The Heap™ – Consultancy ©® - 2023-01-31

« first day (4410 days earlier) ← previous day next day → last day (451 days later) »

12:36 AM

If I disable an index, can I still use my dishwasher?

> Yes, they're completely different things

I see, tyvm

@Zikato

I disabled an index and now my car won't start

1:11 AM

> my car won't start

Perhaps it has no tank?

Since you have nowhere to go, you might as well start the dishwasher

1:28 AM

At least the day won't be a total loss

Hmmmm

I wonder if that feature is named because once you use it, your service will be broker

Much to think about

2 hours later…

3:13 AM

> Lisa Loring, best known as the first actress to play Wednesday Addams in the original The Addams Family sitcom, has died at the age of 64.

A black day

1 hour later…

4:17 AM

Ugh, dying on a Monday.

Yeah

> Cindy Williams, who played Shirley alongside Penny Marshall's Laverne in "Laverne & Shirley," one of the most popular U.S. TV shows of the 1970s and 1980s, has died at 75, the AP reported on Monday, citing a statement from her family.

2 hours later…

6:05 AM

Morning

6:46 AM

It should be named Service Borker

Morning

7:14 AM

@ErikDarling set @variable = @Max

2 hours later…

9:24 AM

'Morning all.

2 hours later…

Master Database

11:03 AM

0

Q: ChatGPT answers?

I know that ChatGPT is banned on Stack Overflow. Lately, I saw an increase of answers by ChatGPT . Join twice on same table with different conditions and no PK / FK relations Get table data not in pivot table Are the answers generated by ChatGPT banned/allowed on Database Administrators ?

11:18 AM

I marked that featured because I think it's important we have an agreed policy on this.

2

12:03 PM

It’s also interesting to think what will happen when chatgpt is being trained on content it produced on sites that it was originally trained on

Maybe it would just go mad

12:15 PM

I would see one use for it

Helping people find answers that exist while typing questions

But that’s not a policy matter

1 hour later…

1:21 PM

@PaulWhite agreed

@ErikDarling maybe that's when it becomes sentient

cool then it can pay rent

ConcernedOfTunbridgeWells

I think Azure SQL needs a hierarchical structure for schemas so you can use them as a proper namespace for stuff that has to reside within the same database.

congratulations on inventing xml

ConcernedOfTunbridgeWells

@ErikDarling Nice singing voice, though. Reminds me of Amy Lee.

@SeanGallardy Started out life as a freebie before SQL Server 2005 came out, so I think about 20 years or so.

@ErikDarling No, I was thinking of something like schemas in the form of foo.bar.wibble so you could use them to organise a database that has to have a lot of stuff cohabiting in it because of Azure licensing.

In related news, Azure is expensive and slow.

6

so a foo walks into a bar and orders a wibble

microsoft is really blowing the whole sql server cloud thing

1:37 PM

@ErikDarling I can't wait till it reads all the posts saying not to believe it; it'll become a loner with no friends and will think all humans are not worthy.

@HannahVernon well brent said it's bad today so we have nothing to fear

Oh well then awesome

ConcernedOfTunbridgeWells

@ErikDarling I think the problem is more that shared storage is expensive and slow, and any hosted service is going to be stuck using it because of the nature of the beast.

yeah not exactly what you want your enterprise relational database stuck with

ConcernedOfTunbridgeWells

Back when I were a wee lad, conventional wisdom is that direct attach storage is an order of magnitude cheaper per TB than shared and two orders of magnitude cheaper per IOPS. That was back in the days of spinning rust - I suspect that the difference is even bigger now.

1:40 PM

now compare to the cost of memory, ho ho ho

ConcernedOfTunbridgeWells

I'm getting deja vu from 2006-7 when I first started hitting blade/SAN architectures and wondered why everything was so slow.

don't worry, pcie5 will save us all

56.6k connections to the storage block

time to go listen to chinese drums

bbl

ConcernedOfTunbridgeWells

@HannahVernon That's what they said about Infiniband.

@ErikDarling The I/O on the servers at work is terrible as their hosting provider went bankrupt and they had to shift at short notice. At one point I was benchmarking table scans at about 2MB/sec. They put 768GB on one of the servers to get around it.

@HannahVernon ... and remote memory

2:27 PM

Your tempdb is taking too much load and is definitely on slower hardware or something is wrong in underlying hardware where Tempdb resides. I believe under heavy load sql server freezed. — Shanky 5 hours ago

well okay then

2:56 PM

Wow poor SQL Server

3:14 PM

@ErikDarling when I worked for the government agency here, we had those I/O delays every night when every.single.server ran maintenance.

we split the schedules up and that took care of most of the issue.

sounds like a SAN problem

it does

also, index rebuilds are awesome at performing lots of IO

imagine falling for that meme

lol

@HannahVernon SQL Server doesn't do I/O, ask Sean.

3:16 PM

memes were a mistake

3

IO was a mistake

3:30 PM

Imagine rebuilding indexes every night

3:44 PM

@ConcernedOfTunbridgeWells [foo.bar.wibble] - then it's just about managing all your brackets, including the ones where it would be useful to show them in SSMS...

user image

Actually worked on a project that had some stuff like that.

4:12 PM

I remember discussing something like this with Paul once, and he pointed out that putting dots in object names does not comply with the rules for "regular identifiers".

Which is annoying, and might cause problems with 3rd party tools.

lotta quotename

Quotename would make a good column name

@JoshDarnell It causes problems with 1st party tool!

I don't use it often, but I have some particular set of views where it's important to surface multiple columns with their original source but which have identical names, and . is the separator closest and easiest to copy and paste and modify.

4:50 PM

@CadeRoux You're a monster.

5:01 PM

This is the worst: learn.microsoft.com/en-us/troubleshoot/sql/database-engine/…

@JoshDarnell 😅😅😅

@SeanGallardy Thanks, Sean! 😜 Seriously, I'm going through the steps with a guy on our team that upgraded to Win 11. So far we've been unable to get SQL Server 2022 Express to install successfully on his machine with the TF or with the reg key.

Just move your files to a different disk, preferably in Azure

The reg key specifies a very specific driver: stornvme

Are they using nvme?

@SeanGallardy That reg path was already there, so I imagine so. But I can confirm. He's uninstalling the partially installed Express Edition and rebooting before we try again.

5:07 PM

because there are some spinning media that have 8k, 16k, and 32k sizes

We didn't reboot after the initial reg edit, so maybe that's what we missed.

14

Q: Does SQL Server 2017, including older versions, support 8k disk sector sizes?

Disk (loosely worded to include not only rotational media but non-rotational media [SSD, NVMe, etc.]) drives are continuing to evolve in their underlying formats and hardware. Part of this was an "enhancement" from 512 byte physical sector sizes to 4k physical sector sizes, which changes the on d...

sql-server hardware

yeah when editing the registry a reboot is generally a good idea

@SeanGallardy Ah, okay.

If that does the trick, maybe the article can be updated.

We can do that

Or maybe I was just supposed to be smart enough to know to reboot after that.

5:12 PM

If the master of documentation didn't know, then it is safe to say the genpop won't know either

🤣🤣🤣

@HannahVernon are you all using dhcp in your AG clusters?

@SeanGallardy no. I started out using DHCP for me AG machines at home, but they are statically assigned now.

why?

my AGs at work are all statically assigned, both for the network adapters and the VNNs

I took a look at why the messages might be printed

They are really only printed for AGs when there are changes to the listener

this could be from the cluster side or the SQL Side

@SeanGallardy weird

5:25 PM

for example, for SQL it could be DDL

from the cluster it could be a signal that the object changed somehow - such as Address/registration/offline/online

strangely the messages are occurring once per day at 9:10 am. The prior DBA had some "interesting" jobs, perhaps there is something else going on as the root cause.

😂

@SeanGallardy I appreciate you looking into it for me :-)

np

I'm looking through one of his procs, and am at line 2060 already

5:27 PM

...

just... why

believe me I ask myself the same question almost daily

it has probably 100+ calls to other procs in the DBA database, too

It could also be that at 9:10 am is when the dns refresh occurs

I was going to look at the execution plan, but I don't want to for obvious reasons.

I hear Erik likes looking at them

@SeanGallardy yah, it's interesting for sure

5:29 PM

You could set the cluster log level to 5, size to 2000, and then generate one at 9:11 am and see if it correlates

might forget

the error number has the mark of the devil, 9666, obviously it's related to Service Broker

5:54 PM

🤣😂🤣😂

6:06 PM

> The design of the machine by the cluster service availability group listener every 24 hours for DNS updates to occur on Network Name This causes the logs will be written to Error log.
> The AlwaysOn 's message is an informational message indicates that the problem behaviour and can be ignored because it does .

thanks for the gibberish.

Wow.

I left out the unintelligible bit, too

Maybe ChatGPT wrote it.

I'm going to ask ChatGPT

apparently it's busy

@SeanGallardy Confirmed, the reboot did the trick. Want me to submit a PR against that page?

6:17 PM

that would save a call to MS Support where they ask you if you've rebooted

don't they tell you to try rebooting in the automated email

"many issues can be resolved by rebooting your device"

6:36 PM

is installing postgres on windows cheating?

@JoshDarnell Sure, I can't approve it though :D

@SeanGallardy FINE!

I have no power here

github.com/MicrosoftDocs/SupportArticles-docs/pull/910

@ErikDarling depends, are you having romantic discussions with it?

ChatGPT:

This error occurs when the Service Broker endpoint is in a disabled or stopped state, which could be due to:

    The Service Broker is not enabled on the database.
    The endpoint has been dropped or altered.
    The service has been stopped or restarted.
    The connection is not open.
    The certificate used for encryption is invalid or missing.
    To resolve the issue, check the Service Broker status, ensure it is enabled on the database, and validate the endpoint, service, and connection state. You may also need to check the encryption certificate.

6:53 PM

@ErikDarling using an installer?

@ConcernedOfTunbridgeWells it's the slowest cloud storage i've experienced, that's for sure

and the AI is completely wrong in your case

@swasheck They must not be galaxy scale

intergalactic planetary

i don't know

i just don't wanna deal with the weirdness around ununtu vms

just use docker

_runs and hides_

what kind of sick human being are you

7:01 PM

i was going to say it's not cheating if you compile from source using VC++

but i suggested docker.

AM I NOT MERCIFUL?

I would need to hire a consultant to do this

i hear that good consultants are hard to find these days

good anything is hard to find these days

the world has moved on

decay and decline lay ahead

get off my lawn

2 hours later…

9:34 PM

PaulW - there's a typo in here "lower in thsi test"

1 hour later…

10:42 PM

Definitely a rough year if Erik is moving to postgres

11:13 PM

@MichaelGreen please to submit PR

11:25 PM

And don't forget to reboot

« first day (4410 days earlier) ← previous day next day → last day (451 days later) »

Transcript for

Jan '2331

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