The Heap™ – Consultancy ©® - 2014-08-05 (page 2 of 2)

« first day (1309 days earlier) ← previous day next day → last day (3567 days later) »

7:03 PM

@swasheck :)

Fool, databases do not feel!

0

Q: Bucardo status does not feel if the slave is down

I am using bucardo as third party to postgresql replication between two sites. everything ok except buacrdo status <sync_name>. I have a Master-slave sync on master server. when I poweroff the slave server and call buacrdo status <sync_name> on master sell Active Last good : Aug ...

replication postgresql-9.3 monitoring

7:32 PM

muut.com/blog/news/april-2014-service-failure.html

7:45 PM

dont you hate it when you don't save a query?

no?

just me?

so did you just call me an idiot?

idiot

:)

If you had ssmstoolspak installed, you wouldn't have lost it

even after a week?

@ypercube TL;DR We didn't have a good DR/backup strategy. Bad things happened. We blamed the weather.

@swasheck Depends on how big/often your queries are ssmstoolspack.com/Features?f=3

7:54 PM

hm

ok

is this guy bat shit crazy or is it just me

Just a logical proof: the only reason for denormalizing normalized database is to gain performance. That means, that normalized database hurts performance some way. A saw a lot of systems with very normalized database containing a lot of denormalized "caching" tables. — Dawid Ferenczy 21 mins ago

@swasheck work with your scripts in a dropbox/drive folder ;). They're backed up automatically.

@Marian BLOCKED!

But there's no web filter for comedy gold

sorry, you're working for maroons.. :/

7:56 PM

nah. it's not a big deal. i can recreate but it'd just be nice to have had it

@Marian they just love their "security"

@Kermit Is this ok?

@DawidFerenczy That's not a proof. The usual reasoning (not reason) for denormalizing normalized database is to gain performance. That doesn't mean it is justified. So, your argument is circular and not a proof. — ypercube 23 secs ago

@ypercube yes. i just don't follow their logic.

@Kermit It's the usual song. "Joins are slow. Denormalize."

Translated "I don't care about my data. Denormalize."

But the answer - besides that opening line - has some valuable points. But using not the terms I'd prefer.

8:04 PM

well, an excessively normalized database can be a PITA

@Lamak I'll take writing 40 JOIN's over a denormalized table with 5702 columns

@swasheck there was a question about how to recover an unsaved script (in SSMS).. but don't seem to find it. (TLDR there's a folder where SSMS saves files..you might be able to find yours there - in case you close SSMS without saving)

@Kermit What's the limit on Joins in Vertica, in a single query?

@Kermit I said "excessively" and I really don't advocate for a very denormalized database

There are plenty of reasons to de-normalize a database but if you're suggesting it will just arbitrarily improve performance you're not looking at it correctly.

8:07 PM

@ypercube that's an excellent question

@Kermit and we are waiting for an answer

@Zane exactly

MySQL is 61, SQL-Server 256 I think.

Oracle, no limit.

@ypercube they just start charging over 1000

2

@DawidFerenczy the issue is that you're talking like de-normalizing auto-magically improves performance. It certainly can depending on the type of workload you are looking to do. Looking to get into cubes you may want to consider de-normalizing. However if you're just de-normalizing to "increase performance" it's pretty much not going to work. — Zane 6 secs ago

@Lamak haha, nice

8:12 PM

It's one of those things where people are looking in absolutes.

> You can write queries with expressions that contain the <=> operator for NULL=NULL joins

> HP recommends that you do not write implicit cross joins

@Zane to be fair, he did say: "But it depends on particular case, of course."

@AaronBertrand Just wanted to clarify that De-Nomalizion <> performance gain.

Truncation = performance gains

@ypercube @Kermit I can't find the maximum number of joins for vertica anywhere

8:18 PM

@billinkc Truncation is for wussies I'm all about drop.

@Lamak @ypercube nor could i

@Kermit well, thanks for asking then. The documentation only talks about restriction on pre-join projections

@billinkc If you do TRUNCATE TABLE WITH NOLOCK it goes like 10x as fast.

oooh

@Kermit I was going to answer 7 but then Vertica would know who I am.

8:26 PM

@JNK NOLOCK is for the suxs. Professionals know that SET TRANSACTION ISOLATION LEVEL READ UNCOMMITTED is webscale performant

@JNK I am teaching a co-worker the wonders of Powershell and he's eating it up

sitting in front of a mirror are we?

3

Ha!

@billinkc Oh ye of little faith and even less brain, Jason Kassay is revelling in the power of the Dark Side.

@MikeFal boooooooooooooooooooooooooooooooooooooooooooooo

8:29 PM

@MikeFal He's probably wishing he could put a light saber to his head and turn it on

If only the PowerShell and SSIS people could unite we could defeat all the naysayers. However I feel this is a day we will never see.

It's disagreement out of habit, there's no belief in what we say

for me it's more of an exercise in cutting off my nose to spite my face

@billinkc obviously.

@swasheck White + Dark Force: 0 joins.

8:43 PM

i'm sure that means something - i'm just not smart enough to figure it out

@Zane answer anyway

@ypercube as in cross join? or just.. no join?

@Lamak i hope you'll be able to sleep better once we find out

yeah, I'm still awake because of it

@Lamak don't hold your breath :)

8:48 PM

@Marian I guess none. "I'll never JOIN you" he says.

Force is denormalized.

@ypercube they could union.. you know?

@ypercube got it.

you can't join if you never APPLY

wait ... that doesnt make sense

@swasheck but sounds poetic

style ?> substance

My name is a verb now. Signed by Shog:

2

Henceforth, comments expressing a lack of surprise shall be known as "dutching" and those taking offense "ypering". — Shog9 ♦ 58 mins ago

8:56 PM

soooooooo ... apparently both my presentation and larock's presentation were accepted

@swasheck bummer

@ypercube don't yper on us, bro

@Kermit why bummer?

in the end it's the best possible outcome, IMO

Yasir Arsanukaev

0

Q: Delete backups from old incarnation in rman

I did a refresh on a non-production database and these are the incarnations: List of Database Incarnations DB Key Inc Key DB Name DB ID STATUS Reset SCN Reset Time 1 1 ORCL 1355555557 PARENT 1 20-DEC-11 2 2 ORCL 1355555557 ...

oracle rman backups

VtC pls before it's too late to mgrt thx.

@swasheck you have to talk now

i wouldnt have submitted if i didnt want to talk

8:58 PM

how much are they paying you

@swasheck Are they that similar?

@Kermit 0

@swasheck so let me get this straight. you're going in front of bunch of people and talking for $0?

@MikeFal other than the fact that his is wrong ... i was going to address the new CE as part of mine.

@Kermit experience, free lunch, and free trinkets

@swasheck trinkets are nice!

9:00 PM

but ... that may get taken out for the sake of redundancy

@MikeFal ... @PaulWhite was in here talking about it with us

> SQL Server 2014 includes a new cardinality estimator. This feature allows for statistics to be updated more frequently than previous versions of SQL Server, resulting in better query performance. Attend this session and understand how, why, and when SQL Server uses statistics and how this new feature will help improve performance.

> SQL Server 2014 includes a new cardinality estimator: TRUE

> This feature allows for statistics to be updated more frequently than previous versions of SQL Server: FALSE

> resulting in better query performance: NULL

Couldn't have put it better.

well you did --- that was your explanation

I know right

@ypercube I prefer "ypercubing"

fo sheezy

9:10 PM

I imagine that @PaulWhite has chat notifications hooked up to his alarm clock

@swasheck So Steve selected the sessions based on people's request. I'm going to dodge the right/wrong thing because I don't know.

@swasheck I'd say if you don't feel comfortable giving your session, then you can ask Steve to switch it.

Or go to Tom's session and be "that guy"

@MikeFal dodge away. it's totally fine and i dont care. it's the one i wanted to present and i'm comfy with it.

@billinkc i'm thinking i'll just email him and let him know that there's a potential for overlap and seeing if he's got a grasp on what he's going to cover

diplomatic

human

9:20 PM

being

diplomatic

shutdown -R 0

Intergalactic

planetary

planetary

bumr

9:24 PM

@Kermit Restart-Computer

@SweatCoder Actually it is exactly the same thing. You'll place your daynumber in the new columns. You can place your existing query without the group by, count inside a subquery and you'll create a new column with the datepart for each day, and then count(CheckInDate) for yourdate in ([1], [2], [3]...) — bluefeet ♦ 5 mins ago

sigh people don't want to read dups and then have a user gold badge reopen with a single vote

you had closed it?

link to the dup so I can close it?

I closed it as a dup of stackoverflow.com/questions/15745042/…

Andomar came in and reopened it with one vote because it can't be a dup

Surprised mod-close as dup can be undone by a gold badge holder

this is why we need a vote weight system

9:26 PM

Ah, the proletariat does not like when the plebians rise up against their mixed word phrase things

@PaulWhite that's the point

in the event we are wrong we can be overruled by the community

closed it again

56

A: Give high-rep users extra weight on close votes

Though this question specifically discusses high-rep users, I would not feel comfortable having my vote be binding. Instead, I would propose a tiered close vote system since close votes can be cast beginning with 3,000 reputation. This would also remove the tag or downvote requirement. Reputatio...

even with a useless vote from @billinkc

@Lamak wow, you hear that @billinkc? you're useless.. utterly completely useless

9:27 PM

@bluefeet I guess. Still, surprised me.

The latency to Chilemosa was too great

CLOSE WAR!!!

@billinkc Chilemosa?, that sounds like something Jar Jar Binks would say

Aren't you a gungan?

not to my knowledge

9:30 PM

@PaulWhite since we can make mistakes, it's a check/balances thing

But you're obnoxious and do nothing to advance the plot...

@billinkc you have a point there

One of the few times

@Lamak no he doesnt

> Posted locked by JarJarBinks♦♦♦ 1 mins ago

9:31 PM

gimme the codez

Bluefeet, that sounds great, but can you show an SQL example? — SweatCoder 44 secs ago

@swasheck I didn't say it was a good point though

Well, that's the point of saying that this is a duplicate: not answer the same question again — Lamak 57 secs ago

A SQL Example? In BlooFoot's answer, there are those curious SQL Fiddle with Demo links. You won't believe what happens when you click those — billinkc 14 secs ago

My sarcasm detector just overflowed a bigint

Should have defined it as numeric(38,0)

6k worth of {sql} tags

yeah

nevermind

9:37 PM

I agree, they were probably poorly tagged question

72

A: How do I specify "close existing connections" in sql script

You can disconnect everyone and roll back their transactions with: alter database [MyDatbase] set single_user with rollback immediate After that, you can safely drop the database :)

Is this answer more correct than the others?

3

A: How do I specify "close existing connections" in sql script

According to the ALTER DATABASE SET documentation, there is still a possibility that after setting a database to SINGLE_USER mode you won't be able to access that database: Before you set the database to SINGLE_USER, verify the AUTO_UPDATE_STATISTICS_ASYNC option is set to OFF. When set to ON...

@SweatCoder how about this or even this — bluefeet ♦ 44 secs ago

they wanted examples with dates

@billinkc No, just longer. I find USE [targetdb] before the SINGLE_USER operation works well.

So there's only one connection, and it's me.

Wasn't sure whether their line about the auto updates bit was something to worry about

Makes sense though, if I'm in there, the background process can't touch it

(yes, I'm doing the MC Hammer dance)

9:44 PM

Can't touch this. Ner ner ner.

@billinkc It just looks 'clever' and makes the answer longer. IMO.

I TOLD YOU KIWIBOY, CAN'T TOUCH THIS

:)

@billinkc bloofoot?

Why not

Variety ~= spice of life

9:46 PM

melange is also the spice of life

@Kermit at least not boofoot

@bluefeet true

I had thought about flueboot but passed

9:58 PM

@billinkc curry ... or maybe cumin

« first day (1309 days earlier) ← previous day next day → last day (3567 days later) »

Transcript for

Aug '145

The Heap™ – Consultancy ©®

General on- and off-site discussion for dba.stackexchange.com....

frogs mysql nosql oracle postgresql sqlite sql-server

join this room
about this room

00:00

06:00

12:00

18:00

all times are UTC

site design / logo © 2024 Stack Exchange Inc; legal

mobile