Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Make it possible to detemine WAL format at runtime #304

Open
wants to merge 56 commits into
base: REL_15_STABLE_neon
Choose a base branch
from
Open
Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
56 commits
Select commit Hold shift + click to select a range
0fc0e97
Neon rebased to REL_15_STABLE (up to 0a9045c9ff)
lubennikovaav Jul 22, 2022
8af27f3
fix regression tests
lubennikovaav Aug 10, 2022
c146655
rebase to the latest origin and resolve conflicts
lubennikovaav Aug 16, 2022
a41058d
Remove contrib neon and neon_test_utils.
lubennikovaav Aug 24, 2022
e06d8c7
Prevent access to uninitalized shaerd memory in InstallXLogFileSegmen…
lubennikovaav Aug 24, 2022
9b2b574
Remove Dockerfile, it's now in the neon repo
lubennikovaav Aug 29, 2022
261fc36
Move backpressure throttling implementation to neon extension (#203)
knizhnik Sep 2, 2022
f5cb05b
Merge last written cache lsn with new main branch (#201)
knizhnik Sep 4, 2022
80512f1
Local prefetch implementation for Postgres 15
hlinnaka Sep 13, 2022
54f74ad
Set last written LSN for the created relation (#212)
hlinnaka Sep 16, 2022
fb19b2c
Use normal install program to install server headers.
hlinnaka Sep 12, 2022
5079178
Update expected output for sysviews test because of changed default v…
knizhnik Sep 12, 2022
95f6956
Undo diasming VM check warning in vacuumlazy.c (#214)
knizhnik Sep 20, 2022
561af18
Set Neon-specific FMGR_ABI_EXTRA
lubennikovaav Sep 21, 2022
4ac76c3
Don't use newline in PG_VERSION file.
lubennikovaav Sep 21, 2022
a771474
Unset ArchiveRecoveryRequested for Neon code path.
lubennikovaav Sep 21, 2022
faa68d0
Fix memory leak in ApplyRecord
knizhnik Sep 21, 2022
808318e
Rebase to Stamp 15.0
lubennikovaav Oct 5, 2022
52da928
Pin pages with speculative insert tuples to prevent their reconstruct…
knizhnik Oct 20, 2022
534b38a
Fix shared memory initialization for last written LSN cache (#226)
knizhnik Oct 24, 2022
887fd35
Fix upper boundary caculation in the chunks loop in SetLastWrittenLSN…
knizhnik Oct 28, 2022
e336bc9
Stamp XLP_FIRST_IS_CONTRECORD only if we start writing with page offset.
arssher Oct 28, 2022
a1066f4
Move walredo process code under pgxn in the main 'neon' repository.
hlinnaka Oct 31, 2022
f34afc9
Misc cleanup, mostly to reduce unnecessary differences with upstream.
hlinnaka Oct 31, 2022
87c5bb6
Optimize prefetch patterns in both heap seqscan and vacuum scans. (#228)
MMeent Nov 7, 2022
ddd669f
Fix prefetch issues in parallel scans and vacuum's cleanup scan (#235)
MMeent Nov 11, 2022
90c4f24
Fix expected results for regression tests (#238)
bayandin Nov 15, 2022
890a699
Use prefetch in pg_prewarm extension (#237)
knizhnik Nov 17, 2022
6fce20c
PG15: Prefetch cleanup (#241)
MMeent Nov 23, 2022
debb925
Drop unlogged table in regress test to avoid noise in tests
lubennikovaav Nov 21, 2022
85bd2ad
Do not produce open file error for unlogged relations (#240)
knizhnik Nov 24, 2022
6f72c84
Maintain last written LSN for each page to enable prefetch on vacuum,…
knizhnik Nov 24, 2022
dcdb9ee
Prefetch cleanup: (#246)
MMeent Dec 5, 2022
fc754e5
Set lsn fix v15 (#252)
knizhnik Dec 8, 2022
7cb2db7
Show prefetch statistic in EXPLAIN (#249)
knizhnik Jan 12, 2023
a52fb5a
Implement efficient prefetch for parallel bitmap heap scan (#258)
knizhnik Feb 14, 2023
9fc4107
Unlogged index fix v15 (#262)
knizhnik Feb 22, 2023
7b55f30
Fix bitmap scan prefetch (#261)
knizhnik Feb 23, 2023
513b58c
Revert handling of UNLOGGED tables on compute side v15.
arssher Feb 24, 2023
6da0174
Allow external main functions to skip config load and make last
kelvich Feb 23, 2023
a029c23
Remove walredo-related hacks from InternalIpcMemoryCreate()
kelvich Mar 6, 2023
78481ee
Adjust prefetch target for parallel bitmap scan (#274)
knizhnik Mar 27, 2023
8a6fd67
Copy iterator result in BitmapHeapNext (#276)
knizhnik Apr 7, 2023
0e771f9
Prefetch for index and index-only scans (#271)
knizhnik Apr 7, 2023
52c0264
[PG15] Feature/replicas (#279)
MMeent Apr 13, 2023
48ebdaf
Fix entering hot standby mode for Neon
lubennikovaav May 5, 2023
ac6ca4d
Do not allow users with CREATEROLE privelege to manage system user
kelvich Jul 7, 2023
38e3bd2
Fix regression tests after the patch with CREATEROLE restrictions
kelvich Jul 15, 2023
f93c725
Add startup logs (#293)
bojanserafimov Jul 17, 2023
4834eb7
Make it possible to grant self created roles (#298)
knizhnik Jul 27, 2023
362e841
Update expected file for create_role test (#301)
knizhnik Jul 28, 2023
fcd0bde
Define NEON_SMGR in smgr.h to make it possible for extensions to use …
knizhnik Jul 29, 2023
026d6b0
Request extension files and libraries from compute_ctl
lubennikovaav Jun 13, 2023
bd06d52
Make it possible to detemine WAL format at runtime
Aug 14, 2023
f6ea27b
Make handling t_cid more clear and less erro prone and move t_cid fro…
Aug 14, 2023
2c76abf
Support work with both multi_insert record formats in compatibility mode
Aug 15, 2023
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
5 changes: 5 additions & 0 deletions .dockerignore
Original file line number Diff line number Diff line change
@@ -0,0 +1,5 @@
.git
.vscode
.circleci
tmp_install
compute_build
86 changes: 86 additions & 0 deletions configure
Original file line number Diff line number Diff line change
Expand Up @@ -717,6 +717,7 @@ with_libxml
with_uuid
with_readline
with_systemd
with_libseccomp
with_selinux
with_ldap
with_krb_srvnam
Expand Down Expand Up @@ -864,6 +865,7 @@ with_bsd_auth
with_ldap
with_bonjour
with_selinux
with_libseccomp
with_systemd
with_readline
with_libedit_preferred
Expand Down Expand Up @@ -1573,6 +1575,7 @@ Optional Packages:
--with-ldap build with LDAP support
--with-bonjour build with Bonjour support
--with-selinux build with SELinux support
--with-libseccomp build with libseccomp support
--with-systemd build with systemd support
--without-readline do not use GNU Readline nor BSD Libedit for editing
--with-libedit-preferred
Expand Down Expand Up @@ -8631,6 +8634,39 @@ fi
{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $with_selinux" >&5
$as_echo "$with_selinux" >&6; }

#
# libseccomp
#
{ $as_echo "$as_me:${as_lineno-$LINENO}: checking whether to build with libseccomp support" >&5
$as_echo_n "checking whether to build with libseccomp support... " >&6; }



# Check whether --with-libseccomp was given.
if test "${with_libseccomp+set}" = set; then :
withval=$with_libseccomp;
case $withval in
yes)
:
;;
no)
:
;;
*)
as_fn_error $? "no argument expected for --with-libseccomp option" "$LINENO" 5
;;
esac

else
with_libseccomp=no

fi



{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $with_libseccomp" >&5
$as_echo "$with_libseccomp" >&6; }

#
# Systemd
#
Expand Down Expand Up @@ -14587,6 +14623,56 @@ else
fi


fi

if test "$with_libseccomp" = yes ; then
{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for seccomp_init in -lseccomp" >&5
$as_echo_n "checking for seccomp_init in -lseccomp... " >&6; }
if ${ac_cv_lib_seccomp_seccomp_init+:} false; then :
$as_echo_n "(cached) " >&6
else
ac_check_lib_save_LIBS=$LIBS
LIBS="-lseccomp $LIBS"
cat confdefs.h - <<_ACEOF >conftest.$ac_ext
/* end confdefs.h. */

/* Override any GCC internal prototype to avoid an error.
Use char because int might match the return type of a GCC
builtin and then its argument prototype would still apply. */
#ifdef __cplusplus
extern "C"
#endif
char seccomp_init ();
int
main ()
{
return seccomp_init ();
;
return 0;
}
_ACEOF
if ac_fn_c_try_link "$LINENO"; then :
ac_cv_lib_seccomp_seccomp_init=yes
else
ac_cv_lib_seccomp_seccomp_init=no
fi
rm -f core conftest.err conftest.$ac_objext \
conftest$ac_exeext conftest.$ac_ext
LIBS=$ac_check_lib_save_LIBS
fi
{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $ac_cv_lib_seccomp_seccomp_init" >&5
$as_echo "$ac_cv_lib_seccomp_seccomp_init" >&6; }
if test "x$ac_cv_lib_seccomp_seccomp_init" = xyes; then :
cat >>confdefs.h <<_ACEOF
#define HAVE_LIBSECCOMP 1
_ACEOF

LIBS="-lseccomp $LIBS"

else
as_fn_error $? "library 'libseccomp' is required for Seccomp BPF support" "$LINENO" 5
fi

fi

# for contrib/uuid-ossp
Expand Down
13 changes: 13 additions & 0 deletions configure.ac
Original file line number Diff line number Diff line change
Expand Up @@ -927,6 +927,14 @@ PGAC_ARG_BOOL(with, selinux, no, [build with SELinux support])
AC_SUBST(with_selinux)
AC_MSG_RESULT([$with_selinux])

#
# libseccomp
#
AC_MSG_CHECKING([whether to build with libseccomp support])
PGAC_ARG_BOOL(with, libseccomp, no, [build with libseccomp support])
AC_SUBST(with_libseccomp)
AC_MSG_RESULT([$with_libseccomp])

#
# Systemd
#
Expand Down Expand Up @@ -1613,6 +1621,11 @@ dnl If you want to use Apple's own Bonjour code on another platform,
dnl just add -ldns_sd to LIBS manually.
fi

if test "$with_libseccomp" = yes ; then
AC_CHECK_LIB(seccomp, seccomp_init, [],
[AC_MSG_ERROR([library 'libseccomp' is required for Seccomp BPF support])])
fi

# for contrib/uuid-ossp
if test "$with_uuid" = bsd ; then
AC_CHECK_HEADERS(uuid.h,
Expand Down
35 changes: 35 additions & 0 deletions contrib/pg_prewarm/autoprewarm.c
Original file line number Diff line number Diff line change
Expand Up @@ -54,6 +54,7 @@
#include "utils/rel.h"
#include "utils/relfilenodemap.h"
#include "utils/resowner.h"
#include "utils/spccache.h"

#define AUTOPREWARM_FILE "autoprewarm.blocks"

Expand Down Expand Up @@ -449,10 +450,12 @@ void
autoprewarm_database_main(Datum main_arg)
{
int pos;
int io_concurrency;
BlockInfoRecord *block_info;
Relation rel = NULL;
BlockNumber nblocks = 0;
BlockInfoRecord *old_blk = NULL;
BlockInfoRecord *prefetch_blk = NULL;
dsm_segment *seg;

/* Establish signal handlers; once that's done, unblock signals. */
Expand Down Expand Up @@ -499,6 +502,7 @@ autoprewarm_database_main(Datum main_arg)
{
relation_close(rel, AccessShareLock);
rel = NULL;
io_concurrency = -1;
CommitTransactionCommand();
}

Expand All @@ -518,6 +522,8 @@ autoprewarm_database_main(Datum main_arg)

if (!rel)
CommitTransactionCommand();
else
io_concurrency = get_tablespace_maintenance_io_concurrency(rel->rd_rel->reltablespace);
}
if (!rel)
{
Expand Down Expand Up @@ -550,6 +556,35 @@ autoprewarm_database_main(Datum main_arg)
continue;
}

/* if prefetching is enabled for this relation */
if (io_concurrency > 0)
{
/* make prefetch_blk catch up */
if (blk > prefetch_blk)
{
prefetch_blk = blk;
}

/* now, prefetch all following blocks */
while (prefetch_blk <= &block_info[apw_state->prewarm_stop_idx])
{
/* unless they're of a different relfilenode */
if (prefetch_blk->filenode != blk->filenode ||
prefetch_blk->forknum != blk->forknum ||
prefetch_blk->blocknum >= nblocks)
break;

/* or unless they are more than io_concurrency blocks ahead */
if (blk + io_concurrency <= prefetch_blk)
break;

PrefetchBuffer(rel, prefetch_blk->forknum, prefetch_blk->blocknum);

/* continue with the next block */
prefetch_blk++;
}
}

/* Prewarm buffer. */
buf = ReadBufferExtended(rel, blk->forknum, blk->blocknum, RBM_NORMAL,
NULL);
Expand Down
18 changes: 16 additions & 2 deletions contrib/pg_prewarm/pg_prewarm.c
Original file line number Diff line number Diff line change
Expand Up @@ -18,12 +18,14 @@
#include "access/relation.h"
#include "fmgr.h"
#include "miscadmin.h"
#include "optimizer/cost.h"
#include "storage/bufmgr.h"
#include "storage/smgr.h"
#include "utils/acl.h"
#include "utils/builtins.h"
#include "utils/lsyscache.h"
#include "utils/rel.h"
#include "utils/spccache.h"

PG_MODULE_MAGIC;

Expand Down Expand Up @@ -183,14 +185,26 @@ pg_prewarm(PG_FUNCTION_ARGS)
}
else if (ptype == PREWARM_BUFFER)
{
BlockNumber prefetch_block = first_block;
Oid nspOid;
int io_concurrency;

nspOid = rel->rd_rel->reltablespace;
io_concurrency = get_tablespace_maintenance_io_concurrency(nspOid);

/*
* In buffer mode, we actually pull the data into shared_buffers.
*/
for (block = first_block; block <= last_block; ++block)
{
Buffer buf;

Buffer buf;
BlockNumber prefetch_stop = block + Min(last_block - block + 1,
io_concurrency);
CHECK_FOR_INTERRUPTS();
while (prefetch_block < prefetch_stop)
{
PrefetchBuffer(rel, forkNumber, prefetch_block++);
}
buf = ReadBufferExtended(rel, forkNumber, block, RBM_NORMAL, NULL);
ReleaseBuffer(buf);
++blocks_done;
Expand Down
1 change: 1 addition & 0 deletions contrib/pg_prewarm/pg_prewarm.control
Original file line number Diff line number Diff line change
Expand Up @@ -3,3 +3,4 @@ comment = 'prewarm relation data'
default_version = '1.2'
module_pathname = '$libdir/pg_prewarm'
relocatable = true
trusted = true
1 change: 1 addition & 0 deletions src/Makefile.global.in
Original file line number Diff line number Diff line change
Expand Up @@ -186,6 +186,7 @@ with_tcl = @with_tcl@
with_ssl = @with_ssl@
with_readline = @with_readline@
with_selinux = @with_selinux@
with_libseccomp = @with_libseccomp@
with_systemd = @with_systemd@
with_gssapi = @with_gssapi@
with_krb_srvnam = @with_krb_srvnam@
Expand Down
3 changes: 2 additions & 1 deletion src/backend/access/brin/brin_xlog.c
Original file line number Diff line number Diff line change
Expand Up @@ -69,7 +69,8 @@ brin_xlog_insert_update(XLogReaderState *record,
}

/* need this page's blkno to store in revmap */
regpgno = BufferGetBlockNumber(buffer);
//ZENITH XXX Don't use BufferGetBlockNumber because wal-redo doesn't pin buffer.
XLogRecGetBlockTag(record, 0, NULL, NULL, &regpgno);

/* insert the index item into the page */
if (action == BLK_NEEDS_REDO)
Expand Down
8 changes: 8 additions & 0 deletions src/backend/access/gin/gininsert.c
Original file line number Diff line number Diff line change
Expand Up @@ -335,6 +335,8 @@ ginbuild(Relation heap, Relation index, IndexInfo *indexInfo)
elog(ERROR, "index \"%s\" already contains data",
RelationGetRelationName(index));

smgr_start_unlogged_build(index->rd_smgr);

initGinState(&buildstate.ginstate, index);
buildstate.indtuples = 0;
memset(&buildstate.buildStats, 0, sizeof(GinStatsData));
Expand Down Expand Up @@ -408,6 +410,8 @@ ginbuild(Relation heap, Relation index, IndexInfo *indexInfo)
buildstate.buildStats.nTotalPages = RelationGetNumberOfBlocks(index);
ginUpdateStats(index, &buildstate.buildStats, true);

smgr_finish_unlogged_build_phase_1(index->rd_smgr);

/*
* We didn't write WAL records as we built the index, so if WAL-logging is
* required, write all pages to the WAL now.
Expand All @@ -417,8 +421,12 @@ ginbuild(Relation heap, Relation index, IndexInfo *indexInfo)
log_newpage_range(index, MAIN_FORKNUM,
0, RelationGetNumberOfBlocks(index),
true);
SetLastWrittenLSNForBlockRange(XactLastRecEnd, index->rd_smgr->smgr_rnode.node, MAIN_FORKNUM, 0, RelationGetNumberOfBlocks(index));
SetLastWrittenLSNForRelation(XactLastRecEnd, index->rd_smgr->smgr_rnode.node, MAIN_FORKNUM);
}

smgr_end_unlogged_build(index->rd_smgr);

/*
* Return statistics
*/
Expand Down
19 changes: 13 additions & 6 deletions src/backend/access/gin/ginxlog.c
Original file line number Diff line number Diff line change
Expand Up @@ -407,6 +407,7 @@ ginRedoSplit(XLogReaderState *record)
rootbuf;
bool isLeaf = (data->flags & GIN_INSERT_ISLEAF) != 0;
bool isRoot = (data->flags & GIN_SPLIT_ROOT) != 0;
XLogRedoAction action;

/*
* First clear incomplete-split flag on child page if this finishes a
Expand All @@ -415,21 +416,27 @@ ginRedoSplit(XLogReaderState *record)
if (!isLeaf)
ginRedoClearIncompleteSplit(record, 3);

if (XLogReadBufferForRedo(record, 0, &lbuffer) != BLK_RESTORED)
action = XLogReadBufferForRedo(record, 0, &lbuffer);
if (action != BLK_RESTORED && action != BLK_DONE)
elog(ERROR, "GIN split record did not contain a full-page image of left page");

if (XLogReadBufferForRedo(record, 1, &rbuffer) != BLK_RESTORED)
action = XLogReadBufferForRedo(record, 1, &rbuffer);
if (action != BLK_RESTORED && action != BLK_DONE)
elog(ERROR, "GIN split record did not contain a full-page image of right page");

if (isRoot)
{
if (XLogReadBufferForRedo(record, 2, &rootbuf) != BLK_RESTORED)
action = XLogReadBufferForRedo(record, 2, &rootbuf);
if (action != BLK_RESTORED && action != BLK_DONE)
elog(ERROR, "GIN split record did not contain a full-page image of root page");
UnlockReleaseBuffer(rootbuf);
if (rootbuf != InvalidBuffer)
UnlockReleaseBuffer(rootbuf);
}

UnlockReleaseBuffer(rbuffer);
UnlockReleaseBuffer(lbuffer);
if (rbuffer != InvalidBuffer)
UnlockReleaseBuffer(rbuffer);
if (lbuffer != InvalidBuffer)
UnlockReleaseBuffer(lbuffer);
}

/*
Expand Down
Loading