On Wed, Jun 14, 2023 at 8:15 PM Ron wrote:
> On 6/14/23 13:02, Marc Millas wrote:
>
>
> On Wed, Jun 14, 2023 at 7:27 PM David G. Johnston <
> david.g.johns...@gmail.com> wrote:
>
>> On Wed, Jun 14, 2023 at 9:42 AM Marc Millas
>> wrote:
>>
>>>
On Wed, Jun 14, 2023 at 8:23 PM Marc Millas wrote:
>
>
>
>
> On Wed, Jun 14, 2023 at 8:15 PM Ron wrote:
>
>> On 6/14/23 13:02, Marc Millas wrote:
>>
>>
>> On Wed, Jun 14, 2023 at 7:27 PM David G. Johnston <
>> david.g.johns...@gmail.com> wro
been vacuum analyzed.
thanks
Marc MILLAS
Senior Architect
+33607850334
www.mokadb.com
On Tue, Jun 20, 2023 at 10:14 PM David Rowley wrote:
> On Wed, 21 Jun 2023 at 07:42, Marc Millas wrote:
> > But if I do the same with clause one OR clause 2, I have to kill the
> request after an hour, seeing the filesystem showing more than 140 Mb of
> increased usage.
&
On Tue, Jun 20, 2023 at 11:19 PM David Rowley wrote:
> On Wed, 21 Jun 2023 at 08:34, Marc Millas wrote:
> >
> > On Tue, Jun 20, 2023 at 10:14 PM David Rowley
> wrote:
> >>
> >> On Wed, 21 Jun 2023 at 07:42, Marc Millas
> wrote:
> >> > But i
On Wed, Jun 21, 2023 at 12:43 PM Tomas Vondra
wrote:
> On 6/21/23 00:26, Marc Millas wrote:
> >
> >
> > On Tue, Jun 20, 2023 at 11:19 PM David Rowley > <mailto:dgrowle...@gmail.com>> wrote:
> >
> > On Wed, 21 Jun 2023 at 08:34, Marc Millas
Marc MILLAS
On Wed, Jun 21, 2023 at 12:43 PM Tomas Vondra
wrote:
> On 6/21/23 00:26, Marc Millas wrote:
> >
> >
> > On Tue, Jun 20, 2023 at 11:19 PM David Rowley > <mailto:dgrowle...@gmail.com>> wrote:
> >
> > On Wed, 21 Jun 2023
syntax as
soon as there is some data volumes.
So, now I am writing some kind of recipe book for the users of that DB.
What should I write ?
Marc MILLAS
Senior Architect
+33607850334
www.mokadb.com
On Sun, Jun 25, 2023 at 9:35 PM David Rowley wrote:
> On Mon, 26 Jun 2023 at 03:02, Marc Millas wrote:
> > When I ask this list, David Rowley suggest to rewrite the SQL, replacing
> the OR by a union.
> >
> > Fine, this do work, even if a bit complex as the original SQL
iting raw
SQL queries, and I cannot let them write queries that are going to go on
for ages, and eventually crash over temp_file_limit after hours every now
and then.
So, my understanding of the above is that I must inform the users NOT to
use OR clauses into joins.
which maybe a pb by itself.
regards
Marc
> regards, tom lane
>
Marc MILLAS
ng 3 merge join.
And I have one more question: the explain analyze plan shows that Postgres
decided to do external sorts using around 2 GB of disk space.
I did a set work_mem to '4GB' to try to have those sorts in memory. No
effect. How can I tell the planner to do those sort in memory ?? th
On Mon, Jun 26, 2023 at 4:05 PM Ron wrote:
> On 6/26/23 07:22, Marc Millas wrote:
>
>
>
> On Mon, Jun 26, 2023 at 5:47 AM Avin Kavish wrote:
>
>> Sounds like the problem you are having is, the server is running out of
>> temporary resources for the operation t
On Tue, Jun 27, 2023 at 8:12 PM Kirk Wolak wrote:
> On Wed, Jun 21, 2023 at 12:10 PM Marc Millas
> wrote:
>
>> Marc MILLAS
>> On Wed, Jun 21, 2023 at 12:43 PM Tomas Vondra <
>> tomas.von...@enterprisedb.com> wrote:
>>
>>> On 6/21/23 00:26, Mar
Hi,
Postgres 14.2 (for one more month)
The explain plan and request is here:
https://explain.depesz.com/s/Opk0
The big table is split in around 130 partitions, one by month.
the ladate column is the partition key, and it does have 1 value for each
partition.
there is a
Marc MILLAS
Senior
, one column at a
time (there are 107 columns...)
I dont understand why the planner use a filter on the date, as its already
within the partition structure.
Thanks,
Marc MILLAS
On Wed, Jun 28, 2023 at 5:22 PM Marc Millas wrote:
> Hi,
>
> Postgres 14.2 (for one more month)
> The
On Wed, Jun 28, 2023 at 6:48 PM Laurenz Albe
wrote:
> On Wed, 2023-06-28 at 17:29 +0200, Marc Millas wrote:
> > https://explain.depesz.com/s/Opk0
> >
> > The big table (10 billions raws) is split in around 130 partitions, one
> by month.
> > the ladate column is
Le mer. 28 juin 2023 à 22:46, Laurenz Albe a
écrit :
> On Wed, 2023-06-28 at 19:25 +0200, Marc Millas wrote:
> > Hi Laurenz, as said, in each partition there is only one value for
> ladate.
>
> The planner doesn't seem to take that into account.
>
Indeed. I did check v
.
If I run vacuum analyze ma_table on my big partitioned table (130+
partitions) it does work for quite a long time.
Can someone clarify ?
thanks
Marc MILLAS
Senior Architect
+33607850334
www.mokadb.com
vue.
and, obviously if you partition by id range, no pk pb.
the difference is the SQL plan to access the data, as to prune partitions,
Postgres need to have the partition key in the where clause.
obviously, if your code do access the data thru id... then partition by id.
>
> Marc MILLAS
>
>
try to set work_mem to 100 MB and check what's happening.
Also check the logs, postgres tells his life there...
Marc MILLAS
Senior Architect
+33607850334
www.mokadb.com
On Mon, Aug 7, 2023 at 3:36 PM KK CHN wrote:
> List ,
>
> *Description:*
>
> Maintaining a DB Server
tgres 12:
https://www.postgresql.org/docs/12/runtime-config-autovacuum.html
Marc MILLAS
Senior Architect
+33607850334
www.mokadb.com
On Tue, Aug 8, 2023 at 1:59 PM Marc Millas wrote:
> Hello,
> in the postgresql.conf joined, 2 things (at least) look strange:
> 1) the values for b
omplete data model and the exact requests
can be considered mandatory, but when I change the request I get the very
same behaviour...
Marc MILLAS
Senior Architect
+33607850334
www.mokadb.com
s persistent volumes.
>
> What is the pipe connecting the database server to the disks? If it's
> NAS, well that would explain it.
>
> HTH,
> Adam
>
>
>
> On Thu, Aug 10, 2023 at 2:37 PM Marc Millas
> wrote:
>
>> Hi,
>>
>> I have a 15 TB db
Hi David,
that's exactly my question.
does the analyze buffers data, generated when track_io_timing is on, keep
track of multiple reloads of the same data while executing one operation ?
I ll do the test asap and report the results.
Marc MILLAS
Senior Architect
+33607850334
www.mokad
ifferently, what could be advantages/disadvantages of such a
setup ?
thanks
Marc MILLAS
Senior Architect
+33607850334
www.mokadb.com
s when the line is executed or at commit time.
thanks.
Marc MILLAS
Senior Architect
+33607850334
www.mokadb.com
Thanks !
Marc MILLAS
Senior Architect
+33607850334
www.mokadb.com
On Wed, Aug 30, 2023 at 8:44 PM David G. Johnston <
david.g.johns...@gmail.com> wrote:
> On Wednesday, August 30, 2023, Marc Millas wrote:
>
>> Hi,
>> the doc v15 states: " pg_event_trigger_ddl_c
Thanks !
Marc MILLAS
Senior Architect
+33607850334
www.mokadb.com
On Wed, Aug 30, 2023 at 6:18 PM Erik Wienhold wrote:
> > On 30/08/2023 17:12 CEST Marc Millas wrote:
> >
> > the doc v15 states: " pg_event_trigger_ddl_commands returns a list of DDL
> > commands
Marc MILLAS
Senior Architect
+33607850334
www.mokadb.com
Testing pg_restore with different --jobs= values will be easier. pg_dump
> is what's going to be reading from a constantly varying system.
>
> Hello,
>
each time I do a replatforming of this kind, with DB up to 2 TB,
by just
getting rid of obsolete partitions which is quite speedy.( no delete, no
vacuum, no index updates, ...)
Marc
> Keep in mind that SELECT queries being used on the partition must use the
> partitioning KEY in the WHERE clause of the query or performance will
> suffer.
>
>
from scratch ?!?
what s your advice ?
thanks
Marc MILLAS
Senior Architect
+33607850334
www.mokadb.com
to me, collate 'C' ask for the raw ascii order which put caps before
because the hexa coding is lower
did you try any thing else (POSIX is same as 'C')
Marc MILLAS
Senior Architect
+33607850334
www.mokadb.com
On Thu, Mar 18, 2021 at 4:18 PM basti wrote:
> This doe
or, maybe, just nothing ?
Marc MILLAS
Senior Architect
+33607850334
www.mokadb.com
On Thu, Mar 18, 2021 at 4:44 PM Ron wrote:
> Exactly. "C" collation is the opposite of what he wants.
>
> OP needs something like "de_DE".
>
> On 3/18/21 10:36 AM, Marc
ll my goal)
thanks,
Marc MILLAS
Senior Architect
+33607850334
www.mokadb.com
TE Scan on numb (cost=0.00..0.10 rows=5 width=12) (actual
time=0.004..0.005 rows=5 loops=1)
Planning Time: 0.070 ms
Execution Time: 0.313 ms
(11 lignes)
Marc MILLAS
Senior Architect
+33607850334
www.mokadb.com
On Wed, Mar 24, 2021 at 3:22 AM David G. Johnston <
david.g.johns...@gmail
Got it :-)
thanks !
Marc MILLAS
Senior Architect
+33607850334
www.mokadb.com
On Wed, Mar 24, 2021 at 4:21 AM David G. Johnston <
david.g.johns...@gmail.com> wrote:
> On Tuesday, March 23, 2021, Marc Millas wrote:
>
>> Hi,
>>
>> I cannot agree.
>> I did an
2) I may have to do a POC with Sap HANA on one side and postgres
on the other
on machines with a lot of RAM.
so...
can someone shade some light ?
thanks,
Marc MILLAS
Senior Architect
+33607850334
www.mokadb.com
...)
Marc MILLAS
Senior Architect
+33607850334
www.mokadb.com
On Mon, May 10, 2021 at 11:13 PM Jonathan Chen wrote:
> Hi,
>
> I am running Postgresql 13 as a backend for Odoo, and I believe I have
> discovered a regression with the query engine.
>
> This (simplified) query ge
Hi,
within a function, I want to create another function.
no pb.
but if I write:
declare bidule text;
begin
bidule:='myfunc';
create function bidule() ...
it does create a function named bidule and not myfunc.
so I am obviously missing something too obvious.
can someone help ?
than
good reading, thanks
Marc MILLAS
Senior Architect
+33607850334
www.mokadb.com
On Thu, Jun 3, 2021 at 9:21 PM Tom Lane wrote:
> Adrian Klaver writes:
> > On 6/3/21 12:01 PM, Bruce Momjian wrote:
> >> On Thu, Jun 3, 2021 at 08:58:03PM +0200, Marc Millas wrote:
> >>
automatize the building of the
whole thing, taking list of columns and constraints (for PK) directly from
pg_catalog.
Now it works :-)
but.. why do you ask that question ? is there any king of hidden wolf we
didnt see ?
Marc MILLAS
Senior Architect
+33607850334
www.mokadb.com
On Thu, Jun 3, 2021
he
PK.*
*OK, we may try to use a function to get the bounding box around the
geometry objects and use the result into a btree index*
*Any idea (I mean: another idea !) to tackle this ?*
*Or any critic on the "solution" ??*
*thanks,*
Marc MILLAS
Senior Architect
+33607850334
www.mokadb.com
I take note of this.
thanks
Marc MILLAS
Senior Architect
+33607850334
www.mokadb.com
On Thu, Jun 3, 2021 at 10:23 PM David G. Johnston <
david.g.johns...@gmail.com> wrote:
> On Thu, Jun 3, 2021 at 1:02 PM Marc Millas wrote:
>
>> about knowing if I should...
>> We
Hi Paul,
obviously the bounding box is not a perfect choice. we also think of one of
the "centers" point, but quite similar non uniqueness.
so, if no "tech" solution, we continue to work with the business to try to
find an appropriate PK
thanks
Marc MILLAS
Senior Ar
no pb: I am french, so quite skilled on that topic :-)
there is only 50 bottles of various malt on the presentoir close to my desk
so I must stay reasonnable :-)
Marc MILLAS
Senior Architect
+33607850334
www.mokadb.com
On Thu, Jun 3, 2021 at 11:17 PM Adrian Klaver
wrote:
> On 6/3/21 1
text in a btree
index and NOT a geom column which wkt is 10 000 bytes long.
there, sure, is a reason.
please, let me know :-)
Marc MILLAS
Senior Architect
+33607850334
www.mokadb.com
On Tue, Jun 8, 2021 at 12:09 AM Peter J. Holzer wrote:
> On 2021-06-07 10:20:22 -0700, David G. Johnston
entry cannot exceed approximately one-third of
a page (after TOAST compression, if applicable)."
but as this looks quite hard coded, it means that for long utf8 things the
data length is not so long before hitting the limit.
is there any plan to adress this ?
kind of index toast ?? :-)
tha
a PK, is there any
way to make something useful and not this unuseful "thing" ?
Marc MILLAS
Senior Architect
+33607850334
www.mokadb.com
On Tue, Jun 8, 2021 at 10:51 PM Thomas Kellerer wrote:
> Marc Millas schrieb am 03.06.2021 um 22:51:
> > on a table we need a prima
reating an index that will never be of any use, just for technical reasons.
People in this list have been quite clear that the btree limitation will
NOT change. Which ends the point.
regards
Marc MILLAS
Senior Architect
+33607850334
www.mokadb.com
On Thu, Jun 10, 2021 at 10:33 PM Alban Hertr
s on that difference.
best regards,
Marc Mamin
From: Joe Conway [mailto:m...@joeconway.com]
>Sent: Donnerstag, 24. Juni 2021 14:47
>To: Marc Mamin ; pgsql-general
>
>Subject: Re: removing "serial" from table definitions.
>
>On 6/24/21 8:33 AM, Marc Mamin wrote:
>> Hi,
>>
>> Is there a way to c
t;It's 'nextval('admin.db_jobs_history_id_seq'::regclass)' on source and
>'nextval('db_jobs_history_id_seq'::regclass)' on target.
I've probably found the origin of our problem:
https://www.postgresql.org/docs/9.3/release-8-1.html
=> Add proper dependencies for arguments of sequence functions (Tom)
But I won't be able to check that in the next few days..
best regards,
marc Mamin
Marc Mamin writes:
> > Yes, I undersand that serial is just a hint at table creation time, but is
> > there a place in catalog where we can see if the table was created using
> > 'serial' ?
>
> No. Where the docs say "these are equivalent", they mea
(' ');
insert into test(ble) values('Marc');
insert into test(ble) values(' Marc');
insert into test(ble) values('marc');
insert into test(ble) values(' marc');
insert into test(ble) values('bobo');
insert into test(ble) values(' bobo
Re-reading my post, I see that even the élise is not sorted correctly on
w10...
Marc MILLAS
Senior Architect
+33607850334
www.mokadb.com
On Tue, Jul 27, 2021 at 7:07 PM Marc Millas wrote:
> Hi,
>
> context:
> one postgres 12 on centos 7
> one postgres 12 on windows 10
> bot
Seems to me that's not an UNION, but a JOIN:
SELECT o.old_count - n.new_count, o.old_sup, n.new_sup
FROM (
SELECT new_sup, COUNT(new_sup)
FROM public."Data"
GROUP BY new_sup
) n
JOIN (
SELECT old_sup, COUNT(old_sup)
FROM public."Data"
GROUP BY old_sup
) o ON o.old_su
upgraded slave (step 2 ) ??
thanks,
Marc MILLAS
Senior Architect
+33607850334
www.mokadb.com
it does work well for logical replication.
still its sad that it doesnt work for physical replication as, then, how to
ensure continuous availability for generic apps ??
Marc MILLAS
Senior Architect
+33607850334
www.mokadb.com
On Wed, Oct 6, 2021 at 5:03 PM Guillaume Lelarge
wrote:
> Le
Hi,
Just one question
Isn't it possible to write a function that return an oid?
So.. One can write a function which create a temp table, feed it according
to parameters and then return the oid of the created temp table.
No??
Le ven. 19 nov. 2021 à 23:48, David G. Johnston
a écrit :
> On Fri, No
Hi,
on the postgres repo, on centos 7 x86, I can found the old postgres 9.5,
and a postgis 2.4
I need to found a postgres 9.5 with a postgis 2.5.1
where can I found it ?
(I know... its obsolete...)
thanks
Marc MILLAS
Senior Architect
+33607850334
www.mokadb.com
partition at a time)
great on perf, too IF the partition key is in the where clause. if not,
postgres will need to aggregate and sort the results gathered for each
partition.
Could still be efficient if it allows to parallel execution.
my 2 cents
Marc MILLAS
Senior Architect
+33607850334
uch a lock,
thanks,
Marc Mamin
here an example of a such a deployment/migration script, all of these scripts
are applied sequentially in separate transactions:
===
SET client_min_messages=error;
CREATE OR REPLACE FUNCTION block_item_cancel()
RETURNS TRIGGER AS
$BOD
hment, that the lock exists without any other apparent
conflicting session.
>Do you need to drop/create the trigger or a CREATE OR REPLACE function
would suffice?
There are different use cases. Sometimes I only need to drop a trigger or
modify its definition (not the function)
&
>
>-Original Message-
>From: Achilleas Mantzios
>Sent: Freitag, 10. Dezember 2021 11:36
>To: pgsql-general@lists.postgresql.org
>Subject: Re: What is the best way to redefine a trigger? (lock issue)
>
>On 10/12/21 12:2
example, within a join, some kind of lock may be
acquired so that the dataset concerned is not changed during execution ?
(for example a delete then autovacuum ??)
thanks
Marc MILLAS
Senior Architect
+33607850334
www.mokadb.com
--synchronous option.
My question: as the synchronous option is supposed to make pg_receivewal
write transaction immediately in the wal files, is there a way to ask the
standby to apply them on the fly ie. without waiting a wal file change ?
thanks,
Marc MILLAS
Senior Architect
+33607850334
... during business hours, obviously !
--but a bunch of read request are sent to the secondary, some of them a bit
big in terms of multiple join, etc..
hence my original question.
So, if its not a locking scheme, what can block hundreds of transaction to
get to the secondary ???
Marc MILLAS
Senior
Excellent !
thanks a lot, it was exactly what I was looking for.
Marc MILLAS
Senior Architect
+33607850334
www.mokadb.com
On Tue, Dec 14, 2021 at 6:14 PM Laurenz Albe
wrote:
> On Tue, 2021-12-14 at 13:38 +0100, Marc Millas wrote:
> > but... when you do setup a streaming replication,
, even the disk option, you
"may" encounter phase pb ie. id existing in one db and not in the other.
but this depends on how you maintain the images.
my 2 cents
Marc MILLAS
Senior Architect
+33607850334
www.mokadb.com
On Wed, Dec 15, 2021 at 9:12 PM Estevan Rech wrote:
> Adrian,
&g
onf file in the old
slave, and restart it.
maybe adding first:
recovery_target_timeline latest in the recovery.conf file
Marc MILLAS
Senior Architect
+33607850334
www.mokadb.com
On Mon, Feb 14, 2022 at 8:59 PM Glen Eustace
wrote:
>
> On 15/02/22 8:39 am, Alan Hodgson wrote:
&g
Hello,
On Sun, Oct 13, 2024 at 08:16:04AM +1300, Thomas Munro wrote:
> > template1=> SELECT COUNT(*) FROM pg_class a, pg_class b, pg_class c;
> >
> > I see only one 100% CPU PostgreSQL process.
>
> If you set set min_parallel_table_scan_size = 0 then it uses
Without it, it uses one CPU and takes
Hello,
on a machine where starting two processes:
perl -e 'while (1) { ; }'
I see two processed at 100% CPU, which is expected (with top).
Now, if I do:
template1=> SELECT COUNT(*) FROM pg_class a, pg_class b, pg_class c;
I see only one 100% CPU PostgreSQL process.
I read that while Postgre
l, as said, I will try to pipe data through the
network,
but this imply that data will be put exactly as psql, on the new machine,
will be able to 'ingest' it.
easy for most things, somewhat more boring for blobs...
Marc MILLAS
Senior Architect
+33607850334
www.mokadb.com
On Tue, F
on a redhat 6.5 ?
If nothing, I will try to pipe the data directly through the network
connection, but this generates other problems.
thanks for your help
Marc MILLAS
Senior Architect
+33607850334
👍👍👍
thanks
Marc MILLAS
Senior Architect
+33607850334
On Tue, Feb 4, 2025 at 7:04 PM Tim Gerber wrote:
> Appears PG 12 is available for RHEL 6 via the following repository:
>
> https://download.postgresql.org/pub/repos/yum/12/redhat/
> <https://download.postgresql.org/p
Hello,
on a customer postgres db, using psql, if I post \ to initiate, for
exemple \x
that psql immediately exits.
somewhat boring as it blocks all psql meta commands.
Is there a way to get the extended display without posting a \ ?
thanks,
Marc MILLAS
Senior Architect
+33607850334
=52655.694..62533.811 rows=346 loops=1)
Buffers: shared hit=10068265 read=396705 dirtied=1858 written=218, temp
read=429687 written=115187
I/O Timings: read=79368.246 write=11.486
So, the total execution time is 52655 ms ok
and the total time for i/o is...79368 ms
how ???
thanks
Marc MILLAS
Senior
Thanks !
Marc MILLAS
Senior Architect
+33607850334
www.mokadb.com
On Thu, Apr 24, 2025 at 3:41 PM David Rowley wrote:
> On Fri, 25 Apr 2025 at 01:32, Marc Millas wrote:
> > Nested Loop Left Join (cost=941400.77..966327.57 rows=3 width=653)
> (actual time=52655.694..62533.811 ro
times I rerun the explain (analyze, buffers) of the same request, each
time, the explain shows the same volume of reads. again and again.
If I check with pg_buffercache, the set of objects stay the same, WITHOUT
the objects of my request, just like if those objects where sticky.
any idea ?
thanks
Hello,
One of my customers tells me he did a replication with the master on a
debian 10
and the slave on a debian 12, both with a pg13.20, one on a pgdg110+1, the
other a pgdg120.1
is this a good idea ??? (I don't think so...)
thanks,
Marc MILLAS
Senior Architect
+33607850334
Sorry,
'someone' launches some kind of batches without telling.
Marc MILLAS
Senior Architect
+33607850334
www.mokadb.com
On Fri, Apr 25, 2025 at 3:42 PM Marc Millas wrote:
> hello,
>
> got something strange to me:
> Same db ie. same data, around 1.2TB,one on pg13, on
I didnt know this.
thanks,
Marc MILLAS
Senior Architect
+33607850334
www.mokadb.com
On Sat, Apr 26, 2025 at 12:46 AM Laurenz Albe
wrote:
> On Fri, 2025-04-25 at 15:42 +0200, Marc Millas wrote:
> > got something strange to me:
> > Same db ie. same data, around 1.2TB,one on pg
type numeric "\."
then the 2 lines stating the copy command that was executed.
what am I missing ?
thanks
Marc MILLAS
Senior Architect
+33607850334
www.mokadb.com
Greg... 👍
works fine now.
thanks !
Marc MILLAS
Senior Architect
+33607850334
www.mokadb.com
On Wed, Mar 5, 2025 at 4:53 PM Greg Sabino Mullane
wrote:
> On Wed, Mar 5, 2025 at 10:22 AM Marc Millas
> wrote:
>
>> Then the flow contains a single line: \. to my understan
uch a kind of
use-case using postgres?
Thanks in advance for your response!
Best regards,
Marc-Antoine Nüssli
,"value":2},
{"time":39815.0,"value":3}]}');
Is there a way to do this ? or is it in development?
Thanks!
Marc
alternative but it does not accept toast
tables.
It can be run on pg_largeobject which is very similar to the toast table
structures.
Why pgstattuple_approx() cannot be run on toast tables?
Regards,
Jean-Marc Lessard
Administrateur de base de donn?es / Database Administrator
[cid:Grey_159x30_4bbe7521
the next vacuum.
I guess that the visibility map is updated only at the end of each pass.
My best regards
Thank you
Jean-Marc Lessard
Administrateur de base de données / Database Administrator
[cid:image001.jpg@01D69D5F.64DB0EA0]
Ultra Electronics Forensic Technology Inc.
T +1 514 489 4247 x4164
Thank you Michael for your suggestion,
I will reduce vacuum_cost_delay and increasing vacuum_cost_limit to speedup
vacuum.
But the I/O system is very solicitated on PROD system and I do not want to
impact end user performance.
If aggressive cost_delay and vacuum_cost_limit slow down significative
curity enabled for table "ibisl1.signatures"
pg_dump: reading policies for table "ibisl1.signatures"
pg_dump: reading large objects
pg_dump: reading dependency data
pg_dump: saving encoding = UTF8
pg_dump: saving standard_conforming_strings = on
out of memory
Jean-Marc Lessard
le?
So LOB will be more suitable for frequently updated table where the binary data
rarely change.
Jean-Marc Lessard
Administrateur de base de données / Database Administrator
Ultra Electronics Forensic Technology Inc.
T +1 514 489 4247 x4164
www.ultra-forensictechnology.com<http://www.ultra-forensictechnology.com>
f-line values change.
Does it means, no incurs cost to generate the out of line toast, but that a
copy of the bytea is still made for the new line?
Jean-Marc Lessard
Administrateur de base de données / Database Administrator
Ultra Electronics Forensic Technology Inc.
T +1 514 489 4247 x4164
www.ult
s it means, no incurs cost to generate the out of line toast, but that a
> copy of the bytea is still made for the new line?
I bench mark it as follow:
UPDATE table SET mod_tim=mod_tim;
The relpages of the table doubled, but the relpages of the toast table did not
changed.
Jean-Marc Lessard
A
sion and only 3 pieces can fit in a block.
Can we change the LOBLKSIZE to 2000 bytes to fit 4 rows in a block as follow?
#define LOBLKSIZE 2000
Must the LOBLKSIZE be a power-of-2?
Is there any optimization expecting a power-of-2 value?
Thank you.
Jean-Marc Lessard
Administrateur de
on x86_64-apple-darwin16.7.0, compiled by Apple
LLVM version 8.1.0 (clang-802.0.42), 64-bit"
Best regards,
Marc-Olaf Jaschke
101 - 195 of 195 matches
Mail list logo
