e --
it couldn't possibly be worth much of any risk. I can see the appeal
of consistency, but I also see the appeal of sticking to how things
work there: continually and explicitly inserting into and compacting
the heap seems like a good enough way of framing what a top-n heap
does,
ng, LGTM.
Thanks. I suggest spending at least as much time on unsympathetic
cases (e.g., only 2 or 3 tapes must be merged). At the same time, I
suggest focusing on a type that has relatively expensive comparisons,
such as collated text, to make differences clearer.
--
Peter Geoghegan
--
Sen
On Tue, Sep 6, 2016 at 2:46 PM, Peter Geoghegan wrote:
> Feel free to make a counter-proposal for a cap. I'm not attached to
> 500. I'm mostly worried about blatant waste with very large workMem
> sizings. Tens of thousands of tapes is just crazy. The amount of data
> th
thing that I've put so much work into. Robert has been
involved with 100% of all sorting patches I've written, generally with
far less input from anyone else, and at this point, that's really
rather a lot of complex patches.
> Let's begin with patch 1:
>
> On 08/02/2016 01
On Tue, Sep 6, 2016 at 12:51 PM, Tom Lane wrote:
> I rewrote the comment and pushed it.
Thank you.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
On Tue, Sep 6, 2016 at 12:39 PM, Peter Geoghegan wrote:
> On Tue, Sep 6, 2016 at 12:08 AM, Heikki Linnakangas wrote:
>>> I attach a patch that changes how we maintain the heap invariant
>>> during tuplesort merging.
>
>> Nice!
>
> Thanks!
BTW, the way that
-way merge heap
gets one element smaller). I can write a patch to do this renaming, if
you're interested. Someone should fix it, because independent of all
this, it's just wrong.
[1]
https://www.postgresql.org/message-id/CAM3SWZQKM=Pzc=cahzrixkjp2eo5q0jg1sofqqexfq647ji...@mail.gmail.com
On Tue, Sep 6, 2016 at 12:08 PM, Peter Geoghegan wrote:
> Offhand, I would think that taken together this is very important. I'd
> certainly want to see cases in the hundreds of megabytes or gigabytes
> of work_mem alongside your 4MB case, even just to be able to talk
> informal
r this is very important. I'd
certainly want to see cases in the hundreds of megabytes or gigabytes
of work_mem alongside your 4MB case, even just to be able to talk
informally about this. As you know, the default work_mem value is very
conservative.
--
Peter Geoghegan
--
Sent via pgsql-hackers
ed. I also knew any choice of
constant would be criticized (e.g., "that's voodoo"), so pointed out
specifically that it was non-critical.
What threshold would you use?
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
6, but that seems
unnecessary. I can reproduce it with my parallel CREATE INDEX patch
applied, with just the right test case and right number of workers
(it's rather delicate). After careful consideration, I can think of no
reason why 9.6 would be unaffected.
--
Peter Geoghegan
From 850af5f9f7b1b
, priority, and possibly other considerations. I see the
9.6 work on external sort as a building piece for that, as it removed
the one thing that was sensitive to work_mem in a surprising,
unpredictable way.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
On Fri, Sep 2, 2016 at 11:16 AM, Peter Geoghegan wrote:
> There are only tiny differences, which in any case you can see in the
> commit log on Github. There is no reason why code review needs to
> block on this V3, IMV.
Also, I don't think that we need to include V2's test
g at.)
There are only tiny differences, which in any case you can see in the
commit log on Github. There is no reason why code review needs to
block on this V3, IMV.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://
On Thu, Aug 18, 2016 at 5:14 PM, Peter Geoghegan wrote:
> I'd certainly welcome that. There are Debian packages available from
> the Github version of amcheck, which is otherwise practically
> identical to the most recent version of the patch posted here:
>
> https://githu
d on the off chance
that it mattered, without actually being justified. I would like to be
able to run pg_buffercache in production from time to time.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
o use it on all production databases. It
wouldn't have mattered that the verification was no less effective,
since the bugs it found would simply never have been observed in
practice.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
rmation about a problem. And, ideally, we'd also have some
indication of how big a difference that would make, it terms of
measurable performance impact.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
On Wed, Aug 31, 2016 at 3:08 PM, Andres Freund wrote:
> On August 31, 2016 3:06:23 PM PDT, Peter Geoghegan wrote:
>
>>In other painfully pedantic news, I should point out that
>>sizeof(size_t) isn't necessarily word size (the most generic
>>definition of word size fo
), contrary to my reading
of the 0002-* patch comments. I'm mostly talking thinking about x86_64
here, of course.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
On Sun, Nov 22, 2015 at 7:29 PM, Andreas Karlsson wrote:
> Sorry for dropping this patch, but now I have started looking at it again.
Any chance of picking this up again soon, Andreas? I think it's an
important project. I would like to review it.
--
Peter Geoghegan
--
Sent v
On Wed, Aug 31, 2016 at 2:09 PM, Peter Geoghegan wrote:
> The only thing that stuck out to any degree is that we don't grow the
> "reln->md_seg_fds[forknum]" array within the new _fdvec_resize()
> function geometrically.
That new function looks like this:
>
ly doesn't matter, given what that
array tracks. I'm just pointing out that that aspect did give me
pause. The struct MdfdVec is small enough that that *might* be
worthwhile.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your
iguration
> option perhaps in initdb to change the default so that, say, "fr_FR"
> uses ICU and "fr_FR%posix" uses the old stuff.
I suspect that we'd be better off adding a mechanism for adding a new
collation after initdb runs, on a live production instance. Maybe t
dual OS packagers of
ICU deciding what exact CLDR data to use, which may or may not be of
any significant consequence in practice.
[1] http://unicode.org/reports/tr10
[2] http://site.icu-project.org/design/size/collation
[3] http://userguide.icu-project.org/icudata
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
ery or on standby after WAL replay.
Right you are -- while BTP_INCOMPLETE_SPLIT is set during recovery,
BTP_SPLIT_END is not. Still, most of the btpo_flags flags that are
masked in the patch shouldn't be.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresq
st, because the other flags have clear-cut roles
in various atomic operations that we WAL-log.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
s special case for speculative insertion (to WAL-log
the virtually useless speculative insertion token value)? I'm certain
that the answer must be "no": This tool ought to deal with speculative
insertion as a special case, and not vice-versa.
--
Peter Geoghegan
--
Sent via pgsql-h
k,
because WAL segments consist of zeroes at the end when archive_timeout
is applied (at least from 9.4 on). We compress the WAL segments, and
many zeroes compress very well.
I admit that I haven't looked at it in much detail, but that is my
current understanding.
--
Peter Geoghegan
On Thu, Aug 25, 2016 at 12:59 PM, Peter Geoghegan wrote:
> Maybe we should change the ordering of those IndexInfo structs to
> something more suitable, but it must be immutable (it cannot hinge
> upon the details of one particular DML statement).
I meant that it must be stable (not
lly also
wrote a patch to prefer insertion into the primary key first, which
also went nowhere (I gave up on that one, to be fair).
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
ct is initially found, and so no guarantees
here.)
Anyway, I don't have a lot of sympathy for this point of view, because
the scenario is completely contrived. You have to draw the line
somewhere.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postg
27;t upsert while using more than one index
as an arbiter index. This is true unless they're more or less
equivalent, in which case multiple arbiter indexes can be inferred,
but that clearly doesn't apply here.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hacke
concern here:
https://www.postgresql.org/message-id/30619.1428157...@sss.pgh.pa.us
ISTM that we don't even care about Windows performance to a minimal
degree. Hopefully, the ICU stuff Peter Eisentraut is working on will
level the playing field here a little bit, if only as an accidental
side
is an
interesting discussion of the matter here:
http://www.unicode.org/reports/tr36/#Bidirectional_Text_Spoofing
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
I haven't looked at the patch, but offhand I wonder if it's worth
considering control characters added by unicode, if you haven't already.
--
Peter Geoghegan
thousands of servers. For some reason, both cases involved
strings with code points from the Arabic alphabet, even though each
case was from a totally unrelated customer database.
I'll go update the Wiki page for this [1] now.
[1] https://wiki.postgresql.org/wiki/Abbreviated_keys_glibc_issue
On Mon, Aug 1, 2016 at 3:18 PM, Peter Geoghegan wrote:
> Attached WIP patch series:
This has bitrot, since commit da1c9163 changed the interface for
checking parallel safety. I'll have to fix that, and will probably
take the opportunity to change how workers have maintenance_work_mem
app
yet, even though any
instances of corruption of text indexes I've seen originated before
the point release in which strxfrm() became distrusted. I guess that
not that many Heroku users use the "C" locale, which would still be
affected with the latest point release.
--
Peter Geogheg
fraction of all databases tested, so I don't think it's very common in
the wild.
I'd be surprised if amcheck does not bring more bugs like this to my
attention before too long. We should work on improving it, so that we
have greater visibility into problems that occur in the f
;s much more likely
that they'd be "false positive" bugs. In any case, I haven't seen any
issue with the tool itself yet, having now run the tool on thousands
of servers.
I think I'll have a lot more information in about a week, when I've
had time to work through more data
On Thu, Aug 18, 2016 at 5:42 PM, Andres Freund wrote:
> How large was the index & table in question? I mean this really only
> comes into effect at 100+ segments.
Not that big, but I see no reason to take the chance, I suppose.
--
Peter Geoghegan
--
Sent via pgsql-hackers ma
t that would be a
waste of time.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
On Thu, Aug 18, 2016 at 5:26 PM, Andres Freund wrote:
> Rebased version attached. A review would be welcome. Plan to push this
> forward otherwise in the not too far away future.
I can review this next week.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-h
y, I would like to make amcheck verifying the heap through a
B-Tree index as a next step. There is also a good case for the tool
directly verifying heap relations, without involving any index, but
that can come later.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgres
s necessary to be very ambitious in order
to solve a problem. The understandable and usually well-reasoned
approach of making progress as incremental as possible occasionally
works against contributors. It's worth considering if this is such a
case.
--
Peter Geoghegan
--
Sent via pgs
address this problem is with a duplicate list
and/or prefix compression in leaf pages.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
On Sat, Mar 12, 2016 at 12:38 PM, Peter Geoghegan wrote:
> Only insofar as it helps diagnose the underlying issue, when it is a
> more subtle issue. Actually fixing the index is almost certainly a
> REINDEX. Once you're into the messy business of diagnosing a
> problematic opclas
tuff in the frame pointer at that point. You either need to use
> dwarf or lbr to get accurate ones.
Is it worth doing that here, and redoing the test, so that the glibc
attributions are correct?
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To
ated that the tuplesort CLUSTER takes just under 3
minutes (this includes writing out the new heap, of course).
--
Peter Geoghegan
cycles-cluster-presorted-flamegraph.svg.gz
Description: GNU Zip compressed data
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make ch
tuples, the hash lookups in
> rewrite_heap_tuple(), ...?
Perhaps the attached svg "flamegraph" will give you some idea. This is
based on the "perf cache-misses" event type.
--
Peter Geoghegan
cache-misses-cluster-presorted-flamegraph.svg.gz
Description: GNU Zip compressed da
aightforward to have
reform_and_rewrite_tuple() work occur in parallel workers instead,
which buys us a lot.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
The ordering dependencies happen to be quite naturally
across one leader process and one or more worker processes.
I do see value in this, though.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
workers that arrive after batch 0 is complete.
Is that really so bad? In general, I don't tend to think of workers as
the cost to worry about. Rather, we should be concerned about the
active use of CPU cores as our major cost.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list
the old hash table goes away. Of course, there are
> some tricky issues with reading tapes that were originally created by
> other backends, but if I understand correctly, Peter Geoghegan has
> already done some work on that problem, and it seems like something we
> can eventually solve,
This makes a C++ port seem less
compelling to me than the idea first appears.
Note, for example, that ICU is implemented in C++, but still has C
stub functions, not necessarily for the exclusive benefit of C client
code.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-ha
On Tue, Aug 16, 2016 at 1:29 PM, Peter Geoghegan wrote:
> IMV, it would be useful to use C++ classes (and even template classes)
> for a small number of data structures, while still largely adhering to
> earlier practices (this is what GCC did). Specifically, a few modules
> such a
ngInfo, could be made to follow the RAII/scope bound
resource management usefully, which doesn't seem incompatible with
memory contexts. However, this doesn't seem terribly exciting to me.
[1] https://lwn.net/Articles/542457/
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing lis
On Wed, Aug 3, 2016 at 2:13 PM, Peter Geoghegan wrote:
> Since merging is a big bottleneck with this, we should probably also
> work to address that indirectly.
I attach a patch that changes how we maintain the heap invariant
during tuplesort merging. I already mentioned this over
ween these two functions add clarity.
tuplesort_heap_siftup() comments will also need to be updated, since
"sift up" is mentioned there.
Should I write a patch?
[1] https://en.wikipedia.org/wiki/Heap_(data_structure)
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
h any effort to consolidate the number
of spinlock acquisitions? In other words, maybe the most common idioms
should be baked into the ConditionVariable interface, which could save
callers from having to use their own mutex variable.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-
it works well with this. You
really need to create a scenario with a real sort, and all the
conditions I describe.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
I would think, which seems like a bug that is
not dodged by simply not defining TRUST_STRXFRM. Isn't its assumption
that that matching the ordering used elsewhere not really hold on
FreeBSD builds?
[1] https://wiki.postgresql.org/wiki/Abbreviated_keys_glibc_issue
--
Peter Geoghegan
--
Se
e accepted. It's not
practical to go to the trouble of preventing it entirely. So, the
comparison with quicksort works on a couple of levels.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
er, you cannot perform the final merge on-the-fly; you must
produce a serialized tape as output, which is used subsequently to
support random seeks. There is no penalty when you manage to do the
sort in memory, though (not that that has anything to do with parallel
sort).
--
Peter Geoghegan
--
initially input.
I need to do some more research before posting a patch, but right now
I can see that it makes merging presorted numeric values more than 2x
faster. And that's with 8 tapes, on my very I/O bound laptop. I bet
that the benefits would also be large for text (temporal loc
postgresql.org/message-id/CA%2BTgmoY5JYs4R1g_ZJ-P6SkULSb19xx4zUh7S8LJiXonCgVTuQ%40mail.gmail.com
[3] http://pages.cs.wisc.edu/~dewitt/includes/paralleldb/parsort.pdf
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
runs.
All of these factors are why I believe I'm able to compete well with
other systems with this relatively straightforward, evolutionary
approach. I have a completely open mind about partitioning, but my
approach makes sense in this context.
--
Peter Geoghegan
--
Sent via pgsql-hackers m
other things. Sadly, I
>>> don't have time to look at it right now.
>>
>> I would be happy to look at generalizing that further, to help
>> parallel hash join. As you know, Thomas Munro and I have discussed
>> this privately.
>
> Right.
By the way, the patch is in better shape from that perspective, as
compared to the early version Thomas (CC'd) had access to. The BufFile
stuff is now credible as a general-purpose abstraction.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
ow this, and write code
defensively because of this). The bug is that there is any ERROR for
VACUUM that isn't absolutely unavoidable (the damage has been done
anyway -- the index is already corrupt).
I'll need to think about this some more, when I have more time.
Perhaps tomorrow.
--
Pet
ndirectly.
> The work on making the logtape infrastructure parallel-aware seems
> very interesting and potentially useful for other things. Sadly, I
> don't have time to look at it right now.
I would be happy to look at generalizing that further, to help
parallel hash join. As you know, Thomas Munro and I have discussed
this privately.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
-recommend-it things.
+1.
This also has value in the context of automatically surfacing
situations where "can't happen" errors do in fact happen at scale.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
the utility statement equivalent of
max_parallel_workers_per_gather. This is clearly necessary, since
we're using up to maintenance_work_mem per worker, which is of course
typically much higher than work_mem. I didn't feel the need to create
a new maintenance-wise variant GUC for th
d given our awful track record about maintaining this
> documentation, I'm not sure that going into more detail is really
> a good idea. Comments?
+1
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http:/
now not so sure that that's actually the case.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
8, for example. There were other locales that were
affected less severely, and I think the majority were not shown to be
affected at all.
That being said, it probably wouldn't have caught that particular
issue if we had broad coverage. It probably would catch a broken test,
though.
--
Pete
> on a buildfarm animal should work" anyway. I'm much more interested in
> supporting locales that someone cares enough about to configure a
> buildfarm animal for.
That seems like a high standard to me. Locale rules are known to
change, and are explicitly versioned by glibc, for e
for.
>
> Nah, we have a hard enough time with reproducibility of buildfarm results
> without deliberately injecting transient failures.
It could be pseudo-random, and so deterministic per buildfarm animal.
That's what I did.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing
27;s more or less what I did with the amcheck regression tests.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
rt test
cases [1] take as much as 15% less time to execute overall. That's a
big difference. I looked at the disassembly, and the number of
instructions for varstrfastcmp_c() was reduced from 113 to 29. That's
the kind of difference that could add up to a lot.
[1] https://github.com/peter
e
called batch memory) for just a few modules, and what remains isn't
several broad swathes that can be delineated easily. I can see a
"palloc a lot and don't worry too much about pfrees" allocator having
some value, but I suspect that that isn't going to move the needle
greed.
Also, for what it's worth, I should point out to Jun that
GetCurrentTimestamp() should definitely not be called when a spinlock
is held like that.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
htt
. It's unfortunate that there isn't a good
third-party tool that does that, but there is nothing that prevents
it.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
On Thu, Apr 14, 2016 at 4:03 PM, Peter Geoghegan wrote:
> Attached patch removes obsolete comment from fmgr.c.
This patch seems to have been overlooked. It's a pretty straightforward case.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
correlation performs compared to a bitmap index scan of a B-Tree (due
to a less selective bitmap index scan qual, as in your example), but
offhand I guess that could be faster in general, making the bottleneck
you're addressing relatively greater there.
--
Peter Geoghegan
--
Sent via pgsql-hack
rom 2461.622
> to 952.161.
That's pretty great. I wonder what this would look like with a BRIN
index, since l_shipdate looks like a good candidate for BRIN indexing.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
On Wed, Jul 13, 2016 at 12:24 PM, Tom Lane wrote:
>> I am happy with the adjustment. Please commit the adjusted patch.
>
> Done with minor adjustments.
Thanks. I'm pleased that we found a way forward that addressed every concern.
--
Peter Geoghegan
--
Sent via pgsql-hac
and | lthousand | tenthous | ltenthous
> -+--+--+---+--+---
> (0 rows)
It independently occurred to me that I should have done something like
this afterwards. I agree.
> If you're good with that adjustment, I'm happy to commit this.
I am happy with the adjustmen
structures, leaving a
cache-friendly layout).
I suspect that there are not that many places where it is worth it to
even contemplate batch or dense allocators, so I doubt that what we
will see all that many more instances of "local allocators".
--
Peter Geoghegan
--
Sent via pgs
t select returning id
> ) insert into bar(foo_id,i)
> select id,2 from _foo;
I gather that the point of this pseudo SQL is to show how you might be
able to project and select the values not successfully inserted. Can't
you just pipeline together some CTEs instead?
--
Peter Geoghe
th checkpoints starting
> because of XLOG, but there's no indication of that being a bad thing.
I agree. checkpoint_warning exists for the benefit of novice DBAs.
I've seen those warnings in customer logs on several occasions, at
least back when I was a consultant.
--
Peter Geoghegan
On Sat, Jul 9, 2016 at 1:48 PM, Alvaro Herrera wrote:
>> How about
>>
>> Time: 1234567.666 ms (20m 34.6s)
>>
>> ?
>
> +1 LGTM
+1
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription
Testing replacement selection in the second CLUSTER is made very
convenient by the fact that we just ran CLUSTER, so input should be
presorted.
--
Peter Geoghegan
0001-Add-test-coverage-for-CLUSTER-external-sorts.patch.gz
Description: GNU Zip compressed data
--
Sent via pgsql-hacker
orrectly)? Apparently that's something that's been discussed a few
times among senior community members, and I think it has promise.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
On Thu, Jul 7, 2016 at 2:52 PM, Corey Huinker wrote:
> Wouldn't it be great if we had a way of printing timing in more human
> friendly formats?
Yes, it would. I've thought about doing this myself. So, +1 to the idea from me.
--
Peter Geoghegan
--
Sent via pgsql-hackers mai
On Thu, Jul 7, 2016 at 10:51 AM, Robert Haas wrote:
> Thanks for testing. I've committed this patch, breaking off one
> unrelated bit of into a separate commit.
Thank you. To be clear, I still intend to follow up with a CLUSTER
external sort test case, as outlined to Noah.
--
Pete
ately described steps to reproduce?
>
> I can confirm that (after 62 minutes) your test procedure reached SIGSEGV
> today and then completed successfully with your patch.
Thanks for going to the trouble of confirming that the test procedure
causes a segmentation fault, and that my patch app
On Tue, Jul 5, 2016 at 12:58 PM, Tom Lane wrote:
> Perhaps we could change the wording of temp_file_limit's description
> from "space that a session can use" to "space that a process can use"
> to help clarify this?
That's all that I was looking for, real
use temp_buffers directly in practice. max_files_per_process is
already clearly per process, so no change needed there either.
I don't see a case other than temp_file_limit that appears to be even
marginally in need of a specific note.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgs
n limit, whereas users are quite
used to the idea that work_mem might be doled out multiple times for
multiple executor nodes.
--
Peter Geoghegan
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
601 - 700 of 3472 matches
Mail list logo