On Thu, Mar 28, 2024 at 2:02 PM Thomas Munro wrote:
> ... In practice on a non-toy system, that's always going to be
> io_combine_limit. ...
And to be more explicit about that: you're right that we initialise
max_pinned_buffers such that it's usually at least io_combine_l
On Mon, Mar 25, 2024 at 2:02 AM Thomas Munro wrote:
> On Wed, Mar 20, 2024 at 4:04 AM Heikki Linnakangas wrote:
> > > /*
> > >* Skip the initial ramp-up phase if the caller says we're going to
> > > be
> > >* reading the
On Thu, Mar 28, 2024 at 10:52 AM Thomas Munro wrote:
> I think 1 is good, as a rescan is even more likely to find the pages
> in cache, and if that turns out to be wrong it'll very soon adjust.
Hmm, no I take that back, it probably won't be due to the
strategy/ring... I se
On Thu, Mar 28, 2024 at 9:43 AM Melanie Plageman
wrote:
> For sequential scan, I added a little reset function to the streaming
> read API (read_stream_reset()) that just releases all the buffers.
> Previously, it set finished to true before releasing the buffers (to
> indicate it was done) and th
On Wed, Mar 27, 2024 at 1:40 AM Heikki Linnakangas wrote:
> Is int16 enough though? It seems so, because:
>
> max_pinned_buffers = Max(max_ios * 4, buffer_io_size);
>
> and max_ios is constrained by the GUC's maximum MAX_IO_CONCURRENCY, and
> buffer_io_size is constrained by MAX_BUFFER_IO_SIZ
On Tue, Mar 26, 2024 at 3:34 AM Pankaj Raghav wrote:
> One question: Does ZFS do something like FUA request to force the device
> to clear the cache before it can update the node to point to the new page?
>
> If it doesn't do it, there is no guarantee from device to update the data
> atomically un
ublic functions _begin(), _next(),
_end() to be next to each other after the static helper functions.
Working on perf regression/tuning reports today, more soon...
From edd3d078cf8d4b0c2f08df82295825f7107ec62b Mon Sep 17 00:00:00 2001
From: Thomas Munro
Date: Mon, 26 Feb 2024 23:48:31 +1300
On Wed, Mar 20, 2024 at 4:04 AM Heikki Linnakangas wrote:
> On 12/03/2024 15:02, Thomas Munro wrote:
> > src/backend/storage/aio/streaming_read.c
> > src/include/storage/streaming_read.h
>
> Standard file header comments missing.
Fixed.
> It would be nice to have
On Fri, Mar 22, 2024 at 10:56 PM Pankaj Raghav (Samsung)
wrote:
> My team and I have been working on adding Large block size(LBS)
> support to XFS in Linux[1]. Once this feature lands upstream, we will be
> able to create XFS with FS block size > page size of the system on Linux.
> We also gave a
On Fri, Mar 8, 2024 at 6:53 AM Robert Haas wrote:
> But I think that's really only necessary if we're actually going to
> get rid of the idea of segmented relations altogether, which I don't
> think is happening at least for v17, and maybe not ever.
Yeah, I consider the feedback on ext4's size li
On Sat, Mar 23, 2024 at 6:26 AM Tom Lane wrote:
> conftest.c:139:5: error: no previous prototype for 'does_int64_work'
> [-Werror=missing-prototypes]
> 139 | int does_int64_work()
> | ^~~
> cc1: all warnings being treated as errors
> configure:17003: $? = 1
> configure: pr
Hmm, this discussion seems to assume that we only use
copy_file_range() to copy/clone whole segment files, right? That's
great and may even get most of the available benefit given typical
databases with many segments of old data that never changes, but... I
think copy_write_range() allows us to go
On Sun, Mar 17, 2024 at 8:10 AM Andres Freund wrote:
> I don't think zeroextend on the one hand and and on the other hand a normal
> write or extend are really the same operation. In the former case the content
> is hard-coded in the latter it's caller provided. Sure, we can deal with that
> by sp
On Tue, Mar 19, 2024 at 11:55 AM Tom Lane wrote:
> Jeff Davis writes:
> > On Mon, 2024-03-18 at 18:04 -0400, Tom Lane wrote:
> >> This is causing all CI jobs to fail the "compiler warnings" check.
>
> > I did run CI before checkin, and it passed:
> > https://cirrus-ci.com/build/5382423490330624
>
On Tue, Mar 12, 2024 at 10:03 AM Melanie Plageman
wrote:
> I've rebased the attached v10 over top of the changes to
> lazy_scan_heap() Heikki just committed and over the v6 streaming read
> patch set. I started testing them and see that you are right, we no
> longer pin too many buffers. However,
I am planning to push the bufmgr.c patch soon. At that point the new
API won't have any direct callers yet, but the traditional
ReadBuffer() family of functions will internally reach
StartReadBuffers(nblocks=1) followed by WaitReadBuffers(),
ZeroBuffer() or nothing as appropriate. Any more though
I canvassed Andres off-list since smgrzeroextend() is his invention,
and he wondered if it was a good idea to blur the distinction between
the different zero-extension strategies like that. Good question. My
take is that it's fine:
mdzeroextend() already uses fallocate() only for nblocks > 8, bu
On Sat, Mar 16, 2024 at 7:27 AM Tom Lane wrote:
> Are there limits on the runtime of CI or cfbot jobs? Maybe
> somebody should go check those systems.
Those get killed at a higher level after 60 minutes (configurable but
we didn't change it AFAIK):
https://cirrus-ci.org/faq/#instance-timed-out
For me it seems that the LLVMRunPasses() call, new in
commit 76200e5ee469e4a9db5f9514b9d0c6a31b496bff
Author: Thomas Munro
Date: Wed Oct 18 22:15:54 2023 +1300
jit: Changes for LLVM 17.
is reaching code that segfaults inside libLLVM, specifically in
llvm::InlineFunction(llvm::CallBase
em to be a
> combination we have covered in the buildfarm.
Yeah, 18.1 (note they switched to 1-based minor numbers, there was no
18.0) just came out a week or so ago. Despite testing their 18 branch
just before their "RC1" tag, as recently as
commit d282e88e50521a457fa1b36e55f43bac02a3
On Fri, Mar 15, 2024 at 11:19 AM Tom Lane wrote:
> Heikki Linnakangas writes:
> > Somehow the 'gin-leave-leaf-split-incomplete' injection point was active
> > in the 'intarray' test. That makes no sense. That injection point is
> > only used by the test in src/test/modules/gin/. Perhaps that ran
On Fri, Mar 15, 2024 at 3:18 AM Tomas Vondra
wrote:
> So, IIUC this means (1) the patched code is more aggressive wrt
> prefetching (because we prefetch more data overall, because master would
> prefetch N pages and patched prefetches N ranges, each of which may be
> multiple pages. And (2) it's n
On Fri, Mar 15, 2024 at 7:00 AM Alexander Lakhin wrote:
> Could it be that the timeout (360 sec?) is just not enough for the test
> under the current (changed due to switch to meson) conditions?
Hmm, well it looks like he switched over to meson around 42 days ago
2024-02-01, looking at "calliphor
On Thu, Mar 14, 2024 at 3:27 PM Michael Paquier wrote:
> Hmm. Perhaps 8af25652489? That looks like the closest thing in the
> list that could have played with the way WAL is generated, hence
> potentially impacting the records that are replayed.
Yeah, I was wondering if its checkpoint delaying
On Wed, Mar 13, 2024 at 10:53 AM Thomas Munro wrote:
> https://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=skink&dt=2024-02-23%2015%3A44%3A35
Assuming it is due to a commit in master, and given the failure
frequency, I think it is very likely to be a change from this 3 day
window of
Hi,
I noticed 3 regression test failures like $SUBJECT in cfbot runs for
unrelated patches that probably shouldn't affect GIN, so I guess this
is probably a problem in master. All three happened on FreeBSD, but I
doubt that's relevant, it's just that the FreeBSD CI task was randomly
selected to a
Short sales pitch for these patches:
* the default settings eat data on Macs and Windows
* nobody understands what wal_sync_method=fsync_writethrough means anyway
* it's a weird kludge that it affects not only WAL, let's clean that up
eduction to do so by merging, like this.
From 61b351b60d22060e5fc082645cdfc19188ac4841 Mon Sep 17 00:00:00 2001
From: Thomas Munro
Date: Sat, 9 Mar 2024 16:04:21 +1300
Subject: [PATCH v5 1/3] Use smgrwritev() for both overwriting and extending.
Since mdwrite() and mdextend() were basically the same and both nee
On Sun, Mar 3, 2024 at 11:41 AM Tomas Vondra
wrote:
> On 3/2/24 23:28, Melanie Plageman wrote:
> > On Sat, Mar 2, 2024 at 10:05 AM Tomas Vondra
> > wrote:
> >> With the current "master" code, eic=1 means we'll issue a prefetch for B
> >> and then read+process A. And then issue prefetch for C and
rnal details of the smgr implementation, it's
part of the "contract" for the API.
From 0a57274e29369e61712941e379c24f7db1dec068 Mon Sep 17 00:00:00 2001
From: Thomas Munro
Date: Sat, 9 Mar 2024 16:04:21 +1300
Subject: [PATCH v4 1/3] Merge smgrzeroextend() and smgrextend() with
smgr
On Wed, Mar 13, 2024 at 9:57 PM Heikki Linnakangas wrote:
> Let's bite the bullet and merge the smgrwrite and smgrextend functions
> at the smgr level too. I propose the following signature:
>
> #define SWF_SKIP_FSYNC 0x01
> #define SWF_EXTEND 0x02
> #define SWF_ZERO
On Wed, Mar 13, 2024 at 4:50 AM Maxim Orlov wrote:
> I looked at the changes and I liked them. Here are my thoughts:
Thanks for looking! Pushed.
Hi,
Several animals are timing out while waiting for catchup,
sporadically. I don't know why. The oldest example I have found so
far by clicking around is:
https://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=skink&dt=2024-02-23%2015%3A44%3A35
So perhaps something was committed ~3 weeks ago
One more observation while I'm thinking about bulk_write.c... hmm, it
writes the data out and asks the checkpointer to fsync it, but doesn't
call smgrwriteback(). I assume that means that on Linux the physical
writeback sometimes won't happen until the checkpointer eventually
calls fsync() sequent
On Tue, Mar 12, 2024 at 7:40 PM Thomas Munro wrote:
> possible. So in the current patch you say "hey please read these 16
> blocks" and it returns saying "only read 1", you call again with 15
Oops, typo worth correcting: s/15/16/. Point being that the caller is
inter
On Tue, Mar 12, 2024 at 7:15 PM Dilip Kumar wrote:
> I am planning to review this patch set, so started going through 0001,
> I have a question related to how we are issuing smgrprefetch in
> StartReadBuffers()
Thanks!
> + /*
> + * In theory we should only do this if PrepareReadBuffers() had to
On Tue, Mar 12, 2024 at 2:56 PM Andrew Dunstan wrote:
> On 2024-03-11 Mo 04:21, Oleg Tselebrovskiy wrote:
> > Greetings, everyone!
> >
> > While running "installchecks" on databases with UTF-8 encoding the test
> > citext_utf8 fails because of Turkish dotted I like this:
> >
> > SELECT 'i'::citex
On Mon, Mar 11, 2024 at 5:31 AM Melanie Plageman
wrote:
> On Wed, Mar 6, 2024 at 6:47 PM Melanie Plageman
> wrote:
> > Performance results:
> >
> > The TL;DR of my performance results is that streaming read vacuum is
> > faster. However there is an issue with the interaction of the streaming
> >
tings. We write out the index 128kB at a time, but
the WAL 8kB at a time.
From 793caf2db6c00314f5bd8f7146d4797508f2f627 Mon Sep 17 00:00:00 2001
From: Thomas Munro
Date: Sat, 9 Mar 2024 16:04:21 +1300
Subject: [PATCH v3 1/3] Provide vectored variant of smgrextend().
Since mdwrite() and mdextend(
On Mon, Mar 11, 2024 at 9:59 AM Thomas Munro wrote:
> On Mon, Mar 11, 2024 at 9:30 AM Heikki Linnakangas wrote:
> > Hmm, I'm not sure if we need even smgrreleaseall() here anymore. It's
> > not required for correctness AFAICS. We don't do it in
On Mon, Mar 11, 2024 at 9:30 AM Heikki Linnakangas wrote:
> Barring objections, I'll commit the attached.
+1
I guess the comment for smgrreleaseall() could also be updated. It
mentions only PROCSIGNAL_BARRIER_SMGRRELEASE, but I think sinval
overflow (InvalidateSystemCaches()) should also be men
On Sun, Mar 10, 2024 at 6:48 PM Thomas Munro wrote:
> I won't be surprised if the answer is: if you're holding a reference,
> you have to get a pin (referring to bulk_write.c).
Ahhh, on second thoughts, I take that back, I think the original
theory still actually works just fine.
On Sun, Mar 10, 2024 at 5:02 PM Thomas Munro wrote:
> Thanks, reproduced here (painfully slowly). Looking...
I changed that ERROR to a PANIC and now I can see that
_bt_metaversion() is failing to read a meta page (block 0), and the
file is indeed of size 0 in my filesystem. Which is not c
On Sat, Mar 9, 2024 at 9:48 AM Tomas Vondra
wrote:
> I spent a bit of time investigating this today, but haven't made much
> progress due to (a) my unfamiliarity with the smgr code in general and
> the patch in particular, and (b) CLOBBER_CACHE_ALWAYS making it quite
> time consuming to iterate an
Slightly better version, adjusting a few obsoleted comments, adjusting
error message to distinguish write/extend, fixing a thinko in
smgr_cached_nblocks maintenance.
From c786f979b0c38364775e32b9403b79303507d37b Mon Sep 17 00:00:00 2001
From: Thomas Munro
Date: Sat, 9 Mar 2024 16:04:21 +1300
Bmag4BF%2BzHo7qo%3Do9CFheB8%3Dg6uT5TUm2gkvA%40mail.gmail.com
From 4611cb121bbfa787ddbba4bc0e80ac6c732345d0 Mon Sep 17 00:00:00 2001
From: Thomas Munro
Date: Sat, 9 Mar 2024 16:04:21 +1300
Subject: [PATCH 1/2] Provide vectored variant of smgrextend().
Since mdwrite() and mdextend() were basical
On Sat, Mar 9, 2024 at 2:36 AM Tomas Vondra
wrote:
> On 3/8/24 13:21, Tomas Vondra wrote:
> > My guess would be 8af25652489, as it's the only storage-related commit.
> >
> > I'm currently running tests to verify this.
> >
>
> Yup, the breakage starts with this commit. I haven't looked into the
> r
Happened again. I see this is OpenSUSE. Does that mean the file
system is Btrfs?
+1
://www.postgresql.org/message-id/flat/CA%2BhUKG%2B2hZ0sBztPW4mkLfng0qfkNtAHFUfxOMLizJ0BPmi5%2Bg%40mail.gmail.com
From 85678257fef94aa3ca3efb39ce55fb66df7c889e Mon Sep 17 00:00:00 2001
From: Thomas Munro
Date: Fri, 26 May 2023 01:41:11 +1200
Subject: [PATCH v3] Allow relation segment size to be set by initdb.
MIME
ever really uses --with-segsize (do they?), but
if we make it an initdb option it will be more popular and this will
become a problem. Hmm.
From 1d183245e9676ef45ca6a93e7d442ee903a2a14c Mon Sep 17 00:00:00 2001
From: Thomas Munro
Date: Wed, 6 Mar 2024 17:44:19 +1300
Subject: [PATCH] Fix potential
On Wed, Mar 6, 2024 at 3:16 PM Thomas Munro wrote:
> Here's a rebase.
Now with a wait event and a paragraph of documentation.
From 9d5a60e9a9cc4a4312de3081be99c254a8876e42 Mon Sep 17 00:00:00 2001
From: Thomas Munro
Date: Sat, 2 Sep 2023 22:21:49 +1200
Subject: [PATCH v4] CREATE
On Wed, Oct 11, 2023 at 7:40 PM Peter Eisentraut wrote:
> On 07.10.23 07:51, Thomas Munro wrote:
> > Here is an experimental POC of fast/cheap database cloning.
>
> Here are some previous discussions of this:
>
> https://www.postgresql.org/message-id/flat/20131001223108.G
On Wed, Mar 6, 2024 at 2:43 AM Peter Eisentraut wrote:
> As far as I can tell, the original pg_upgrade patch has been ready to
> commit since October. Unless Thomas has any qualms that have not been
> made explicit in this thread, I suggest we move ahead with that.
pg_upgrade --copy-file-range p
On Sat, Dec 2, 2023 at 3:30 PM Thomas Munro wrote:
> On Sat, Dec 2, 2023 at 2:18 PM Thomas Munro wrote:
> > On Fri, Dec 1, 2023 at 6:13 PM Justin Pryzby wrote:
> > > $ kill -9 2524495; sleep 0.05; pg_ctl -D ./pgdev.dat1 stop -m fast #
> > > 2524495 is a child'
These two animals seem to have got mixed up about about the size of
this relation in the same place:
https://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=avocet&dt=2024-02-28%2017%3A34%3A30
https://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=trilobite&dt=2024-03-01%2006%3A47%3A53
+++
/hom
On Sat, Mar 2, 2024 at 3:12 AM Peter Eisentraut wrote:
> 0001-Return-ssize_t-in-fd.c-I-O-functions.patch
>
> This patch looks correct to me.
Thanks, I'll push this one.
> 0002-Fix-theoretical-overflow-in-Windows-pg_pread-pg_pwri.patch
>
> I have two comments on that:
>
> For the overflow of the
Rebased over 8d140c58.
v2-0001-Make-wal_sync_method-fdatasync-the-default-on-all.patch
Description: Binary data
v2-0002-Remove-fsync_writethrough-add-fsync-full-macOS-on.patch
Description: Binary data
On Wed, Feb 28, 2024 at 9:24 AM Heikki Linnakangas wrote:
> Here's a patch to fully remove AIX support.
--- a/doc/src/sgml/installation.sgml
+++ b/doc/src/sgml/installation.sgml
@@ -3401,7 +3401,7 @@ export MANPATH
PostgreSQL can be expected to work on current
versions of these operat
Patches attached.
PS Correction to my earlier statement about POSIX: the traditional K&R
interfaces were indeed in the original POSIX.1 1988 but it was the
1990 edition (approximately coinciding with standard C) that adopted
void, size_t, const and invented ssize_t.
0001-Return-ssize_t-in-fd.c-I
On Wed, Feb 7, 2024 at 11:54 PM Nazir Bilal Yavuz wrote:
> 0001-Provide-vectored-variant-of-ReadBuffer:
>
> - Do we need to pass the hit variable to ReadBuffer_common()? I think
> it can be just declared in the ReadBuffer_common() now.
Right, thanks! Done, in the version I'll post shortly.
> 00
[Sorry to those who received this message twice -- the first time got
bounced by the list because of a defunct email address in the CC
list.]
Here is a rebase of Palak's v2 patch. I didn't change anything except
for the required resource manager API change, a pgindent run, and
removal of a stray
On Sun, Feb 25, 2024 at 11:16 AM Thomas Munro wrote:
> On Sun, Feb 25, 2024 at 11:06 AM Heikki Linnakangas wrote:
> > Regarding the issue at hand, perhaps we should define PG_IO_ALIGN_SIZE as
> > 16 on AIX, if that's the best the linker can do on that platform.
>
> You
On Sun, Feb 25, 2024 at 11:06 AM Heikki Linnakangas wrote:
> Regarding the issue at hand, perhaps we should define PG_IO_ALIGN_SIZE as 16
> on AIX, if that's the best the linker can do on that platform.
You'll probably get either an error or silently fall back to buffered
I/O, if direct I/O is e
On Sun, Feb 25, 2024 at 9:12 AM Thomas Munro wrote:
> On Sun, Feb 25, 2024 at 8:50 AM Noah Misch wrote:
> > On GNU/Linux x64, gcc correctly records alignment=2**12 for the associated
> > section (.rodata for bulk_write.o zero_buffer, .bss for pg_prewarm.o
> > blockbuffer).
On Sun, Feb 25, 2024 at 8:50 AM Noah Misch wrote:
> On GNU/Linux x64, gcc correctly records alignment=2**12 for the associated
> section (.rodata for bulk_write.o zero_buffer, .bss for pg_prewarm.o
> blockbuffer). If I'm reading this right, neither AIX gcc nor xlc is marking
> the section with su
On Sun, Feb 25, 2024 at 6:24 AM Noah Misch wrote:
> On Fri, Feb 23, 2024 at 04:27:34PM +0200, Heikki Linnakangas wrote:
> > Committed this. Thanks everyone!
>
> https://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=mandrill&dt=2024-02-24%2015%3A13%3A14
> got:
> TRAP: failed Assert("(uintptr_t)
On Thu, Feb 22, 2024 at 5:37 PM Andrei Lepikhov
wrote:
> On 22/2/2024 06:42, Thomas Munro wrote:
> > extreme skew for one version of the problem, but even with zero/normal
> > skewness and perfect estimation of the number of partitions, if you
Sorry, I meant to write "but eve
On Wed, Feb 21, 2024 at 7:34 PM Andrei Lepikhov
wrote:
> I see in [1] that the reporter mentioned a delay between the error
> message in parallel HashJoin and the return control back from PSQL. Your
> patch might reduce this delay.
> Also, I have the same complaint from users who processed gigabyt
On Thu, Feb 22, 2024 at 10:30 AM Thomas Munro wrote:
> collisions arbitrarily far apart (just decide how many bits to use).
. o O ( Perhaps if you also allocated slots using a FIFO freelist,
instead of the current linear search for the first free slot, you
could maximise the time before a s
On Thu, Feb 22, 2024 at 8:19 AM Heikki Linnakangas wrote:
> - Separate dsm_handle, used by backend code to interact with the high
> level interface in dsm.c, from dsm_impl_handle, which is used to
> interact with the low-level functions in dsm_impl.c. This gets rid of
> the convention in dsm.c of
On Sat, Jul 2, 2022 at 11:10 AM Thomas Munro wrote:
> On Sat, Jul 2, 2022 at 1:15 AM Robert Haas wrote:
> > Changing the default on certain platforms to 'posix' or 'sysv'
> > according to what works best on that platform seems reasonable to me.
>
>
On Fri, Feb 16, 2024 at 5:29 PM Robert Haas wrote:
> 3. Reserve lots of address space and then only use some of it. I hear
> rumors that some forks of PG have implemented something like this. The
> idea is that you convince the OS to give you a whole bunch of address
> space, but you try to avoid
On Thu, Oct 5, 2023 at 9:07 PM David Rowley wrote:
> Thanks. Pushed.
FYI somehow this plan from a8a968a8212e flipped in this run:
=== dumping
/home/bf/bf-build/mylodon/HEAD/pgsql.build/testrun/recovery/027_stream_regress/data/regression.diffs
===
diff -U3
/home/bf/bf-build/mylodon/HEAD/pgsql/s
On Tue, Feb 13, 2024 at 9:25 AM Jeff Davis wrote:
> On Sun, 2024-02-11 at 22:04 +0530, Robert Haas wrote:
> > "icu_multilib must be loaded via shared_preload_libraries.
> > icu_multilib ignores any ICU library with a major version greater
> > than
> > that with which PostgreSQL was built."
> >
> >
On Wed, Feb 14, 2024 at 3:23 AM Heikki Linnakangas wrote:
> On 29/01/2024 14:06, Heikki Linnakangas wrote:
> > If you call dsa_allocate_extended(DSA_ALLOC_NO_OOM), it will still
> > ereport an error if you run out of space (originally reported at [0]).
> >
> > Attached patch adds code to test_dsa.
On Sat, Feb 10, 2024 at 8:36 AM Andres Freund wrote:
> Also, yikes, that's an ugly way of doing hardware detection. Jumping out of a
> signal handler into normal code. Brrr.
Maybe it's a little baroque but what's actually wrong with it?
OpenSSL does something similar during initialisation as a fa
On Tue, Jan 16, 2024 at 8:52 AM Robert Haas wrote:
> On Wed, Dec 6, 2023 at 8:45 PM Kyotaro Horiguchi
> wrote:
> > > So I think we should just hard-code the error messages in English and
> > > move on. However, English is my language so perhaps I should abstain
> > > and leave it to others to de
On Thu, Feb 8, 2024 at 3:38 PM Thomas Munro wrote:
> Perhaps you could wrap it in a branch-free sign() function so you get
> a narrow answer?
>
> https://stackoverflow.com/questions/14579920/fast-sign-of-integer-in-c
Ah, strike that, it is much like the suggested (a > b) - (a <
On Thu, Feb 8, 2024 at 3:06 PM Andres Freund wrote:
> On 2024-02-07 19:52:11 -0600, Nathan Bossart wrote:
> > On Wed, Feb 07, 2024 at 04:42:07PM -0800, Andres Freund wrote:
> > > On 2024-02-07 16:21:24 -0600, Nathan Bossart wrote:
> > >> The assembly for that looks encouraging, but I still need to
On Tue, Jan 30, 2024 at 5:06 PM Tom Lane wrote:
> Thomas Munro writes:
> > On Sat, Jan 13, 2024 at 1:51 PM Tom Lane wrote:
> >> Time for a bug report to IO::Tty's authors, I guess.
>
> > Ahh, there is one: https://github.com/cpan-authors/IO-Tty/issues/38
>
&
On Fri, Feb 2, 2024 at 12:56 PM Yugo NAGATA wrote:
> On Fri, 2 Feb 2024 11:18:18 +0100
> Thomas Munro wrote:
> > One simple way to address that would be to make XLogFileInitInternal()
> > wait for InstallXLogFileSegment() to finish. It's a little
>
> Or, can we ma
Hi,
New WAL space is created by renaming a file into place. Either a
newly created file with a temporary name or, ideally, a recyclable old
file with a name derived from an old LSN. I think there is a data
loss window between rename() and fsync(parent_directory). A
concurrent backend might open
On Wed, Nov 29, 2023 at 1:42 PM Heikki Linnakangas wrote:
> I spent some more time digging into this, experimenting with different
> approaches. Came up with pretty significant changes; see below:
Hi Heikki,
I think this approach is good. As I wrote in the first email, I had
briefly considered
On Thu, Jan 25, 2024 at 4:44 PM Thomas Munro wrote:
> ... A few build farm animals will
> now fail in the configure step as discussed, and need some adjustment
> (ie disable LLVM or upgrade to LLVM 10+ for the master branch).
Owners pinged.
On Wed, Jan 24, 2024 at 1:39 PM Andres Freund wrote:
> On 2024-01-23 17:26:19 -0600, Tristan Partin wrote:
> > On Tue Jan 23, 2024 at 4:23 PM CST, Andres Freund wrote:
> > > A fork() while threads are running is undefined behavior IIRC, and
> > > undefined
> > > behavior isn't limited to a single
Thanks all for the discussion. Pushed. A few build farm animals will
now fail in the configure step as discussed, and need some adjustment
(ie disable LLVM or upgrade to LLVM 10+ for the master branch).
Next year I think we should be able to do a much bigger cleanup, by
moving to LLVM 14+.
On Wed, Jan 3, 2024 at 6:04 PM Thomas Munro wrote:
> LLVM 16 provided a new function name[1], and LLVM 18 (not shipped yet)
> has started complaining[2] about the old spelling.
>
> Here's a patch.
And pushed.
Just in case anyone else is confused by this, be aware that they
On Thu, Jan 25, 2024 at 8:51 AM Kirk Wolak wrote:
> getrusage(RUSAGE_SELF, &usage);
> memory_usage_bytes = usage.ru_maxrss * 1024;
FWIW log_statement_stats = on shows that in the logs. See ShowUsage()
in postgres.c.
On Fri, Jan 19, 2024 at 1:47 AM Anton Voloshin
wrote:
> I believe there is a small problem in the 039_end_of_wal.pl's
> "xl_tot_len zero" test. It supposes that after immediate shutdown the
> server, upon startup recovery, should always produce a message matching
> "invalid record length at .*: wa
On Sat, Jan 13, 2024 at 1:51 PM Tom Lane wrote:
> Time for a bug report to IO::Tty's authors, I guess.
Ahh, there is one: https://github.com/cpan-authors/IO-Tty/issues/38
In the meantime, will look into whether I can pin that package to 1.17
somewhere in the pipeline, hopefully later today...
On Sat, Jan 13, 2024 at 9:32 AM Tom Lane wrote:
> It looks like every recent cfbot run has failed in the
> FreeBSD-13-Meson build, even if it worked in other ones.
> The symptoms are failures in the TAP tests that try to
> use interactive_psql:
>
> Can't call method "slave" on an undefined value a
On Fri, Jan 12, 2024 at 3:31 AM Heikki Linnakangas wrote:
> Ok. It feels surprising to have three steps. I understand that you need
> two steps, one to start the I/O and another to wait for them to finish,
> but why do you need separate Prepare and Start steps? What can you do in
> between them? (
On Thu, Jan 11, 2024 at 8:58 AM Heikki Linnakangas wrote:
> On 10/01/2024 06:13, Thomas Munro wrote:
> > Bikeshedding call: I am open to better suggestions for the names
> > PrepareReadBuffer() and CompleteReadBuffers(), they seem a little
> > grammatically clumsy.
>
&g
ommit/5ac12951b4e9bbfcc5791282d0961ec2b65575e9
From bc2a07e0012aa58af2cf97a202d181f473a4d7bc Mon Sep 17 00:00:00 2001
From: Thomas Munro
Date: Wed, 3 Jan 2024 17:45:30 +1300
Subject: [PATCH] Track LLVM 18 changes.
https://github.com/llvm/llvm-project/commit/1b97645e56bf321b06d1353024339958b64fd242
https://github.com
Hi Ranier,
I doubt it really matters, unless perhaps it's upsetting some static
analysis tool? It's bounded by FD_SETSIZE, a small number.
FWIW, I would like to delete pgwin32_select() in PG17. Before PG16
(commit 7389aad6), the postmaster used it to receive incoming
connections, but that was r
On Thu, Dec 28, 2023 at 4:02 AM Justin Pryzby wrote:
> My main question is why an IO error would cause the DB to abort, rather
> than raising an ERROR.
In CommitTransaction() there is a stretch of code beginning s->state =
TRANS_COMMIT and ending s->state = TRANS_DEFAULT, from which we call
out t
On Mon, Dec 25, 2023 at 7:09 AM Tom Lane wrote:
> Coverity whinged this morning about the following bit in
> the new pg_combinebackup code:
>
> 644 unsignedrb;
> 645
> 646 /* Read the block from the correct source, except if
> dry-run. */
> 647
On Sat, Dec 23, 2023 at 9:40 AM Peter Eisentraut wrote:
> On 13.11.23 08:15, Peter Eisentraut wrote:
> > On 08.10.23 07:15, Thomas Munro wrote:
> >>> About your patch:
> >>>
> >>> I think you should have a "check" function called from
> &
On Fri, Dec 15, 2023 at 9:53 AM Thomas Munro wrote:
> ... We've seen a system with ~30GB of files in there
> (note: full/untruncated be would be 2³² xids × sizeof(uint64_t) =
> 32GB). It's not just a gradual disk space leak: according to disk
> space monitoring, this syste
301 - 400 of 2304 matches
Mail list logo