take place after lazy_scan_heap() returns, which seems clearer.
Author: Peter Geoghegan <pg@bowt.ie>
Reviewed-By: Andres Freund <andres@anarazel.de>
Discussion: https://postgr.es/m/20211211045710.ljtuu4gfloh754rs@alap3.anarazel.de
Discussion: https://postgr.es/m/CAH2-WznYsUxVT156rCQ+q=YD4S4=1M37hWvvHLz-H1pwSM8-Ew@mail.gmail.com
We called the argument totally_frozen in its function prototype as well
as in code comments, even though totally_frozen_p was used in the
function definition. Standardize on totally_frozen.
Commit 8b069ef5d changed this function to look at pg_constraint.conindid
rather than searching pg_depend. That was a good performance improvement,
but it failed to preserve the exact semantics. The old code would only
return an index that was "owned by" (internally dependent on) the
specified constraint, whereas the new code will also return indexes that
are just referenced by foreign key constraints. This confuses ALTER
TABLE, which was implicitly expecting the previous semantics, into
failing with errors like
ERROR: relation 146621 has multiple clustered indexes
or
ERROR: "pk_attbl" is not an index for table "atref"
We can fix this without reverting the performance improvement by adding
a contype check in get_constraint_index(). Another way could be to
make ALTER TABLE check it, but I'm worried that extension code could
also have subtle dependencies on the old semantics.
Tom Lane and Japin Li, per bug #17409 from Holly Roberts.
Back-patch to v14 where the error crept in.
Discussion: https://postgr.es/m/17409-52871dda8b5741cb@postgresql.org
Fail with a suitable error message instead. We can't inject the backup
manifest into the output tarfile without decompressing it, and if
we did that, we'd have to recompress the tarfile afterwards to produce
the result the user is expecting. While we have enough infrastructure
in pg_basebackup now to accomplish that whole series of steps without
much additional code, it seems like excessively surprising behavior.
The user probably did not select server-side compression with the idea
that the client was going to end up decompressing it and then
recompressing.
Report from Justin Pryzby. Fix by me.
Discussion: http://postgr.es/m/CA+Tgmob6Rnjz-Qv32h3yJn8nnUkLhrtQDAS4y5AtsgtorAFHRA@mail.gmail.com
"Hot Standby" was capitalized in a couple of places in the docs, as the
style primarily used when it was introduced, but this has not been much
respected across the years. Per discussion, it is more natural for the
reader to use "hot standby" (aka lower-case only) when in the middle of
a sentence, and "Hot standby" (aka capitalized) in a title. This commit
adjusts all the places in the docs to be consistent with this choice,
rather than applying one style or the other midway.
Author: Daniel Westermann
Reviewed-by: Kyotaro Horiguchi, Aleksander Alekseev, Robert Treat
Discussion: https://postgr.es/m/GVAP278MB093160025A779A1A5788D0EAD2039@GVAP278MB0931.CHEP278.PROD.OUTLOOK.COM
wal_compression gains a new value, "zstd", to allow the compression of
full-page images using the compression method of the same name.
Compression is done using the default level recommended by the library,
as of ZSTD_CLEVEL_DEFAULT = 3. Some benchmarking has shown that it
could make sense to use a level lower for the FPI compression, like 1 or
2, as the compression rate did not change much with a bit less CPU
consumed, but any tests done would only cover few scenarios so it is
hard to come to a clear conclusion. Anyway, there is no reason to not
use the default level instead, which is the level recommended by the
library so it should be fine for most cases.
zstd outclasses easily pglz, and is better than LZ4 where one wants to
have more compression at the cost of extra CPU but both are good enough
in their own scenarios, so the choice between one or the other of these
comes to a study of the workload patterns and the schema involved,
mainly.
This commit relies heavily on 4035cd5, that reshaped the code creating
and restoring full-page writes to be aware of the compression type,
making this integration straight-forward.
This patch borrows some early work from Andrey Borodin, though the patch
got a complete rewrite.
Author: Justin Pryzby
Discussion: https://postgr.es/m/20220222231948.GJ9008@telsasoft.com
Per project policy, all system and library headers need to be declared
in the backend code after "postgres.h" and before the internal headers,
but 4035cd5 broke this policy when adding support for LZ4 in
wal_compression.
Noticed while reviewing the patch to add support for zstd in this area.
This only impacts HEAD, so there is no need for a back-patch.
Tom Lane
Thomas Munro
Michael Paquier
Robert Haas
Tomas Vondra
Peter Eisentraut
Alvaro Herrera
Daniel Gustafsson
Alexander Korotkov
Amit Kapila
Peter Geoghegan
Andres Freund