inconvenient if
that wasn't possible.
Do you have an example on where a grant prevents dropping an object?
hp
--
_ | Peter J. Holzer| Story must make more sense than reality.
|_|_) ||
| | | h...@hjp.at |-- Charles Stross, "Cre
ex: 19). But in Postgres
> the
> same query returns result as "19 days".
Which PostgreSQL version is this?
I get 19 with PostgreSQL 11.16 and 14.0.
hp
--
_ | Peter J. Holzer| Story must make more sense than reality.
|_|_) ||
| | | h...@h
er". (Please include relevant details when you ask a
question here - don't expect people to look at a stack overflow
question).
> How can I fix that `();` issue? Is this documented behavior?
My guess is that's a bug in DBeaver.
hp
--
e
changes also result in an atomic disk write seems to be both pointless
and impossible (that might be many gigabytes of data).
hp
--
_ | Peter J. Holzer| Story must make more sense than reality.
|_|_) ||
| | | h...@hjp.at |-- Charle
On 2022-07-12 13:07:41 -0600, Rob Sargent wrote:
> I thought OP was hinting at WAL stuff defn here
So did I.
hp
--
_ | Peter J. Holzer| Story must make more sense than reality.
|_|_) ||
| | | h...@hjp.at |-- Charles Stross, "Creative
r to connect to an oracle database that's too old or
too new (or you may be able to connect and then get weird errors -
BTDT). PostgreSQL is in my experience rather tolerant of client/server
mismatches, but I wouldn't be surprised if some stuff wouldn't work if
the versions are too diff
r is quite a bit faster that copying into the
database (and therefore also copying out AND copying in).
hp
[1]
https://github.com/hjp/dbbench/blob/master/import_pg_comparison/results/akran.2019-12-15/results.png
--
_ | Peter J. Holzer| Story must make more sense than re
aster, as last time I looked (it's been
some time) the optimizer wasn't especially good at handlung DISTINCT
FROM (probably because it's so rarely used).
hp
--
_ | Peter J. Holzer| Story must make more sense than reality.
|_|_) ||
| | | h...@h
ght.
Otherwise: For the same reason I would prefer B, I would prefer the one
with the most up-to-date data. But there might be other considerations,
e.g. the network connections (bandwidth and delays) between the
surviving members and the clients.
hp
--
_ | Pete
On 2022-08-06 15:06:06 -0500, Ron wrote:
> On 8/6/22 03:40, Peter J. Holzer wrote:
> > Using sync replication on an unstable link is probably not a good idea.
> > Every time the link goes down, A freezes. Is this what you want?
>
> I had to fight my end users about how to r
nd an fqdn at most 255
bytes. So without the scan id we are at 285 bytes. maybe a bit more due
to overhead. That leaves about 2400 bytes for the scan id. I don't know
what a scanid is, but 2000+ bytes for an id seems excessive.
hp
--
_ | Peter J. Holzer| Story
valuated) followed by a very quick count up (while
the result is transmitted to the client). Probably not what you want.
hp
--
_ | Peter J. Holzer| Story must make more sense than reality.
|_|_) ||
| | | h...@hjp.at |-- Charles Stross, "Cr
seconds, the progress
indicator is useless.
You are stuffing the whole result into an array and THEN counting the
number of elements. So when you get to the count all of the work (except
sending the result to the client) is already done, so there is little
point in displaying a progress indicator.
er clarify that he's fetching a lot of data for each row, so
«select count(*) ,,,» is indeed much faster than «select * ...» due to
the sheer amount of data to be transferred. That wasn't obvious from his
first message, but I hedged against the possibility in my answer.)
ursive query. That's theoretically possible but I would be
surprised if it actually did this. (It didn't in my tests, but my test
data set was too small to get it to even use indexes with normal
queries).
hp
--
_ | Peter J. Holzer| Story must make more sense
ed int
to an int halves the positive range. But it seems that this is already
capped at 2**31 days (5874897 AD), so that wouldn't be a problem here.
hp
--
_ | Peter J. Holzer| Story must make more sense than reality.
|_|_) ||
| | | h...@hjp
uncertainty of +/- 50 years?
I guess to really store "what do I know about when something happened"
you would have to be able to store a number of constraints (like
"between year x and y", "at least d years after event e", "between month
m and n", etc.)
ve your problem.
Isn't that what logical replication basically does?
I also think I've seen other tools parsing the WAL stream and doing
something useful with the results.
hp
--
_ | Peter J. Holzer| Story must make more sense than reality.
|_|_) |
. If I run
% sudo -u postgres -H /usr/lib/postgresql/13/bin/psql
(which is not a symlink)
I get the same behaviour. So it seems that psql changes to its basedir
and then can't change back again.
And sure enough, strace shows:
chdir("/usr/lib/postgresql/13/bin") = 0
chdir("/h
this affects auto_explain.
hp
--
_ | Peter J. Holzer| Story must make more sense than reality.
|_|_) ||
| | | h...@hjp.at |-- Charles Stross, "Creative writing
__/ | http://www.hjp.at/ | challenge!"
signature.asc
Description: PGP signature
PG-USERNAME
localusers rootpostgres
Then root can invoke `psql -U postgres ...`, but other users can't.
hp
--
_ | Peter J. Holzer| Story must make more sense than reality.
|_|_) ||
| | | h...@hjp.at |-- Charles Stross, "Creative writing
__/ | http://www.hjp.at/ | challenge!"
signature.asc
Description: PGP signature
On 2022-09-01 20:49:56 -0400, Jeffrey Walton wrote:
> On Thu, Sep 1, 2022 at 8:23 PM Peter J. Holzer wrote:
> >
> > On 2022-09-01 18:16:14 -0400, Tom Lane wrote:
> > > Jeffrey Walton writes:
> > > > We are having a heck of a time getting PostreSQL utilities to
eld4`: I have no
idea what this is supposed to do, so it's hard to tell if there is a
better way. Using `a` to refer to 3 different things doesn't help
either.
hp
--
_ | Peter J. Holzer| Story must make more sense than reality.
|_|_) ||
| | | h...@hjp.a
efense. The database frequently won't
be accessible from the open internet (or even the company network)
directly. Only a middle tier of application servers running vetted
client code will connect directly. Even those servers may not be
accessible directly to end users. There may be a layer of proxy s
t; users. There may be a layer of proxy servers above them. Each of these
> > > layers may implement additional checks, rate limits and monitoring.
>
> If no one has direct SQL access to the database, then there's no problem with
> a
> role being able to pg_terminate_backe
ons ...)
hp
--
_ | Peter J. Holzer| Story must make more sense than reality.
|_|_) ||
| | | h...@hjp.at |-- Charles Stross, "Creative writing
__/ | http://www.hjp.at/ | challenge!"
signature.asc
Description: PGP signature
oles by another name" or
"just data".
hp
--
_ | Peter J. Holzer| Story must make more sense than reality.
|_|_) ||
| | | h...@hjp.at |-- Charles Stross, "Creative writing
__/ | http://www.hjp.at/ | challenge!"
signature.asc
Description: PGP signature
bout it" abstraction which puts it closer to the "roles by another
name" camp.
> Platform Multitenant Architecture
> https://architect.salesforce.com/fundamentals/platform-multitenant-architecture
That's *too* high-level for me. There's any number of techniques which
cou
s of course assumes that the behaviour is intentional and not a
bug.)
hp
--
_ | Peter J. Holzer| Story must make more sense than reality.
|_|_) ||
| | | h...@hjp.at |-- Charles Stross, "Creative writing
__/ | http://www.hjp.at/ | challenge!"
signature.asc
Description: PGP signature
itches to a generic plan if it
thinks that the generic plan isn't worse than the specialized plans.
If the plan suddenly gets worse after 5 executions, you've probably run
into a case where the generic plan is worse although the cost computed
by the planner isn't.
hp
--
_
On 2022-10-01 20:24:21 +0800, Julien Rouhaud wrote:
> On Sat, Oct 01, 2022 at 02:05:53PM +0200, Peter J. Holzer wrote:
> > On 2022-09-30 17:59:01 -0700, Bryn Llewellyn wrote:
> > > set rls.tenant_id=42;
> >
> > This works because there is a "." in the name.
║
╟──╢
║ 0.000100 ║
╚══╝
(1 row)
So the number of decimals by default isn't sufficient to represent
10^-18. You have to explicitely increase it.
hp
--
_ | Peter J. Holzer| Story must
imate in the first tuple changes
without any actual data change (although the only reason I can think of
right now would be an ANALYZE (in another session or by autovacuum)).
But the actual rows definitely shouldn't change.
hp
--
_ | Peter J. Holzer
nd when I check it says
[doesn't use /home/dmartuser/pgsql/14/data]
>
> How can I point the service to read the new path ( /home/dmartuser/pgsql/14/
> data )?
Check the systemd configuration file for this service.
hp
--
_ | Peter J. Holzer| Story must
initions the
same?
(Just trying to rule other other possible error sources.)
> I do apologize, but I do not understand the value of doing that select
> juggling.
I think Allan may have misread your mail.
hp
--
_ | Peter J. Holzer| Story must make more sens
On 2022-10-27 15:07:06 +0300, Kristjan Mustkivi wrote:
> On Thu, Oct 27, 2022 at 12:18 PM Peter J. Holzer wrote:
> > On 2022-10-27 10:55:31 +0300, Kristjan Mustkivi wrote:
> > > We use dockerized postgres.
> >
> > So that means you aren't just replaci
t at that time or does it just
issue a notice and continue to rebuild the other indexes? In the latter
case it migh be easy to miss a problem.
hp
--
_ | Peter J. Holzer| Story must make more sense than reality.
|_|_) ||
| | | h...@hjp.at |--
dex, but why would you?)
> I do like this. I add oid in pg_am.dat and pg_proc.dat for my index.
> And I add codes in contrib and backend/access/myindex_name, is there
> any other places I need to add some infos?
What? Why are you doing these things?
hp
--
_ | Peter J. Ho
/s user is
> logging
> in using their own name as the requested login role.
I think that's not quite correct. The -U option affects which user name
psql uses to connect to the server. It is psql which defaults to the
OS user name in the absence of the -U option (or the PGUSER environment
arted: "su mary".
>
> 2. Then I want to start a session (I use "psql" here an an example) like
> this:
> "psql -d postgres".
>
> 3. Then, at the "psql" prompt, I want "select session_user" to show "bob".
Set the PGUSER=
AICS your test users aren't supposed to be that.
> uid=1001(postgres) gid=1001(postgres) groups=1001(postgres),27(sudo),114
> (ssl-cert)
And is there a reason for posgres to be in group sudo?
hp
--
_ | Peter J. Holzer| Story must make more sense than reality.
|_|_) |
nant:~ 22:46 :-) 1016# su - 'mac$crooge'
mac@trintignant:~$ id
uid=1002(mac$crooge) gid=1003(mac$crooge) groups=1003(mac$crooge)
mac@trintignant:~$
I'm not saying that doing this is a good idea ...
hp
--
_ | Peter J. Holzer| Sto
t; select * from b;
╔╤═══╤══╗
║ id │ a │ t ║
╟┼───┼──╢
║ 1 │ 1 │ foo1 ║
║ 2 │ 1 │ foo2 ║
║ 3 │ 2 │ bar1 ║
╚╧═══╧══╝
(3 rows)
And the data in the table is also unchanged.
hp
--
_ | Peter J. Holzer| Story must make more sense than reality.
|_|_) ||
| | | h...@hjp.at |-- Charles Stross, "Creative writing
__/ | http://www.hjp.at/ | challenge!"
signature.asc
Description: PGP signature
nd even those that do
need only one key, so it is sufficient that only the server has a
certificate.
hp
--
_ | Peter J. Holzer| Story must make more sense than reality.
|_|_) ||
| | | h...@hjp.at |-- Charles Stross, "Creative writing
__/ | h
sudo vi /etc/postgresql/14/main/pg_hba.conf
> [sudo] password for aklaver:
>
> which opens pg_hba.conf for editing.
Well, yes. Root can edit the file, too. But root can edit anything[1].
hp
[1] Except ... lots of stuff, actually.
--
_ | Peter J.
s the latter, your programming language's postgresql library
probably has a method for invoking copy.
> Has anything changed in the last ten years? Or, is there a
> better way to copy file contents in a remote database?
COPY is the fastest way to load data.
hp
--
_
you use copy_from()
you don't have to parse it (but then why use Python at all?)
hp
--
_ | Peter J. Holzer| Story must make more sense than reality.
|_|_) ||
| | | h...@hjp.at |-- Charles Stross, "Creative writing
__/ | http://www.hjp.at/ | challenge!"
signature.asc
Description: PGP signature
On 2022-11-09 12:57:23 -0600, Ron wrote:
> On 11/9/22 10:17, Peter J. Holzer wrote:
> > On 2022-11-07 14:40:40 -0600, Ron wrote:
> > > On 11/7/22 10:57, Вадим Самохин wrote:
> > > I have an application that must copy a local file in csv format to a
> > >
pdates, so again some may be HOT updated and some not. If you are
updating the same tupel several times, you may get a few HOT updates
first, then a non-HOT update, then HOT updates again.
hp
--
_ | Peter J. Holzer| Story must make more sense than reality.
|_|_) |
an appropriately configured role with
> "nosuperuser".
One important task that can AFAIK only be performed by superusers is the
creation of functions in untrusted languages like plpython3u and
plperlu.
If your application uses functions in those languages you ne
h
sounds reasonable) while the other reads 9995216 buffers (or almost one
full buffer per row). Why? The entries should be dense in the index in
both cases and since it's an index only scan (and explain says there
were 0 heap fetches) I would not expect extra accesses. Where do these
buffer reads
On 2022-11-18 15:59:46 -0500, Tom Lane wrote:
> "Peter J. Holzer" writes:
> > Both do a parallel index only scan. Both perform 0 heap fetches.
> > But one reads 27336 buffers (or about 22 bytes per index entry, which
> > sounds reasonable) while the other reads
On 2022-11-18 13:09:16 -0800, Peter Geoghegan wrote:
> On Fri, Nov 18, 2022 at 12:46 PM Peter J. Holzer wrote:
> > Both do a parallel index only scan. Both perform 0 heap fetches.
> > But one reads 27336 buffers (or about 22 bytes per index entry, which
> > sounds reason
t editing errors happen and are not always
caught by a reviewer. Having to get the database, the role and the
password consistently wrong is a much greater hurdle that getting just
the database wrong.
So you can give these credentials to you developers or devops folks
(whom you trus
On 2022-11-18 16:21:18 -0600, Ron wrote:
> On 11/18/22 16:13, Peter J. Holzer wrote:
> > So you can give these credentials to you developers or devops folks
> > (whom you trust not attack the system -
>
> They like to "fix" things without documenting what they
╔═══╗
║ ascii ║
╟───╢
║ 3 ║
╚═══╝
(1 row)
Same for the other ctrl characters.
hp
[1] There are usually four Ctrl-Characters which need only a single
key: Ctrl-I (TAB), Ctrl-M (CR), Ctrl-[ (ESC) and Ctrl-H (BS) or Ctrl-?
(DEL).
(On Unix systems CR is
red by the server. So you can't simply
tell whether the stored/used version corresponds to the code you
installed.
I don't know how reproducable that tokenization process is. Can you just
do it again and compere the results?
hp
--
_ | Peter J. Holzer| Stor
are not what I want.
Can you elaborate why you can't use those?
hp
--
_ | Peter J. Holzer| Story must make more sense than reality.
|_|_) ||
| | | h...@hjp.at |-- Charles Stross, "Creative writing
__/ | http://www.hjp.at/ |
that a reason why it should).
> * your logic only works by accident for some languages (try to upcase
> a `ß` or a `ı`)
This is also true of upper() and lower() and SQL does provide those.
hp
--
_ | Peter J. Holzer| Story must make more sense than reality.
|_|_) |
(x, '(.*_)(.*)(_.*)', '\2')) ||
regexp_replace(x, '(.*_)(.*)(_.*)', '\3')
FROM test;
╔═╗
║ ?column? ║
╟─╢
║ abc_DEF_ghi ║
╚═╝
(1 row)
hp
--
_ | Peter J. Holzer| Story must make more sense than reality.
|_|_) ||
| | | h...@hjp.at |-- Charles Stross, "Creative writing
__/ | http://www.hjp.at/ | challenge!"
signature.asc
Description: PGP signature
On 2022-12-10 13:44:37 +, Gianni Ceccarelli wrote:
> On 2022-12-10 "Peter J. Holzer" wrote:
> > > * your logic only works by accident for some languages (try to
> > > upcase a `ß` or a `ı`)
> >
> > This is also true of upper() and lo
.
but that would be insane even for the 26 letters of the basic Latin
alphabet, much less the myriad of accented letters (and other alphabets
like Cyrillic or Greek ...).
On second thought you could probably use NFD normalization to separate
base letters from accents, uppercase the base letters an
'(.)(.)(.)', '\2')) ||
UPPER(REGEXP_REPLACE(x, '(.)(.)(.)', '\3'))
FROM test;
╔═╤═╗
║ x │ ?column? ║
╟─┼─╢
║ abc_def_ghi │ A_DEF_GHIB_DEF_GHIC_DEF_GHI ║
╚══
On 2022-12-10 15:48:58 +0100, Peter J. Holzer wrote:
> On second thought you could probably use NFD normalization to separate
> base letters from accents, uppercase the base letters and then
> (optionally) NFC normalize everything again.
Of course I had to try that:
wds=> select
e not on the same box. They are in a HSM. A dedicated piece of
> tamper-proof hardware that stores secrets (keys).
> The Oracle-server needs to talk to the HSM to get the keys.
If the hacker has root access: What prevents them from talking to the
HSM?
hp
--
_ | Peter J. Holze
On 2022-12-22 11:15:57 +0100, Rainer Duffner wrote:
>
>
> Am 22.12.2022 um 10:46 schrieb Peter J. Holzer :
>
> If the hacker has root access: What prevents them from talking to the
> HSM?
>
>
>
> I wasn’t involved in setting it up here, but AFAIK you
with multiple active nodes running
already.
hp
--
_ | Peter J. Holzer| Story must make more sense than reality.
|_|_) ||
| | | h...@hjp.at |-- Charles Stross, "Creative writing
__/ | http://www.hjp.at/ | challenge!"
signature.asc
Description: PGP signature
living room.
Isn't that how you normally vacuum your living room?
hp
--
_ | Peter J. Holzer| Story must make more sense than reality.
|_|_) ||
| | | h...@hjp.at |-- Charles Stross, "Creative writing
__/ | http://www.hjp.at/ |
On 2023-01-05 12:34:08 +0100, Karsten Hilbert wrote:
> > Von: "Peter J. Holzer"
> > On 2023-01-04 09:34:42 -0600, Ron wrote:
> > > I don't think VACUUM FULL (copy the table, create new indices and other
> > > metadata all in one command) actually
rate.
I'd go for a middle ground: Instead of expiring once per day, use a
shorter interval, maybe once per hour or once per minute. That will
(probably) make each expire job really quick but still create much less
load overall.
hp
--
_ | Peter J. Holzer| Story must make m
On 2023-01-07 07:40:01 -0600, Ron wrote:
> On 1/7/23 05:29, Peter J. Holzer wrote:
> If I understood correctly, you have to delete about 3 million records
> (worst case) from the main table each day. Including the other 8 tables
> those are 27 million DELETE queries e
g like
this to PostgreSQL would be worthwhile?
hp
--
_ | Peter J. Holzer| Story must make more sense than reality.
|_|_) ||
| | | h...@hjp.at |-- Charles Stross, "Creative writing
__/ | http://www.hjp.at/ | challenge!"
signature.asc
Description: PGP signature
On 2023-01-14 06:32:03 +0100, Peter J. Holzer wrote:
> Oracle has a function which returns the internal representation of a
> value as a series of (decimal) byte values. Back in the days when I was
> new to Oracle I used this to figure out how Oracle stores NUMBER, but
> now I'
eted[1]. Whether that means that the space is
"immediately" available again is up to the operating system.
hp
[1] Possibly delayed until commit.
--
_ | Peter J. Holzer| Story must make more sense than reality.
|_|_) ||
| | | h...@hjp.at
ocal time + time
zone, not timestamptz (the time zone can be implicit).
hp
--
_ | Peter J. Holzer| Story must make more sense than reality.
|_|_) ||
| | | h...@hjp.at |-- Charles Stross, "Creative writing
__/ | http://www.hjp.at/ | challenge!"
signature.asc
Description: PGP signature
On 2023-02-02 10:22:09 -0500, Benedict Holland wrote:
> Well... until two processes generate an identical UUID. That happened to me
> several times.
How did that happen? Pure software implementation with non-random seed?
Hardware with insufficient entropy source?
hp
--
_ | P
On 2023-02-05 18:57:13 -0600, Ron wrote:
> Why are you specifying the collation to be "C" when the default db encoding
> is UTF8, and UTF-8 has Greek, Chinese and English encodings?
C is equally bad for Greek, Chinese and English ;-)
hp
--
_ | Peter J. Holzer|
e may be negligible or huge.
It really depends on your access patterns.
hp
[1] There was even a discussion about making that much faster on the
LKML recently.
--
_ | Peter J. Holzer| Story must make more sense than reality.
|_|_) ||
| | | h...@hjp.a
les, schemas and databases and covered by the "usual" SQL privilege
system.
hp
--
_ | Peter J. Holzer| Story must make more sense than reality.
|_|_) ||
| | | h...@hjp.at |-- Charles Stross, "Creative writing
__/ | http:
so during migrations, restores from backups and other
infrequent events.
With random Ids you don't have to worry about this.
hp
--
_ | Peter J. Holzer| Story must make more sense than reality.
|_|_) ||
| | | h...@hjp.at |-- Charles Str
══════════╝
(10 rows)
The latter is almost 1000 times faster. Saving 1.8 ms on planning time
doesn't help you if you the
On 2023-02-11 16:21:49 +0100, Peter J. Holzer wrote:
> On 2023-02-09 18:35:42 +0100, Dominique Devienne wrote:
> > Right. The goal is to (re)use a prepared statement (i.e. plan once), and
> > bind
> > the RHS (binary) array
> > and do a single exec (single round-tri
nd
should be fixed.
OTOH it could also be argued that the optimizer should be able to
perform the same simplifications as I did above and produce the same
code for WHERE (("id" > ? OR "id" IS NULL)) AND (("id" <= ?))
as for WHERE (("id" > ?))
On 2023-02-14 15:36:32 -0700, Rob Sargent wrote:
> But if the query is supposed to be generic and re-used in a situation where id
> could be null, wouldn't the null id records be fetched every time?
No, they will never be fetched because of the AND (("id" <= ?)).
e case (the exact same query is submitted repeatedly) is
sufficiently rare that it isn't all that effective in practice.
(The other techniques mentioned are of course also used by other
databases.)
hp
--
_ | Peter J. Holzer| Story must make m
n be a DNS or LDAP client). (And some programs are
even server and client for the same protocol)
3) A machine intended for running server programs.
You are thinking of the 3rd meaning. My guess is that Albert meant the
first.
hp
--
_ | Peter J. Holzer| Story must make
That doesn't work. A unique constraint can't exist without a (unique)
index. Think about it: With a unique constraint PostgreSQL needs to
check for every insert whether the value already exists in the table.
Without an index this would mean a full table scan.
hp
--
_ | Peter
n is pointing out that CONCURRENTLY cannot be used for
that purpose.
(I realize that your idea is not to create the constraint in the first
place.)
hp
--
_ | Peter J. Holzer| Story must make more sense than reality.
|_|_) ||
| | | h...@hjp.at |-- Charles Stross, "Creative writing
__/ | http://www.hjp.at/ | challenge!"
signature.asc
Description: PGP signature
On 2023-03-04 02:34:02 -0600, Ron wrote:
> On 3/4/23 02:03, Peter J. Holzer wrote:
> [snip]
> > So your plan is to create a unique constraint (backed by a unique
> > index) and then to drop the index and keep the constraint?
> >
> > That doesn't work. A uniq
oubles backslashes. Other
| byte values are represented literally.
So the byte 0x19 is converted to a single character U+0019 (EM) which is
then displayed as '\x19', while bytes >= 0x80 are converted to
four-character escape sequences.
hp
--
_ | Peter J. Holzer
roblems anyway, so it is questionable if one wants to enable external
> # automatic restarts.
> #Restart=on-failure
So I'd try this despite the comment.
hp
--
_ | Peter J. Holzer| Story must make more sense than reality.
|_|_) ||
| | | h...@h
On 2023-03-13 09:55:50 -0800, Israel Brewster wrote:
> On Mar 13, 2023, at 9:43 AM, Peter J. Holzer wrote:
> > On 2023-03-13 09:21:18 -0800, Israel Brewster wrote:
> >> I’m running a postgresql 13 database on an Ubuntu 20.04 VM that is a bit
> >> more
> >>
an alternate locale:
hjp=> select upper('ß');
╔═══╗
║ upper ║
╟───╢
║ ß ║
╚═══╝
(1 row)
hjp=> select upper('ß' collate "de-AT-x-icu");
╔═══╗
║ upper ║
╟───╢
║ SS ║
╚═══╝
(1 row)
The challenge now is to find a
Or possibly counting stuff far more often than necessary. If an exact
count is necessary more frequently than it changes it is probably a good
idea to store that somewhere and update it in a trigger.
(If the count doesn't have to be totally up-to-date, caching it in the
OPY some time.)
With COPYing the output of a SELECT I don't see any savings. On the
contrary, it's an extra step.
hp
--
_ | Peter J. Holzer| Story must make more sense than reality.
|_|_) ||
| | | h...@hjp.at |-- Charles S
wards. The
BSD file system and its descendants (like ext4) don't like getting
completely full.)
hp
--
_ | Peter J. Holzer| Story must make more sense than reality.
|_|_) ||
| | | h...@hjp.at |-- Charles Stross, "Creative writing
__/ | http://www.hjp.at/ | challenge!"
signature.asc
Description: PGP signature
se do I have to
configure? (I know about wal_keep_size, but it was my understanding that
this isn't needed when slots are used)
hp
--
_ | Peter J. Holzer| Story must make more sense than reality.
|_|_) ||
| | | h...@hjp.at |-- Charles Stros
hild(id) on delete cascade,
grandparent int not null references parent(id) on delete cascade,
name text not null,
unique(grandparent, parent, name)
);
hp
--
_ | Peter J. Holzer| Story must make more sense than reality.
|_|_) ||
| | | h...@hjp.at |-- Charles Stross, "Creative writing
__/ | http://www.hjp.at/ | challenge!"
signature.asc
Description: PGP signature
On 2023-03-28 11:07:04 -0400, Jeremy Smith wrote:
> On Tue, Mar 28, 2023 at 10:55 AM Peter J. Holzer wrote:
>
> The configuration includes `use_slots: true` and I can see a slot in
> pg_replication_slots on the leader.
>
> I was under the impression that this woul
On 2023-03-28 17:08:38 +0200, Alexander Kukushkin wrote:
> On Tue, 28 Mar 2023 at 16:55, Peter J. Holzer wrote:
>
>
> However, when we took down one node for about two hours for some tests
> recently (with some moderate traffic on the remaining node), the replica
>
201 - 300 of 640 matches
Mail list logo
