I'm comfortable
reviewing or committing right now. So maybe move those to the next
commitfest or close the entry as returned with feedback?
--
Peter Eisentraut http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services
ntifier has confused me a bit. (What's a "set
>> identifier"?) Maybe use CopyAndSetIdentifier? (We similarly have
>> MemoryContextResetAndDeleteChildren.)
>
> No objection, do you want to make the change?
and done
--
Peter Eisentraut http://w
m going to look at this again today, and will post something within
12 hours. Please hold off on committing until then.
--
Peter Geoghegan
about applying following patch on the top of v14?
It's clearly necessary. Looks fine to me.
--
Peter Geoghegan
t were not within
scope for v11. In some cases, these were addressed with only modest
effort. It's pretty obvious to me that some aspects of MERGE's
architecture are accidents. Not necessarily happy accidents.
--
Peter Geoghegan
On Fri, Apr 6, 2018 at 4:39 PM, Robert Haas wrote:
> Committed. Thanks to David for the report and analysis and to Peter
> for the patch and study.
Thanks!
--
Peter Geoghegan
eature freeze, to
make sure that it is in good shape. I have a general interest in
making sure that amcheck gains acceptance as a way of validating a
complicated patch like this one after commit.
[1] https://www.postgresql.org/message-id/15195.1490988897%40sss.pgh.pa.us
--
Peter Geoghegan
ssive usage of
> ItemPointerGetBlockNumberNoCheck(&(itup->t_tid)), suggest to wrap it to
> macro something like this:
> #define BTreeInnerTupleGetDownLink(itup) \
> ItemPointerGetBlockNumberNoCheck(&(itup->t_tid))
Agreed. We do that with GIN.
--
Peter Geoghegan
On Sat, Apr 7, 2018 at 1:02 PM, Teodor Sigaev wrote:
> Thanks to everyone, pushed.
I'll keep an eye on the buildfarm, since it's late in Russia.
--
Peter Geoghegan
robably be returning NULL for most
> properties except 'returnable'.
>
> I can look at fixing these for you if you like?
I'm happy to accept your help with it, for one.
--
Peter Geoghegan
already occurs because relids_logged would be empty? Worth a comment in
> that case)
Added an assertion and a comment.
Committed with those changes.
--
Peter Eisentraut http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services
This part of the patch didn't end up being necessary, since the updated
implementation of logical decoding of TRUNCATE could do without it.
--
Peter Eisentraut http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services
least add a note to the nbtree README that
explains the high level theory behind the optimization, as part of
post-commit clean-up. I'll ask him to say something about how it might
affect extent-based storage, too.
--
Peter Geoghegan
On Sun, Apr 8, 2018 at 11:18 AM, Teodor Sigaev wrote:
> Thank you, fixed
I suggest that we remove some unneeded amcheck tests, as in the
attached patch. They don't seem to add anything.
--
Peter Geoghegan
From 0dbbee5bfff8816cddf86961bf4959192f62f1ff Mon Sep 17 00:00:00 2001
Fro
items. I would strongly
> suggest fixing
> this prior to the 11 release as it is unintuitive from a user standpoint
> to use ‘TRUE’
I think this is actually more accurately classified as an existing bug
in PostgreSQL 10.
--
Peter Eisentraut http://www.2ndQuadrant.com/
Postg
re serious practical problems. I have personally scoffed at stories
involving Postgres databases corruption that gets attributed to
running out of disk space. Looks like I was dead wrong.
[1] https://danluu.com/file-consistency/ -- "Filesystem correctness"
--
Peter Geoghegan
backpatch it, isn't it?
+1 to committing to master as a documented behavioral change.
--
Peter Geoghegan
libpq.so
It's not clear to me why you would need these, given that Debian has
been doing this for many years without this.
--
Peter Eisentraut http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services
ing about (perhaps
just renaming the macro).
I do not have the time to write a patch right away, but I should be
able to post one in a few days. I want to avoid sending several small
patches.
--
Peter Geoghegan
s and it seems to work as
> expected. Unforunately, there are no tests for libpq, afaik. But testing
> such features would become complicated quite quickly, with the need to
> generate certificates and such...
There are tests in src/test/ssl/ that would probably be a good fit to
extend for this.
ation.
--
Peter Eisentraut http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services
From eaccc63b2f9943633f957f80d67acf54ad70098d Mon Sep 17 00:00:00 2001
From: Peter Eisentraut
Date: Tue, 10 Apr 2018 12:32:05 -0400
Subject: [PATCH] pg_
_bt_mark_page_halfdead() looked like it had a problem, but it now
looks like I was wrong. I also verified every other
BTreeInnerTupleGetDownLink() caller. It now looks like everything is
good here.
--
Peter Geoghegan
ing down the server using immediate mode
could lead to torn pages, that crash recovery will need to repair at a
later stage. I think that some strong caveats around this are required
in the pg_verify_checksums docs, at a minimum.
--
Peter Geoghegan
On Tue, Apr 10, 2018 at 1:37 PM, Peter Geoghegan wrote:
> _bt_mark_page_halfdead() looked like it had a problem, but it now
> looks like I was wrong.
I did find another problem, though. Looks like the idea to explicitly
represent the number of attributes directly has paid off already:
pg@
On Tue, Apr 10, 2018 at 4:44 PM, Michael Paquier wrote:
> Peter, the code does the right thing as it requires the instance's
> control file state to be either DB_SHUTDOWNED_IN_RECOVERY or
> DB_SHUTDOWNED. The documentation, on the contrary, implies that
> the instance just ne
;>
>>
>> Might be a left-over a previous version which might have used MOD(a, b), but
>> a/b are rather used for fp numbers and i/j for integers.
>
> Ah that seems possible.
committed
--
Peter Eisentraut http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services
ilege on the replication target table, but it is not
> described in the manual.
Committed, thanks.
--
Peter Eisentraut http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services
act similarly.
>
> So apparently I missed something with that. Will look more closely.
This was listed as an open item, but it was already fixed by
877cdf11eaa9cabcb9b1e3c1bef0760fe08efdc3, so I'll remove it.
--
Peter Eisentraut http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services
the resolved items to a separate
section on the page, instead of just removing them? I'm not sure that
the resolved sections are useful, compared to just using the git log.
--
Peter Eisentraut http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services
not NaNs
>
> Attached is a patch for the first issue. I'll look at fixing the rest
> later this week.
Committed #1. Please keep more patches coming. :)
--
Peter Eisentraut http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services
without making hash of the
wiki page.
It's not a very large deal, but I doubt the result is very useful. The
current "resolved before 11beta1" section is pretty much useless.
--
Peter Eisentraut http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services
is theoretically allowed to affect the behavior of equality,
even though so far we've never tried to make that work for any
collatable datatype.
Maybe the right thing to do is to say that any collation will work
equally well (as will any opclass). Maybe that's the same thing as
what you said, though.
> +1 for ii_IndexAttrNumbers.
+1
--
Peter Geoghegan
On 4/11/18 12:06, Andres Freund wrote:
> On 2018-03-22 15:19:12 +0100, Pavel Stehule wrote:
>> attached patch should to fix it
>
> This is still broken, and has been an open item for a bit. Peter, Could
> you check whether Pavel's fix resolves the issue for you?
Y
On 4/11/18 17:08, Peter Geoghegan wrote:
>> However, I don't see any point in defining collations here, because
>> INCLUDE attributes exist solely for index-only scans. So, index just
>> can return value of INCLUDE attribute "as is", no point to do something
>&
t that at child creation.
I agree it should do one or the other, but not what it's doing now.
--
Peter Eisentraut http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services
On Wed, Apr 11, 2018 at 2:29 PM, Peter Eisentraut
wrote:
> But in this case it doesn't even do equality comparison, it just returns
> the value.
That's the idea that I tried to express. The point is that we need to
tell the user that there is no need to worry about it, rather t
On 4/11/18 17:38, Peter Geoghegan wrote:
> On Wed, Apr 11, 2018 at 2:29 PM, Peter Eisentraut
> wrote:
>> But in this case it doesn't even do equality comparison, it just returns
>> the value.
>
> That's the idea that I tried to express. The point is that we nee
s my attempt to clean this up a bit. Thoughts?
--
Peter Eisentraut http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services
From db7f5eacc23114f24230dd087837d94bcefc4c88 Mon Sep 17 00:00:00 2001
From: Peter Eisentraut
Date: Wed, 11 Apr
On Tue, Apr 10, 2018 at 5:45 PM, Peter Geoghegan wrote:
> I did find another problem, though. Looks like the idea to explicitly
> represent the number of attributes directly has paid off already:
>
> pg@~[3711]=# create table covering_bug (f1 int, f2 int, f3 text);
> create uniqu
le also referring to more advanced
techniques for compressing (not truncating) leaf tuples as prefix
compression.
I suggested suffix truncation because it seemed to be the dominant way
of referring to the technique. And, because it seemed more logical:
the suffix is what gets truncated away.
--
Peter Geoghegan
b
= 15, c = 120 WHERE b = 4;
In short, it looks like the tests added to update.sql by commit
2f178441 ("Allow UPDATE to move rows between partitions") lead to this
failure, since I always hit a problem when update.sql is reached. I
haven't gone to the trouble of digging any deeper than that just yet.
--
Peter Geoghegan
ecially on boolean-like data that is brought into
fulltext search scope (so for example, if some attribute "balcony" is
false/not checked, you could index that as "no balcony" which then makes
both the presence AND the absence of it searchable...)
That's basically it. Thoughts?
-Peter
On Thu, Apr 12, 2018 at 11:47 AM, Peter Geoghegan wrote:
> In short, it looks like the tests added to update.sql by commit
> 2f178441 ("Allow UPDATE to move rows between partitions") lead to this
> failure, since I always hit a problem when update.sql is reached. I
> haven
t most once
per page. Is it worth noting the exception?
--
Peter Geoghegan
On 3/22/18 10:19, Pavel Stehule wrote:
> attached patch should to fix it
Committed.
I had to make a small tweak to make INOUT + DEFAULT parameters work in
PL/pgSQL.
--
Peter Eisentraut http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Train
;t replace the equivalent subexpression
> of MaxIndexTuplesPerPage with MinIndexTupleSize?".
I had the same reaction.
--
Peter Geoghegan
since fuzzing is inherently a computationally intensive
process; CPU costs are bound to dominate. A strategy like this would
also make it easier to run amcheck on indexes left behind by the
regression tests.
How can we do better?
--
Peter Geoghegan
tch added tests to. At no point do they leave behind any
INCLUDE indexes. I'll do something about that as part of the INCLUDE
patch that I'm working on at the moment.
--
Peter Geoghegan
On Sat, Apr 14, 2018 at 8:02 PM, Tom Lane wrote:
> So we've got *some*, but it sure looks like they were all added by the
> patch to fix covering indexes for partitions. I'd want to see some for
> plain-table cases as well.
Will do.
--
Peter Geoghegan
uot;can't happen" elog error). Finally, I made sure that we
don't drop all tables in the regression tests, so that we have some
pg_dump coverage for INCLUDE indexes, per a request from Tom.
[1] https://queue.acm.org/detail.cfm?id=2220317
--
Peter Geoghegan
From db777cad48f1
o had good success using syslog. While syslog is not very
structured, the setting syslog_split_messages allows sending log entries
that include newlines in one piece, which works well if you have some
kind of full-text search engine at the receiving end.
--
Peter Eisentraut http://www.2nd
full-text search engine at the receiving end.
> syslog suffers from the possibility to lose messages if I recall
> correctly, right? This may matter for some critical environments.
Depends on whether you configure it to use TCP or UDP.
--
Peter Eisentraut http://www.2ndQuadra
t your database.
I'm fine if Teodor wants to commit that change separately, of course.
--
Peter Geoghegan
ot to change.
Andres mentioned that he has prototyped an approach to buffer
management that uses a Radix tree, which is generally assumed to be
the right long-term fix.
--
Peter Geoghegan
On 4/17/18 16:14, Andres Freund wrote:
> I still think cmake is the least unreasonable path going forward.
I would rather try to make Meson work and if needed add features back
into Meson.
--
Peter Eisentraut http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Rem
ou posted.
> Attached patch is rebased to current head and contains some comment
> improvement in index_truncate_tuple() - you save some amount of memory with
> TupleDescCopy() call but didn't explain why pfree is enough to free all
> allocated memory.
Makes sense.
--
Peter Geoghegan
ached patch contains all
> changes suggested in my previous email.
Looks new BTreeTupSetNAtts () assertion good to me.
I suggest committing this patch as-is.
Thank you
--
Peter Geoghegan
On Wed, Apr 18, 2018 at 1:45 PM, Peter Geoghegan wrote:
> I suggest committing this patch as-is.
Actually, I see one tiny issue with extra '*' characters here:
> +* The number of attributes won't be explicitly represented if the
> +* negative infi
seem like a good idea? This could get pretty expensive if it
was overused, even by the standards of what we expect from
assertion-enabled builds, but we could make it optional if the
overhead got out of hand.
--
Peter Geoghegan
_attr(). AssertNotInCriticalSection() certainly found
several bugs when it was first added.
--
Peter Geoghegan
s first added.
>
> Yep.
I wrote a simple prototype of this. No assertion failed during a "make
installcheck". Assertions were added to heap_tuple_fetch_attr(), and a
couple of other places.
--
Peter Geoghegan
inity
items to the same comment block. While it's true that _bt_insertonpg()
cannot truncate downlinks to make new minus infinity items, I see that
as a narrower issue.
--
Peter Geoghegan
From dced00be29775965a45c7889ca99e19b96d9e4d0 Mon Sep 17 00:00:00 2001
From: Peter Geoghegan
Date: Wed
On Wed, Apr 18, 2018 at 10:29 PM, Teodor Sigaev wrote:
> Will see...
I'll take a look tomorrow.
--
Peter Geoghegan
UDE patch didn't actually
change the page deletion algorithm in any way.
> Trivial and naive fix is attached, but for me it looks a bit annoing that we
> store pointer (leafhikey) somewhere inside unlocked page.
Well, it has worked that way for a long time. No reason to change it now.
I also think that we could have better conventional regression test
coverage here.
--
Peter Geoghegan
g will break unless we do a detailed audit of all that code.
+1. I think that it is plainly a bad idea to do something like that at
this point in the cycle.
--
Peter Geoghegan
In summary, I think that we should find a way to never use
ItemPointerGetBlockNumberNoCheck() in nbtree, while still using it in
amcheck.
--
Peter Geoghegan
been
there already). Maybe we can add it to the amcheck regression tests
instead, since those are run less often. This also allows us to add a
test case specifically as part of the amcheck enhancement, which makes
a bit more sense.
--
Peter Geoghegan
ing soon.
> Hm, it seems to me, that 350ms is short enough to place it in both core and
> amcheck test. I think, basic functionality should be covered by core tests
> as we test insert/create.
I think you're right. There is really no excuse for not having
thorough code coverage for a module like nbtree.
--
Peter Geoghegan
ut what general area this deals with.
Perhaps something like the following:
"Get/set leaf page highkey's link. During the second phase of
deletion, the target leaf page's high key may point to an ancestor
page (at all other times, the leaf level high key's link is not used).
See th
em around the
absence of a needed downlink. There is an absent downlink, but that's
beside the point, and in any case missing a downlink is not
*inherently* wrong (as I said, it's not inherently wrong because of
the issue of interrupted multi-level page deletes).
--
Peter Geoghegan
On 2/27/17 10:55, Peter Eisentraut wrote:
> It appears we need pandoc 1.13 to get the good output. This won't be
> available until Debian stretch.
I understand that borka is updated to stretch now. So we could give
this another try. Updated patch attached.
--
Peter
ve good
> instructions for how to make a local install of it that doesn't affect
> the rest of the system, then that should not matter. And we need such
> instructions anyway, since it won't be on every distribution.
Did we decide on this?
--
Peter Eisentraut
think that that needs to be updated. The docs should
explain that 'I' relations are not backed by storage, since that will
probably affect users of one or two external tools.
--
Peter Geoghegan
On Sat, Apr 21, 2018 at 6:02 PM, Peter Geoghegan wrote:
> I refined the amcheck enhancement quite a bit today. It will not just
> check that a downlink is not missing; It will also confirm that it
> wasn't a legitimately interrupted multi-level deletion, by descending
> to the l
#x27;s presumably fixed now, so if we see the new error message, we'll
know it's a different issue.
--
Peter Eisentraut http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services
jsonb_plpython does the same
> thing) let you create spec-violating jsonb values that don't round-trip
> via jsonb_out/in.
Yeah this is not good. Is there a way to do this in a centralized way?
Is there a function to check an internal jsonb value for consistency.
Should at least the j
nnected to this patch, but allocation of
> BtreeCheckState is not needed, it could be allocated on stack.
>
> 4) Nevertheless, I suggest to use palloc0 (or memset(0)) for
> BtreeCheckState. Now several fields of that structure could be not inited.
I like the idea of using palloc0(). Do
On 4/23/18 00:33, Amit Kapila wrote:
> Yeah, right, I have missed the point that they can be vacuumed
> separately, however, I think that decision is somewhat questionable.
Manually vacuuming the TOAST table was a way to repair the recently
fixed TOAST bug, so it's kind of useful.
On 4/11/18 11:33, Tom Lane wrote:
> (Wanders away wondering what Peter has done about toasted parameter
> values for procedures in general ...)
I'm not sure. How can a procedure have a toasted parameter value?
--
Peter Eisentraut http://www.2ndQuadrant.com/
PostgreSQL
oint?
I'm not sure what you have in mind. In many cases, you can just change
the LANGUAGE clause. I suppose you could run 2to3 over the function
body? Write a PL/Python function to run 2to3?
--
Peter Eisentraut http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services
regress. :-)
It looks like you pushed v1, which didn't have the tests and other
changes you asked for. Attached patch adds those back.
Thanks
--
Peter Geoghegan
From fc2d158d3d1c52e94a6d3f66c65027bf09b64f7f Mon Sep 17 00:00:00 2001
From: Peter Geoghegan
Date: Thu, 19 Apr 2018 17:45:26 -0700
On Wed, Apr 25, 2018 at 12:06 PM, Teodor Sigaev wrote:
> Oops, I missed, I don't know how... Thank you very much for your quick eye!
Thanks Teodor.
--
Peter Geoghegan
On Wed, Apr 25, 2018 at 1:45 PM, Tom Lane wrote:
> Teodor, are you caught up to the point where it'd be okay to run pgindent,
> or are there still patches waiting?
I can't speak for Teodor, but fwiw I am not aware of any more patches waiting.
--
Peter Geoghegan
ajor difficulty here is choosing an
on-disk representation that lets us do everything that we'll want to
do in the future. That requires strong cooperation.
--
Peter Geoghegan
ou've
heard about, too. Andrey is probably correct in his suspicion that
we'll end up prototyping a number of approaches.
I'm glad that you're thinking about this, in any case. Seems like a
promising area to work on.
--
Peter Geoghegan
uch of a problem is it that we waste memory bandwidth copying to
and from OS cache, particularly on large systems? Might that be the
bigger problem, but also one that can be addressed incrementally?
I think that I may be repeating some of what Andrey said, in another
way (not sure of that).
--
Peter Geoghegan
gt; addresses this.
I'm confused what Fedora and Arch Linux have to do with this. Are they
distributing Windows binaries of third-party software?
--
Peter Eisentraut http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services
robably be made into a local variable now.
The rest looks good to me.
Do we need set_pglocale_pgservice() calls here if we're not doing NLS?
Does the logging stuff require it? I'm not sure.
--
Peter Eisentraut http://www.2ndQuadrant.com/
PostgreSQL Development, 24
l.org/24/2119/>.
What about the two patches? Which one is better?
Have you also considered using memmove() to deal with the overlap issue?
Benchmarks have been posted in this thread. Where is the benchmarking
tool? Should we include that in the source somehow?
--
Peter Eisentraut
I agree with this patch and the reasons for it.
A related point, perhaps we should change the key printed into
postmaster.pid to be in hexadecimal format ("0x08x") so that it matches
what ipcs prints.
--
Peter Eisentraut http://www.2ndQuadrant.com/
PostgreSQL Develop
On 2019-08-30 02:10, Michael Paquier wrote:
> On Thu, Aug 29, 2019 at 10:52:55AM +0200, Julien Rouhaud wrote:
>> That was already suggested by Thomas and seconded by Peter E., see
>> https://www.postgresql.org/message-id/2b1504ac-3d6c-11ec-e1ce-3daf132b3d37%402ndquadrant.com.
>
rvice() calls here if we're not doing NLS?
>> Does the logging stuff require it? I'm not sure.
>
> The logging part does not require it, but this can be used for
> PGSYSCONFDIR, no?
How does PGSYSCONFDIR come into play here?
--
Peter Eisentraut http://www
.patch
Are those the same ones?
--
Peter Eisentraut http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services
On 2019-09-05 04:12, Michael Paquier wrote:
> On Wed, Sep 04, 2019 at 04:38:21PM -0400, Alvaro Herrera wrote:
>> Marked RfC. Can we get on with this?
>
> FWIW, I have been able to test this one on Windows with MSVC and
> things are handled correctly.
committed
--
make a case for it.
--
Peter Eisentraut http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services
On 2019-09-05 01:59, Michael Paquier wrote:
> On Wed, Sep 04, 2019 at 02:37:12PM +0200, Peter Eisentraut wrote:
>>>> Do we need set_pglocale_pgservice() calls here if we're not doing NLS?
>>>> Does the logging stuff require it? I'm not sure.
>>>
>
he shmem key,
> so maybe there's no compatibility issue.)
Let's just leave it decimal then. At least then it's easier to compare
it to ls -i output.
--
Peter Eisentraut http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services
a.c, initdb.c, guc.c, enc_cipher.c). Suggest it would be
better to declare those as string constants in one place only.
---
Kind Regards,
Peter Smith
Fujitsu Australia
On 2019-09-05 22:27, Jonathan S. Katz wrote:
> PostgreSQL 12 Beta 4 will be released on 2019-09-12. Please make sure
> that fixes for bugs and other open items[1] are committed by the end of
> the weekend.
Could we get the list of major items in the release notes done by then?
On 2019-08-14 20:17, Andres Freund wrote:
> On 2019-05-22 16:40:28 +0200, Peter Eisentraut wrote:
>> On 2019-04-29 19:52, Andres Freund wrote:
>>> Hm. It appears that gettext supports expanding PRId64 PRIu64 etc in
>>> translated strings.
>>
>> That won&#
1501 - 1600 of 10756 matches
Mail list logo
