On 15 June 2016 at 06:48, David G. Johnston
wrote:
>
> We could stand to be more explicit here. The docs for version()
> indicated the server_version_num should be used for "machine processing".
>
> The implied correct way to access the canonical server version is
Hi,
On 06/16/2016 01:00 AM, Tom Lane wrote:
Tomas Vondra writes:
Attached is a reworked patch, mostly following the new design proposal
from this thread.
I've whacked this around quite a bit and am now reasonably happy with it.
The issue of planning speed hit
Alvaro Herrera wrote:
> Andrew Gierth wrote:
> > Why is the correct rule not "check for and ignore pre-upgrade mxids
> > before even trying to fetch members"?
>
> I propose something like the attached patch, which implements that idea.
Hm, this doesn't apply cleanly to 9.4. I'll need to come
On Fri, Jun 17, 2016 at 3:20 AM, Robert Haas wrote:
>
> On Thu, Jun 16, 2016 at 12:16 PM, Robert Haas
wrote:
> > On Thu, Jun 16, 2016 at 8:00 AM, Amit Kapila
wrote:
> >>> 1. The case originally reported by Thomas Munro still
On Wed, Jun 15, 2016 at 08:56:52AM -0400, Robert Haas wrote:
> On Wed, Jun 15, 2016 at 2:41 AM, Thomas Munro
> wrote:
> > I spent some time chasing down the exact circumstances. I suspect
> > that there may be an interlocking problem in heap_update. Using the
> >
Amit Kapila writes:
> On Thu, Jun 16, 2016 at 11:49 PM, Tom Lane wrote:
>> But you need to include to use INT_MAX ...
> I wonder why it has not given me any compilation error/warning. I have
> tried it on both Windows and CentOS, before sending the
Andrew Gierth wrote:
> But that leaves an obvious third issue: it's all very well to ignore the
> pre-upgrade (pre-9.3) mxid if it's older than the cutoff or it's in the
> future, but what if it's actually inside the currently valid range?
> Looking it up as though it were a valid mxid in that
On Thu, Jun 16, 2016 at 11:49 PM, Tom Lane wrote:
>
> Amit Kapila writes:
> > On Mon, Jun 13, 2016 at 10:36 PM, Tom Lane wrote:
> >> min_parallel_relation_size, or min_parallelizable_relation_size, or
> >> something like
On 2016/06/16 22:00, Robert Haas wrote:
On Thu, Jun 16, 2016 at 7:05 AM, Etsuro Fujita
wrote:
ISTM that a robuster solution to this is to push down the ft1-ft2-ft3 join
with the PHV by extending deparseExplicitTargetList() and/or something else
so that we can send
Robert Haas writes:
> On Thu, Jun 16, 2016 at 6:40 PM, Tom Lane wrote:
>> Meh. I think this is probably telling us that trying to repurpose Aggref
>> as the representation of a partial aggregate may have been mistaken.
> I don't mind if you feel the
On Thu, Jun 16, 2016 at 6:40 PM, Tom Lane wrote:
> Meh. I think this is probably telling us that trying to repurpose Aggref
> as the representation of a partial aggregate may have been mistaken.
> Or maybe just that fix_combine_agg_expr was a bad idea. It seems likely
> to
On 15 Jun 2016 2:59 pm, "David G. Johnston"
wrote:
>
> IIRC the plan is to have the machine version behave as if the middle
number is present and always zero. It would be (the?) one place that the
historical behavior remains visible but it is impossible to have a
Robert Haas writes:
> On Thu, Jun 16, 2016 at 4:31 PM, Tom Lane wrote:
>> However, before trying to fix any of that, I'd like to demand an
>> explanation as to why aggoutputtype exists at all. It seems incredibly
>> confusing to have both that and
On Thu, Jun 16, 2016 at 4:51 PM, Vik Fearing wrote:
> On 10/06/16 17:01, Robert Haas wrote:
>> Update pg_stat_statements extension for parallel query.
>
> I couldn't readily find a review for this patch, and I am unsatisfied
> with it. I think it's very strange that a 1.4
On Thu, Jun 16, 2016 at 4:31 PM, Tom Lane wrote:
> However, before trying to fix any of that, I'd like to demand an
> explanation as to why aggoutputtype exists at all. It seems incredibly
> confusing to have both that and aggtype, and certainly the code comments
> give no
On Thu, Jun 16, 2016 at 12:16 PM, Robert Haas wrote:
> On Thu, Jun 16, 2016 at 8:00 AM, Amit Kapila wrote:
>>> 1. The case originally reported by Thomas Munro still fails. To fix
>>> that, we probably need to apply scanjoin_target to each partial
Tom Lane-2 wrote
> The problem with an extension is: when we make a core change that breaks
> one of these views, which we will, how can you pg_upgrade a database
> with the extension installed? There's no provision for upgrading an
> extension concurrently with the core upgrade. Maybe there
I believe I've narrowed down the cause of this failure [1]:
regression=# explain SELECT min(col) FROM enumtest;
ERROR: type matched to anyenum is not an enum type: anyenum
The core reason for this seems to be that apply_partialaggref_adjustment
fails to consider the possibility that it's
On 16/06/16 20:31, Jim Nasby wrote:
> On 6/13/16 12:16 PM, Tom Lane wrote:
>> At the same time, I'm pretty suspicious of putting stuff like this in
>> core, because the expectations for cross-version compatibility go up
>> by orders of magnitude as soon as we do that.
>
> On a first go-round, I
2016-06-16 20:31 GMT+02:00 Jim Nasby :
> On 6/13/16 12:16 PM, Tom Lane wrote:
>
>> At the same time, I'm pretty suspicious of putting stuff like this in
>> core, because the expectations for cross-version compatibility go up
>> by orders of magnitude as soon as we do
On 2016-06-16 11:57:48 -0700, Andres Freund wrote:
> See
> https://www.postgresql.org/message-id/20160616183207.wygoktoplycdz...@alap3.anar
For posterity's sake, that was supposed to be
https://www.postgresql.org/message-id/20160616183207.wygoktoplycdz...@alap3.anarazel.de
--
Sent via
On 2016-06-16 13:53:01 -0500, Kevin Grittner wrote:
> On Thu, Jun 16, 2016 at 1:19 PM, Kevin Grittner wrote:
> > On Thu, Jun 16, 2016 at 11:54 AM, Andres Freund wrote:
> >> On 2016-06-16 12:43:34 -0400, Robert Haas wrote:
>
> >>> Maybe it would help if you
On Thu, Jun 16, 2016 at 1:32 PM, Andres Freund wrote:
> With old_snapshot_threshold=1 I indeed can reproduce the issue. I
> disabled autovacuum, to make the scheduling more predictable. But it
> should "work" just as well with autovacuum.
>
> S1: CREATE TABLE toasted(largecol
On Thu, Jun 16, 2016 at 1:19 PM, Kevin Grittner wrote:
> On Thu, Jun 16, 2016 at 11:54 AM, Andres Freund wrote:
>> On 2016-06-16 12:43:34 -0400, Robert Haas wrote:
>>> Maybe it would help if you lay out the whole sequence of events, like:
>>>
>>> S1: Does
Hi,
On 2016-06-16 13:19:13 -0500, Kevin Grittner wrote:
> On Thu, Jun 16, 2016 at 11:54 AM, Andres Freund wrote:
> > On 2016-06-16 12:43:34 -0400, Robert Haas wrote:
>
> >> The root of my confusion is: if we prune a tuple, we'll bump the page
> >> LSN, so any session that is
On 6/13/16 12:16 PM, Tom Lane wrote:
At the same time, I'm pretty suspicious of putting stuff like this in
core, because the expectations for cross-version compatibility go up
by orders of magnitude as soon as we do that.
On a first go-round, I don't think we should add entire views, but
On 2016-06-16 13:16:35 -0500, Kevin Grittner wrote:
> On Thu, Jun 16, 2016 at 1:01 PM, Andres Freund wrote:
>
> > The relevant part is the HeapTupleSatisfiesMVCC check, we're using
> > SatisfiesToast for toast lookups.
> >
> > FWIW, I just tried to reproduce this with
Amit Kapila writes:
> On Mon, Jun 13, 2016 at 10:36 PM, Tom Lane wrote:
>> min_parallel_relation_size, or min_parallelizable_relation_size, or
>> something like that?
> You are right that such a variable will make it simpler to write tests for
>
On Thu, Jun 16, 2016 at 11:54 AM, Andres Freund wrote:
> On 2016-06-16 12:43:34 -0400, Robert Haas wrote:
>> The root of my confusion is: if we prune a tuple, we'll bump the page
>> LSN, so any session that is still referencing that tuple will error
>> out as soon as it
On Thu, Jun 16, 2016 at 1:01 PM, Andres Freund wrote:
> The relevant part is the HeapTupleSatisfiesMVCC check, we're using
> SatisfiesToast for toast lookups.
>
> FWIW, I just tried to reproduce this with old_snapshot_threshold = 0 -
> but ran into the problem that I couldn't
As of HEAD you can exercise quite a lot of parallel query behavior
by running the regression tests with these settings applied:
force_parallel_mode = regress
max_parallel_workers_per_gather = 2-- this is default at the moment
min_parallel_relation_size = 0
parallel_setup_cost = 0
On 2016-06-16 13:44:34 -0400, Robert Haas wrote:
> On Thu, Jun 16, 2016 at 12:54 PM, Andres Freund wrote:
> >> The root of my confusion is: if we prune a tuple, we'll bump the page
> >> LSN, so any session that is still referencing that tuple will error
> >> out as soon as it
Hi,
2016-06-16 9:48 GMT-03:00 Michael Paquier :
> On Thu, Jun 16, 2016 at 8:37 PM, Martín Marqués
> wrote:
>> El 16/06/16 a las 00:08, Michael Paquier escribió:
>>> On Wed, Jun 15, 2016 at 7:19 PM, Martín Marqués
>>>
On Thu, Jun 16, 2016 at 12:54 PM, Andres Freund wrote:
>> The root of my confusion is: if we prune a tuple, we'll bump the page
>> LSN, so any session that is still referencing that tuple will error
>> out as soon as it touches the page on which that tuple used to exist.
>
>
On Thu, Jun 16, 2016 at 12:50 PM, Tom Lane wrote:
> Amit Kapila writes:
>> On Thu, Jun 16, 2016 at 9:26 AM, Robert Haas wrote:
>>> 3. In https://www.postgresql.org/message-id/25521.1465837...@sss.pgh.pa.us
>>> Tom proposed
On 2016-06-16 12:43:34 -0400, Robert Haas wrote:
> On Thu, Jun 16, 2016 at 12:14 PM, Andres Freund wrote:
> >> > The issue isn't there without the feature, because we (should) never
> >> > access a tuple/detoast a column when it's invisible enough for the
> >> > corresponding
Amit Kapila writes:
> On Thu, Jun 16, 2016 at 9:26 AM, Robert Haas wrote:
>> 3. In https://www.postgresql.org/message-id/25521.1465837...@sss.pgh.pa.us
>> Tom proposed adding a GUC to set the minimum relation size required
>> for consideration of
On Thu, Jun 16, 2016 at 12:14 PM, Andres Freund wrote:
>> > The issue isn't there without the feature, because we (should) never
>> > access a tuple/detoast a column when it's invisible enough for the
>> > corresponding toast tuple to be vacuumed away. But with
>> >
Teodor Sigaev writes:
>> So I think there's a reasonable case for decreeing that should only
>> match lexemes *exactly* N apart. If we did that, we would no longer have
>> the misbehavior that Jean-Pierre is complaining about, and we'd not need
>> to argue about whether <0>
On Thu, Jun 16, 2016 at 8:00 AM, Amit Kapila wrote:
>> 1. The case originally reported by Thomas Munro still fails. To fix
>> that, we probably need to apply scanjoin_target to each partial path.
>> But we can only do that if it's parallel-safe. It seems like what we
>>
On 2016-06-16 12:02:39 -0400, Robert Haas wrote:
> On Thu, Jun 16, 2016 at 11:37 AM, Andres Freund wrote:
> >> If that were really true, why would we not have the problem in
> >> current production versions that the toast table could be vacuumed
> >> before the heap, leading
On Thu, Jun 16, 2016 at 11:37 AM, Andres Freund wrote:
>> If that were really true, why would we not have the problem in
>> current production versions that the toast table could be vacuumed
>> before the heap, leading to exactly the issue you are talking
>> about?
>
> The
Michael Paquier writes:
> On Wed, Jun 15, 2016 at 10:34 PM, Tom Lane wrote:
>> My feeling is that we'd keep
>> the pg_attribute.attnotnull field and continue to drive actual enforcement
>> off that, but it would just reflect a summary of the
On 2016-06-16 09:50:09 -0500, Kevin Grittner wrote:
> On Wed, Jun 15, 2016 at 8:57 PM, Andres Freund wrote:
>
> > The simplest version of the scenario I'm concerned about is that a tuple
> > in a tuple is *not* vacuumed, even though it's elegible to be removed
> > due to STO.
On Thu, Jun 16, 2016 at 10:03 PM, Robert Haas wrote:
> On Thu, Jun 16, 2016 at 2:33 AM, Masahiko Sawada
> wrote:
>> Option name DISABLE_PAGE_SKIPPING is good to me.
>> I'm still working on this, but here are some review comments.
>>
>> ---
>>
On Wed, Jun 15, 2016 at 8:57 PM, Andres Freund wrote:
> The simplest version of the scenario I'm concerned about is that a tuple
> in a tuple is *not* vacuumed, even though it's elegible to be removed
> due to STO. If that tuple has toasted columns, it could be the that the
>
On Thu, Jun 16, 2016 at 4:35 PM, Euler Taveira wrote:
> On 16-06-2016 09:05, Magnus Hagander wrote:
> > Shouldn't pg_upgrade turn off vacuum cost delay when it vacuums the new
> > cluster? Not talking about the post-analyze script, but when it runs
> > vacuumdb to analyze
On 16-06-2016 09:05, Magnus Hagander wrote:
> Shouldn't pg_upgrade turn off vacuum cost delay when it vacuums the new
> cluster? Not talking about the post-analyze script, but when it runs
> vacuumdb to analyze and freeze before loading the new schema, in
> prepare_new_cluster()? Those run during
On 01-06-2016 20:52, Andreas Karlsson wrote:
> I think at least three of the four aggregate functions are little used,
> so I do not think many users would be affected. And only min(citext) and
> max(citext) can make use of the parallel aggregation.
>
> The functions are:
>
> min(citext)
>
On 06/14/2016 09:55 PM, Robert Haas wrote:
On Tue, Jun 14, 2016 at 1:51 PM, Robert Haas wrote:
On Tue, Jun 14, 2016 at 6:37 AM, Andreas Karlsson wrote:
I have rebased all my patches on the current master now (and skipped the
extensions I previously
El 16/06/16 a las 09:48, Michael Paquier escribió:
> On Thu, Jun 16, 2016 at 8:37 PM, Martín Marqués
> wrote:
>
>> This problem came up due to a difference between pg_dump on 9.1.12 and
>> 9.1.22 (I believe it was due to a patch on pg_dump that excluded the
>> dependent
On Thu, Jun 16, 2016 at 2:33 AM, Masahiko Sawada wrote:
> Option name DISABLE_PAGE_SKIPPING is good to me.
> I'm still working on this, but here are some review comments.
>
> ---
> +CREATE FUNCTION pg_truncate_visibility_map(regclass)
> +RETURNS void
> +AS
On Thu, Jun 16, 2016 at 7:05 AM, Etsuro Fujita
wrote:
> ISTM that a robuster solution to this is to push down the ft1-ft2-ft3 join
> with the PHV by extending deparseExplicitTargetList() and/or something else
> so that we can send the remote query like:
>
> select
On Thu, Jun 16, 2016 at 8:37 PM, Martín Marqués wrote:
> El 16/06/16 a las 00:08, Michael Paquier escribió:
>> On Wed, Jun 15, 2016 at 7:19 PM, Martín Marqués
>> wrote:
>>>
>>> How would the recovery process work? We expect the schema to be there
Shouldn't pg_upgrade turn off vacuum cost delay when it vacuums the new
cluster? Not talking about the post-analyze script, but when it runs
vacuumdb to analyze and freeze before loading the new schema, in
prepare_new_cluster()? Those run during downtime, so it seems like you'd
want those to run
On Thu, Jun 16, 2016 at 9:26 AM, Robert Haas wrote:
>
> On Tue, Jun 14, 2016 at 12:18 PM, Tom Lane wrote:
> > Amit Kapila writes:
> >> On Mon, Jun 13, 2016 at 9:37 PM, Tom Lane wrote:
> >>> ... I got a
El 16/06/16 a las 00:08, Michael Paquier escribió:
> On Wed, Jun 15, 2016 at 7:19 PM, Martín Marqués
> wrote:
>>
>> How would the recovery process work? We expect the schema to be there
>> when restoring the tables?
>
> pg_dump creates the schema first via the CREATE
On 2016/06/15 9:13, Amit Langote wrote:
On 2016/06/15 0:50, Robert Haas wrote:
On Tue, Jun 14, 2016 at 4:06 AM, Amit Langote wrote:
Attached new version of the patch with following changes:
OK, I committed this version with some cosmetic changes.
Thank you all for working on this!
While
> "Alvaro" == Alvaro Herrera writes:
Alvaro> I think that was a good choice in general so that
Alvaro> possibly-data-eating bugs could be reported, but there's a
Alvaro> problem in the specific case of tuples carried over by
Alvaro> pg_upgrade whose Multixact is
On Thu, Jun 16, 2016 at 10:39 AM, Jakob Egger wrote:
> Hi!
>
> I've looked at the way libpq handles TLS certificates and plaintext
> fallback, and I am somewhat surprised.
>
> The default ssmode is prefer. According to the documentation, this will
> make libpq use an SSL
Hi!
I've looked at the way libpq handles TLS certificates and plaintext fallback,
and I am somewhat surprised.
The default ssmode is prefer. According to the documentation, this will make
libpq use an SSL connection if possible, but will use a plain text connection
as a fallback. The
On Tue, May 10, 2016 at 5:39 PM, Amit Kapila
wrote:
>
>
> Incomplete Splits
> --
> Incomplete splits can be completed either by vacuum or insert as both
> needs exclusive lock on bucket. If vacuum finds split-in-progress flag on
> a bucket then it
On Thu, Jun 16, 2016 at 12:06 PM, Robert Haas wrote:
> [ Changing subject line in the hopes of keeping separate issues in
> separate threads. ]
>
> On Mon, Jun 6, 2016 at 11:00 AM, Tom Lane wrote:
>> Robert Haas writes:
>>> I'm
63 matches
Mail list logo