On 01/17/2016 10:03 PM, Jeff Janes wrote:
On Fri, Jan 15, 2016 at 3:29 PM, Peter Geoghegan wrote:
On Fri, Jan 15, 2016 at 2:38 PM, Constantin S. Pan wrote:
I have a draft implementation which divides the whole process between
N parallel workers, see the patch attached. Instead of a full scan
On Wed, Sep 14, 2016 at 3:32 PM, Tom Lane wrote:
> Michael Paquier writes:
>> From an OSX laptop with -S, -c 1 and -M prepared (9 runs, removed the
>> three best and three worst):
>> - HEAD: 9356/9343/9369
>> - HEAD + patch: 9433/9413/9461.071168
>> This laptop has a lot of I/O overhead... Still
Michael Paquier writes:
> From an OSX laptop with -S, -c 1 and -M prepared (9 runs, removed the
> three best and three worst):
> - HEAD: 9356/9343/9369
> - HEAD + patch: 9433/9413/9461.071168
> This laptop has a lot of I/O overhead... Still there is a slight
> improvement here as well. Looking at
On Wed, Sep 14, 2016 at 2:56 PM, Kuntal Ghosh
wrote:
> Master
> ---
> - If wal_consistency check is enabled or needs_backup is set in
> XLogRecordAssemble(), we do a fpw.
> - If a fpw is to be done, then fork_flags is set with BKPBLOCK_HAS_IMAGE,
> which in turns set has_image flag whi
On Wed, Sep 14, 2016 at 6:34 AM, Michael Paquier
wrote:
>>>2) Regarding page comparison:
>>>We could just use memcpy() here. compareImages was useful to get a
>>>clear image of what the inconsistencies were, but you don't do that
>>>anymore.
>> memcmp(), right?
>
>Yep :)
If I use memcmp(), I won't
On Wed, Sep 14, 2016 at 7:06 AM, Tom Lane wrote:
> It would be good for someone else to reproduce my results though.
> For one thing, 5%-ish is not that far above the noise level; maybe
> what I'm measuring here is just good luck from relocation of critical
> loops into more cache-line-friendly lo
On Mon, Sep 5, 2016 at 9:33 AM, Amit Kapila wrote:
> USE_CONTENT_LOCK on my windows box, you can try by commenting that as
> well, if it works for you). So, in short we have to compare three
> approaches here.
>
> 1) Group mode to reduce CLOGControlLock contention
> 2) Use granular locking model
Hello,
It looks to me pg_buffercache tries to allocate more than 1GB using
palloc(), when shared_buffers is more than 256GB.
# show shared_buffers ;
shared_buffers
280GB
(1 row)
# SELECT buffers, d.datname, coalesce(c.relname, '???')
FROM (SELECT count(*) buffers, reldatab
On Wed, Sep 14, 2016 at 12:06 AM, Tom Lane wrote:
> I wrote:
>> At -j 10 -c 10, all else the same, I get 84928 TPS on HEAD and 90357
>> with the patch, so about 6% better.
>
> And at -j 1 -c 1, I get 22390 and 24040 TPS, or about 7% better with
> the patch. So what I am seeing on OS X isn't conte
On Tue, Sep 13, 2016 at 10:28 PM, Robert Haas wrote:
> On Tue, Sep 6, 2016 at 9:07 AM, Ashutosh Bapat
> wrote:
>> That's not true with the alias information. As long as we detect which
>> relations need subqueries, their RTIs are enough to create unique aliases
>> e.g. if a relation involves RTIs
On Wed, Sep 14, 2016 at 12:21 AM, Robert Haas wrote:
> On Fri, Sep 9, 2016 at 3:04 AM, Masahiko Sawada
> wrote:
> > Attached PoC patch changes the representation of dead tuple locations
> > to the hashmap having tuple bitmap.
> > The one hashmap entry consists of the block number and the TID bit
On Thu, Aug 11, 2016 at 2:47 PM, Robert Haas wrote:
> Another approach to the problem is to use a latch wait loop. That
> almost works. Interrupts can be serviced, and you can recheck shared
> memory to see whether the condition for proceeding is satisfied after
> each iteration of the loop. Th
Spotted a typo in the header comment of execIndexing.c. Attached fixes it.
s/exclusive constraints/exclusion constraints/g
Thanks,
Amit
diff --git a/src/backend/executor/execIndexing.c b/src/backend/executor/execIndexing.c
index 0e2d834..009c1b7 100644
--- a/src/backend/executor/execIndexing.c
+
On Tue, Sep 13, 2016 at 5:46 PM, Robert Haas wrote:
> On Thu, Sep 8, 2016 at 12:32 AM, Amit Kapila wrote:
>> Hmm. I think page or block is a concept of database systems and
>> buckets is a general concept used in hashing technology. I think the
>> difference is that there are primary buckets an
> As I understand the issue, the main reason is that event triggers
> execute procedures, and those exist in a single database only. If you
> were to create an event trigger in database A, then a user gets created
> in database B, your function would not be invoked, which becomes a
> problem.
Can
On 14 Sep. 2016 9:44 am, "Alvaro Herrera" wrote:
>
> Tatsuo Ishii wrote:
> > Simple question: Is there any reason for event trigger to not support
> > CREATE/ALTER ROLE/USER?
>
> As I understand the issue, the main reason is that event triggers
> execute procedures, and those exist in a single dat
Tatsuo Ishii wrote:
> Simple question: Is there any reason for event trigger to not support
> CREATE/ALTER ROLE/USER?
As I understand the issue, the main reason is that event triggers
execute procedures, and those exist in a single database only. If you
were to create an event trigger in database
On Wed, Sep 14, 2016 at 2:17 AM, Robert Haas wrote:
> On Thu, Sep 8, 2016 at 1:32 PM, Masahiko Sawada wrote:
>> Attached patch add --disable-page-skipping option to vacuumed command for
>> 9.6.
>> If by any chance the freeze map causes the serious data corruption bug
>> then the user will want t
> On Tue, Sep 13, 2016 at 3:48 AM, Kouhei Kaigai wrote:
> > Sorry for my late.
> >
> > The attached patch fixed the wording problems on SGML part.
>
> I agree that we should have some way for foreign data wrappers and
> custom scans and perhaps also other executor nodes to find out whether
> ther
On Tue, Sep 13, 2016 at 6:07 PM, Kuntal Ghosh
wrote:
>> For now, I've kept this as a WARNING message to detect all inconsistencies
>> at once. Once, the patch is finalized, I'll modify it as an ERROR message.
>
> Or say FATAL. This way the server is taken down.
What I'd really like to see here is
On Thu, Aug 25, 2016 at 3:12 PM, Haribabu Kommi
wrote:
> On Thu, Jun 9, 2016 at 12:56 AM, Tom Lane wrote:
> > Thom Brown writes:
> >> On 15 May 2014 at 19:56, Bruce Momjian wrote:
> >>> On Tue, May 13, 2014 at 06:58:11PM -0400, Tom Lane wrote:
> A recent question from Tim Kane prompted me
Simple question: Is there any reason for event trigger to not support
CREATE/ALTER ROLE/USER?
>>>
>>> Because it performs a global administrative task, no?
>>
>> So is there any technical difficulty?
>
> I don't recall the exact details but...
>
>> Can you please elaborate?
>
> Roles a
On Wed, Sep 14, 2016 at 9:34 AM, Tatsuo Ishii wrote:
>>> Simple question: Is there any reason for event trigger to not support
>>> CREATE/ALTER ROLE/USER?
>>
>> Because it performs a global administrative task, no?
>
> So is there any technical difficulty?
I don't recall the exact details but...
>> Simple question: Is there any reason for event trigger to not support
>> CREATE/ALTER ROLE/USER?
>
> Because it performs a global administrative task, no?
So is there any technical difficulty? Can you please elaborate?
Best regards,
--
Tatsuo Ishii
SRA OSS, Inc. Japan
English: http://www.srao
On Wed, Sep 14, 2016 at 9:26 AM, Tatsuo Ishii wrote:
> Simple question: Is there any reason for event trigger to not support
> CREATE/ALTER ROLE/USER?
Because it performs a global administrative task, no?
--
Michael
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make
Simple question: Is there any reason for event trigger to not support
CREATE/ALTER ROLE/USER?
Best regards,
--
Tatsuo Ishii
SRA OSS, Inc. Japan
English: http://www.sraoss.co.jp/index_en.php
Japanese:http://www.sraoss.co.jp
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To
On Wed, Sep 14, 2016 at 6:52 AM, Peter Eisentraut
wrote:
> On 9/12/16 3:31 PM, Tom Lane wrote:
>> Hm, there was just a kerfuffle about spelling things like this
>> "--no-clean" etc. I wasn't on board with removing existing spellings,
>> but surely all newly added switches should use the "no-" spe
On Tue, Sep 13, 2016 at 5:06 PM, Petr Jelinek wrote:
> I mostly agree. I think if this was called something like background
> transactions it might be better. It's definitely useful functionality but
> the naming is clearly contentious. It won't stop people using it for same
> use-cases as autonom
Hi,
First read through the current version. Hence no real architectural
comments.
On 2016-09-09 00:59:26 +0200, Petr Jelinek wrote:
> diff --git a/src/backend/commands/publicationcmds.c
> b/src/backend/commands/publicationcmds.c
> new file mode 100644
> index 000..e0c719d
> --- /dev/null
>
I wrote:
> At -j 10 -c 10, all else the same, I get 84928 TPS on HEAD and 90357
> with the patch, so about 6% better.
And at -j 1 -c 1, I get 22390 and 24040 TPS, or about 7% better with
the patch. So what I am seeing on OS X isn't contention of any sort,
but just a straight speedup that's indepe
Hi,
Thanks for looking into this!
On 09/12/2016 04:08 PM, Dean Rasheed wrote:
On 3 August 2016 at 02:58, Tomas Vondra wrote:
Attached is v19 of the "multivariate stats" patch series
Hi,
I started looking at this - just at a very high level - I've not read
much of the detail yet, but here a
On Tue, Sep 13, 2016 at 10:00 PM, Alvaro Herrera
wrote:
> Yeah, I looked into this a few days ago and that was my conclusion also:
> let's drop this.
Okay. Just done so in the CF app.
--
Michael
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subs
On 9/12/16 3:31 PM, Tom Lane wrote:
> Hm, there was just a kerfuffle about spelling things like this
> "--no-clean" etc. I wasn't on board with removing existing spellings,
> but surely all newly added switches should use the "no-" spelling?
This is supposed to be parallel to initdb's option. So
I wrote:
> Mark Dilger writes:
>> there are several places in the code where variables defined as
>> (char *) or as (const char *) are passed to the NameGetDatum()
>> macro. I believe it would be better form to use CStringGetDatum()
>> in these locations. I am aware that these two macros are int
On 13/09/16 22:24, Robert Haas wrote:
On Wed, Sep 7, 2016 at 9:14 PM, Craig Ringer
wrote:
That's probably going to be one of the smaller costs. Doing this with
bgworkers won't be cheap, but you need to consider the alternative.
Factoring out all transaction-specific data currently stored in or
On 7/30/16 2:16 PM, Bruce Momjian wrote:
> The second patch does what Tom suggests above by outputting only "KB",
> and it supports "kB" for backward compatibility. What it doesn't do is
> to allow arbitrary case, which I think would be a step backward. The
> second patch actually does match the
On 13/09/16 02:55, Craig Ringer wrote:
On 13 September 2016 at 06:03, Petr Jelinek wrote:
Oh sure, I don't see that as big problem, the TupleData already contains
type of the data it sends (to distinguish between nulls and text data) so
that's mostly about adding some different type there and
Alvaro Herrera writes:
> Yeah, as I recall the only thing the get_new_node thingy does is assign
> a nonconflicting port number to each instance, and make sure the
> instances are all teared down at END. I don't remember now why didn't
> we just do the port check in the constructor, but we messed
Victor Wagner wrote:
> Hi hackers!
>
> I've encountered need to extend functionality of PostgresNode class
> from the TAP test framework. What I want can easily be done via perl
> object inheritation.
>
> But documentation in the PostgresNode.pm recommends to use get_new_node
> function rather t
Andres Freund writes:
> On 2016-09-13 15:37:22 -0400, Tom Lane wrote:
>> (It's a 4-core CPU so I saw little point in pressing harder than
>> that.)
> I think in reality most busy machines, were performance and scalability
> matter, are overcommitted in the number of connections vs. cores. And
>
On Wed, Sep 7, 2016 at 9:14 PM, Craig Ringer
wrote:
> That's probably going to be one of the smaller costs. Doing this with
> bgworkers won't be cheap, but you need to consider the alternative.
> Factoring out all transaction-specific data currently stored in or
> pointed to by globals into a tran
On Thu, Sep 8, 2016 at 11:21 PM, Jesper Pedersen wrote:
>
> > For the archives, this patch conflicts with the WAL patch [1].
>
> > [1] https://www.postgresql.org/message-id/CAA4eK1JS%2BSiRSQBzEFp
> nsSmxZKingrRH7WNyWULJeEJSj1-%3D0w%40mail.gmail.com
>
Updated the patch it applies over Amit's concu
On 2016-09-13 15:37:22 -0400, Tom Lane wrote:
> Andres Freund writes:
> > On 2016-09-13 14:47:08 -0400, Tom Lane wrote:
> >> Also I notice that the WaitEventSet thread started with a simple
> >> pgbench test, so I don't really buy the claim that that's not a
> >> way that will reach the problem.
>
Andres Freund writes:
> On 2016-09-13 14:47:08 -0400, Tom Lane wrote:
>> Also I notice that the WaitEventSet thread started with a simple
>> pgbench test, so I don't really buy the claim that that's not a
>> way that will reach the problem.
> You can reach it, but not when using 1 core:one pgbenc
On Tue, Sep 13, 2016 at 4:06 PM, Robert Haas wrote:
> On Tue, Sep 13, 2016 at 2:59 PM, Claudio Freire
> wrote:
>> I've finished writing that patch, I'm in the process of testing its CPU
>> impact.
>>
>> First test seemed to hint at a 40% increase in CPU usage, which seems
>> rather steep compar
On Tue, Sep 13, 2016 at 2:59 PM, Claudio Freire wrote:
> I've finished writing that patch, I'm in the process of testing its CPU
> impact.
>
> First test seemed to hint at a 40% increase in CPU usage, which seems
> rather steep compared to what I expected, so I'm trying to rule out
> some methodo
On Fri, Sep 9, 2016 at 4:49 AM, Victor Wagner wrote:
> Random permutation is much more computationally complex than random
> picking.
It really isn't. The pseudocode given at
https://en.wikipedia.org/wiki/Fisher%E2%80%93Yates_shuffle is all of 4
lines long, and one of those lines is a comment.
On Tue, Sep 13, 2016 at 3:51 PM, Robert Haas wrote:
> On Fri, Sep 9, 2016 at 3:04 AM, Masahiko Sawada wrote:
>> Attached PoC patch changes the representation of dead tuple locations
>> to the hashmap having tuple bitmap.
>> The one hashmap entry consists of the block number and the TID bitmap
>>
On 09/13/2016 07:26 AM, Amit Kapila wrote:
Attached, new version of patch which contains the fix for problem
reported on write-ahead-log of hash index thread [1].
I have been testing patch in various scenarios, and it has a positive
performance impact in some cases.
This is especially seen
On Tue, Sep 13, 2016 at 11:51 AM, Robert Haas wrote:
> I think it's probably wrong to worry that an array-of-arrays is going
> to be meaningfully slower than a single array here. It's basically
> costing you some small number of additional memory references per
> tuple, which I suspect isn't all
Andres Freund writes:
> On 2016-09-13 12:43:36 -0400, Tom Lane wrote:
>> Also, if it's only a win on machines with dozens of CPUs, how many
>> people are running *BSD on that kind of iron? I think Linux is by
>> far the dominant kernel for such hardware. For sure Apple isn't
>> selling any machi
On 2016-09-13 14:47:08 -0400, Tom Lane wrote:
> Also I notice that the WaitEventSet thread started with a simple
> pgbench test, so I don't really buy the claim that that's not a
> way that will reach the problem.
You can reach it, but not when using 1 core:one pgbench thread:one
client connection
On Fri, Sep 9, 2016 at 3:04 AM, Masahiko Sawada wrote:
> Attached PoC patch changes the representation of dead tuple locations
> to the hashmap having tuple bitmap.
> The one hashmap entry consists of the block number and the TID bitmap
> of corresponding block, and the block number is the hash ke
On 09/13/2016 07:41 AM, Amit Kapila wrote:
README:
+in_complete split flag. The reader algorithm works correctly, as it will
scan
What flag ?
in-complete-split flag which indicates that split has to be finished
for that particular bucket. The value of these flags are
LH_BUCKET_NEW_PAGE_SPLI
On 2016-09-13 12:43:36 -0400, Tom Lane wrote:
> > I think it's not necessarily about the current system, but more about
> > future uses of the WaitEventSet stuff. Some of that is going to use a
> > lot more sockets. E.g. doing a parallel append over FDWs.
(note that I'm talking about network socke
Hi hackers!
I've encountered need to extend functionality of PostgresNode class
from the TAP test framework. What I want can easily be done via perl
object inheritation.
But documentation in the PostgresNode.pm recommends to use get_new_node
function rather than call PostgresNode constructor dir
Andres Freund writes:
> On 2016-09-13 12:07:35 -0400, Tom Lane wrote:
>> /*
>> - * Don't pull up a subquery that has any set-returning functions in its
>> - * targetlist. Otherwise we might well wind up inserting set-returning
>> - * functions into places where they mustn't go, such a
Hi hackers!
Here is gist cube penaly patch v5.
Changes:
1. union version of pack_float() is used, without warings and with
minimum asm commands emitted
2. test for __STDC_IEC_559__ is added, this is a standard way of C99
to determine float compliance with IEEE754. ANSI-only compiler+libs,
along wi
On 2016-09-13 12:07:35 -0400, Tom Lane wrote:
> diff --git a/src/backend/optimizer/plan/analindex e28a8dc..74e4245 100644
> --- a/src/backend/optimizer/plan/analyzejoins.c
> +++ b/src/backend/optimizer/plan/analyzejoins.c
> @@ -650,6 +650,11 @@ rel_is_distinct_for(PlannerInfo *root, R
> bool
> qu
Andres Freund writes:
> On September 13, 2016 9:07:35 AM PDT, Tom Lane wrote:
>> I'd like to go ahead and push this, since it's a necessary prerequisite
>> for either approach we might adopt for the rest of the patch series,
>> and the improved error reporting and avoidance of expensive
>> expres
Robert Haas writes:
> I agree that we should have some way for foreign data wrappers and
> custom scans and perhaps also other executor nodes to find out whether
> there's a known limit to the number of tuples that they might need to
> produce, but I wonder if we should be doing something more gen
On Thu, Sep 8, 2016 at 1:32 PM, Masahiko Sawada wrote:
> Attached patch add --disable-page-skipping option to vacuumed command for 9.6.
> If by any chance the freeze map causes the serious data corruption bug
> then the user will want to run pg_check_visible() and vacuum with
> DISABLE_PAGE_SKIPPI
On Tue, Sep 6, 2016 at 9:07 AM, Ashutosh Bapat
wrote:
> That's not true with the alias information. As long as we detect which
> relations need subqueries, their RTIs are enough to create unique aliases
> e.g. if a relation involves RTIs 1, 2, 3 corresponding subquery can have
> alias r123 without
On Tue, Sep 13, 2016 at 3:48 AM, Kouhei Kaigai wrote:
> Sorry for my late.
>
> The attached patch fixed the wording problems on SGML part.
I agree that we should have some way for foreign data wrappers and
custom scans and perhaps also other executor nodes to find out whether
there's a known limi
Andres Freund writes:
> On 2016-09-13 16:08:39 +0300, Heikki Linnakangas wrote:
>> So, if I've understood correctly, the purpose of this patch is to improve
>> performance on a multi-CPU system, which has the kqueue() function. Most
>> notably, FreeBSD?
> I think it's not necessarily about the cu
On Wed, Sep 7, 2016 at 9:32 PM, Amit Kapila wrote:
> On Wed, Sep 7, 2016 at 11:49 PM, Jeff Janes wrote:
> > On Thu, Sep 1, 2016 at 8:55 PM, Amit Kapila
> wrote:
> >>
> >>
> >> I have fixed all other issues you have raised. Updated patch is
> >> attached with this mail.
> >
> >
> > I am finding
On Tue, Sep 13, 2016 at 5:59 AM, Robert Haas wrote:
> I think that the only real way to judge whether cost_sort() is any
> good is to see whether it causes the wrong plan to be chosen in some
> cases. For example, it could cause merge joins to be picked too
> frequently or too infrequently, or it
On September 13, 2016 9:07:35 AM PDT, Tom Lane wrote:
>Andres Freund writes:
>> Attached is a significantly updated patch series (see the mail one up
>> for details about what this is, I don't want to quote it in its
>> entirety).
>
>I've reviewed the portions of 0005 that have to do with makin
Andres Freund writes:
> Attached is a significantly updated patch series (see the mail one up
> for details about what this is, I don't want to quote it in its
> entirety).
I've reviewed the portions of 0005 that have to do with making the parser
mark queries with hasTargetSRF. The code as you h
On Tue, Sep 13, 2016 at 11:36 AM, Simon Riggs wrote:
> On 13 September 2016 at 08:08, Heikki Linnakangas wrote:
>> So, if I've understood correctly, the purpose of this patch is to improve
>> performance on a multi-CPU system, which has the kqueue() function. Most
>> notably, FreeBSD?
>
> I'm get
On 13 September 2016 at 08:08, Heikki Linnakangas wrote:
> So, if I've understood correctly, the purpose of this patch is to improve
> performance on a multi-CPU system, which has the kqueue() function. Most
> notably, FreeBSD?
I'm getting a little fried from "self-documenting" patches, from
mul
Hi,
On 2016-09-13 16:08:39 +0300, Heikki Linnakangas wrote:
> So, if I've understood correctly, the purpose of this patch is to improve
> performance on a multi-CPU system, which has the kqueue() function. Most
> notably, FreeBSD?
I think it's not necessarily about the current system, but more a
On 09/13/2016 04:33 PM, Tom Lane wrote:
Heikki Linnakangas writes:
So, if I've understood correctly, the purpose of this patch is to
improve performance on a multi-CPU system, which has the kqueue()
function. Most notably, FreeBSD?
OS X also has this, so it might be worth trying on a multi-CP
On Tue, Aug 2, 2016 at 3:41 AM, Kyotaro HORIGUCHI
wrote:
> Thank you for the comment.
>
> At Mon, 1 Aug 2016 10:44:56 +0530, Amit Khandekar
> wrote in
>> On 21 July 2016 at 15:20, Kyotaro HORIGUCHI > > wrote:
>>
>> >
>> > After some consideration, I found that ExecAsyncWaitForNode
>> > cannot b
On Mon, Aug 29, 2016 at 4:08 AM, Kyotaro HORIGUCHI
wrote:
> [ new patches ]
+/*
+ * We assume that few nodes are async-aware and async-unaware
+ * nodes cannot be revserse-dispatched from lower nodes that is
+ * async-aware. Firing of an async node
On Tue, Sep 13, 2016 at 6:58 PM, Tom Lane wrote:
> Amit Kapila writes:
>> While working on write-ahead-logging of hash indexes, I noticed that
>> this function allocates buckets in batches and the mechanism it uses
>> is that it initialize the last page of batch with zeros and expect
>> that the
Craig Ringer writes:
> While updating an extension for 9.6 I noticed that while the
> $(prove_check) definition is exposed for use by PGXS in
> Makefile.global, extensions can't actually use the TAP tests because
> we don't install the required Perl modules like PostgresNode.pm.
> I don't see any
Heikki Linnakangas writes:
> So, if I've understood correctly, the purpose of this patch is to
> improve performance on a multi-CPU system, which has the kqueue()
> function. Most notably, FreeBSD?
OS X also has this, so it might be worth trying on a multi-CPU Mac.
> If there's no measurable d
Amit Kapila writes:
> While working on write-ahead-logging of hash indexes, I noticed that
> this function allocates buckets in batches and the mechanism it uses
> is that it initialize the last page of batch with zeros and expect
> that the filesystem will ensure the intervening pages read as zer
On Sun, Sep 11, 2016 at 2:05 PM, Peter Geoghegan wrote:
> On Sun, Sep 11, 2016 at 6:28 AM, Heikki Linnakangas wrote:
>> Pushed this "displace root" patch, with some changes:
>
> Attached is rebased version of the entire patch series, which should
> be applied on top of what you pushed to the mast
So, if I've understood correctly, the purpose of this patch is to
improve performance on a multi-CPU system, which has the kqueue()
function. Most notably, FreeBSD?
I launched a FreeBSD 10.3 instance on Amazon EC2 (ami-e0682b80), on a
m4.10xlarge instance. That's a 40 core system, biggest avai
Michael Paquier wrote:
> On Tue, Sep 13, 2016 at 10:42 AM, Kyotaro HORIGUCHI
> wrote:
> > If we take a policy to try to imitate the behavior of some
> > reference platform (specifically Linux) on other platforms, this
> > is required disguising. Another potential policy on this problem
> > is "fol
On Sun, Sep 11, 2016 at 12:13 PM, Peter Geoghegan wrote:
> On Sun, Sep 11, 2016 at 9:01 AM, Tom Lane wrote:
>> Peter Geoghegan writes:
>>> I think that we *can* refine this guess, and should, because random
>>> I/O is really quite unlikely to be a large cost these days (I/O in
>>> general often
While working on write-ahead-logging of hash indexes, I noticed that
this function allocates buckets in batches and the mechanism it uses
is that it initialize the last page of batch with zeros and expect
that the filesystem will ensure the intervening pages read as zeroes
too.
I think to make it
On Thu, Sep 8, 2016 at 12:32 AM, Amit Kapila wrote:
> Hmm. I think page or block is a concept of database systems and
> buckets is a general concept used in hashing technology. I think the
> difference is that there are primary buckets and overflow buckets. I
> have checked how they are referred
On 09/12/2016 10:42 PM, Amit Kapila wrote:
The following script hangs on idx_val creation - just with v5, WAL patch
not applied.
Are you sure it is actually hanging? I see 100% cpu for a few minutes but
the index eventually completes ok for me (v5 patch applied to today's
master).
It comple
Attached, new version of patch which contains the fix for problem
reported on write-ahead-log of hash index thread [1].
[1] -
https://www.postgresql.org/message-id/CAA4eK1JuKt%3D-%3DY0FheiFL-i8Z5_5660%3D3n8JUA8s3zG53t_ArQ%40mail.gmail.com
--
With Regards,
Amit Kapila.
EnterpriseDB: http://www.e
On Mon, Sep 12, 2016 at 11:29 AM, Jeff Janes wrote:
>
>
> My test program (as posted) injects crashes and then checks the
> post-crash-recovery system for consistency, so it cannot be run as-is
> without the WAL patch. I also ran the test with crashing turned off (just
> change the JJ* variables
Craig Ringer wrote:
> I suggest that the above patches be applied to 9.6 and v10. Then for
> v10
I don't object to patching 9.6 in this way, but kindly do not pollute
this thread with future ideas on what to do on pg10, at least until the
current release is sorted out. You'll only distract peopl
On Tue, Sep 13, 2016 at 4:05 PM, Albe Laurenz wrote:
> I just noticed something surprising:
>
> -- create a larger local table
> CREATE TABLE llarge (id integer NOT NULL, val integer NOT NULL);
> INSERT INTO llarge SELECT i, i%100 FROM generate_series(1, 1) i;
> ALTER TABLE llarge ADD PRIMARY
I just noticed something surprising:
-- create a larger local table
CREATE TABLE llarge (id integer NOT NULL, val integer NOT NULL);
INSERT INTO llarge SELECT i, i%100 FROM generate_series(1, 1) i;
ALTER TABLE llarge ADD PRIMARY KEY (id);
-- create a small local table
CREATE TABLE small (id i
- If WAL consistency check is enabled for a rmgrID, we always include
the backup image in the WAL record.
>>>
>>>What happens if wal_consistency has different settings on a standby
>>>and its master? If for example it is set to 'all' on the standby, and
>>>'none' on the master, or vice-ve
On 09/08/2016 09:35 AM, Kyotaro HORIGUCHI wrote:
Returning in UTF-8 bloats the result string by about 1.5 times so
it doesn't seem to make sense comparing with it. But it takes
real = 47.35s.
Nice!
I was hoping that this would also make the binaries smaller. A few dozen
kB of storage is perha
Hello Amit,
[...]
There still seems to be a change in behavior of the -r option due to the
patch. Consider the following example:
select a from a where a = 1 \;
select a+1 from a where a = 1;
...
- statement latencies in milliseconds:
2.889 select a from a where a = 1 ;
vs
select
On Fri, Apr 8, 2016 at 10:09 PM, Peter Geoghegan wrote:
> On Wed, Mar 30, 2016 at 8:02 AM, Alexander Korotkov
> wrote:
> > Hmm... I'm not completely agree with that. In typical usage partial sort
> > should definitely use quicksort. However, fallback to other sort
> methods is
> > very useful.
On Fri, Sep 9, 2016 at 1:25 PM, Tom Lane wrote:
> Andres Freund writes:
>> On 2016-09-08 20:15:46 -0400, Peter Eisentraut wrote:
>>> We don't support build directories with spaces in them, but we support
>>> installation directories with spaces in them. So I guess that means
>>> your point is va
Sorry for my late.
The attached patch fixed the wording problems on SGML part.
Best regards,
--
NEC Business Creation Division / PG-Strom Project
KaiGai Kohei
> -Original Message-
> From: Jeevan Chalke [mailto:jeevan.cha...@enterprisedb.com]
> Sent: Tuesday, September 06, 2016 11:22 PM
On Mon, Sep 12, 2016 at 7:12 PM, Pavan Deolasee
wrote:
> I was thinking about locking, bulk reading (like page-mode API) etc. While
> you've added an API for vacuuming, we would probably also need an API to
> collect dead tuples, pruning etc (not sure if that can be combined with
> vacuum). Of cou
On 13 September 2016 at 14:36, Craig Ringer wrote:
>
>
> prove_check:
> rm -rf $(CURDIR)/tmp_check/log
> cd $(srcdir) && TESTDIR='$(CURDIR)' PATH="$(shell pg_config
> --bindir):$$PATH" PGPORT='6$(DEF_PGPORT)'
> top_builddir='$(CURDIR)/$(top_builddir)' PG_REGRESS='pg_regress'
> $(P
On Tue, Sep 13, 2016 at 10:42 AM, Kyotaro HORIGUCHI
wrote:
> If we take a policy to try to imitate the behavior of some
> reference platform (specifically Linux) on other platforms, this
> is required disguising. Another potential policy on this problem
> is "following the platform's behavior". Fr
100 matches
Mail list logo
