The Heap™ – Consultancy ©® - 2019-11-27

« first day (3249 days earlier) ← previous day next day → last day (1605 days later) »

2:15 AM

@PaulWhitesaysGoFundMonica I'm moderately insulted by this explanation.

2:26 AM

Indeed. 16 bits doesn't count as great length.

Paul White says GoFundMonica

@mustaccio Sounds like the level of insult would fit in 8 bits

Said an 8-bit moderator

Paul White says GoFundMonica

3:28 AM

That represents a four-fold improvement

With a little bit of tango, the two of us might even arrive at one answer to this 0-score question:

0

Q: How to convert an integer that was the end result of a bitwise OR to a table of the individual integers that the bitwise OR was applied against?

How can I convert the final integer of a bitwise OR back to it's original set of integers that the bitwise OR operation was applied to? For example if I have the following set of bit values: {0, 1, 2, 4, 8, 16} and if I have the bitwise OR generated value of 11 then I would like to convert that ...

sql-server sql-server-2016 bitwise-comparison

4 hours later…

Tom V - Team Monica

7:03 AM

Morning

7:30 AM

Morning

I see nibble-conversations

Morning

1 hour later…

Peter Vandivier

8:42 AM

Morning

George.Palacios

8:54 AM

morning

9:40 AM

@Forrest well, it's not a joke

John aka hot2use

10:28 AM

@Forrest congratulations

Morning

Peter Vandivier

11:02 AM

oof @Johnakahot2use that backups question feels a little beefy to be answerable 😬

seems like everything after "...but let me elaborate." is justification on why it's not a duplicate

Peter Vandivier

11:30 AM

happy to give you one more downvote if you want the badge, though 😜

John aka hot2use

11:42 AM

@PeterVandivier If that is your opinion feel free to downvote.

Peter Vandivier

i'm usually very reluctant to downvote

that post (while rambling imo) does not offend me so much that i feel the need to express it with suffrage ;p

John aka hot2use

Downvote would be: This question does not show any research effort; it is unclear or not useful

Peter Vandivier

well it definitely has research effort

and it's not unclear per se, just unfocused

i don't think it's useful atm, but not so useless i want to shame it away without discussion

John aka hot2use

I have done research. It's pretty clear what options I have. Now I'm asking for DBA's implementations based on experience.

I know what you mean.

I might re-formulate parts or introduce more information.

Peter Vandivier

seems like you already have all the information you need - and i know you're a smart guy, so asking such a specific question makes it too niche. it's not generalizeable to anyone but you at any time except for right now

John aka hot2use

11:47 AM

However, I don't see how laying out my whole infrastructure and reasons for not backing up the databases with the existing solution (monetary values) is going to increase the probability of an answer.

How do you backup test databases?

Or not?

Peter Vandivier

exactly the same as production

backup protocol isn't an afterthought, it's foundational

George.Palacios

Sounds expensive

We back up based on service tiers and SLA's

John aka hot2use

@George.Palacios Thanks! :-)

Peter Vandivier

and if you're optimizing for dev, then you're putting the cart before the horse

John aka hot2use

@PeterVandivier Not optimizing for DEV (in question). Dev is on Prod and is backed up.

George.Palacios

11:49 AM

I disagree with that. It's not "optimizing" it's fitting to requirements

If you don't need PIT recovery as an example, FULL mode seems superfluous

John aka hot2use

I'm asking about test databases on test servers. and if to use FULL or SIMPLE.

George.Palacios

What are your SLA's?

What is the level of service impact that can be sustained to that system?

Peter Vandivier

you don't have to back up your laptop or local VMs, (although there's no reason not to if you're building your environment the same there as in prod) - but if it's a shared environment then it gets the shared environment treatment

George.Palacios

If the service can be down for 48 hours, FULL mode isn't necessary

without that information you're just choosing

Peter Vandivier

having a different configuration in dev than in prod means you are deliberately introducing differences to environments that are meant to be similar

George.Palacios

11:51 AM

@PeterVandivier Depends on your path to production and configuration management maturity imho

"differences" aren't inherently bad if they're effectively controlled and governed

But you're effectively trading risk for cost at that point

Peter Vandivier

okay, so drive your ola hallengren backup jobs from config and have a shorter retention period in dev, but keep the architecture the same

George.Palacios

Again - why take log backups if your SLA for that system doesn't require them?

Peter Vandivier

so that dev is similar to prod

George.Palacios

There's no solution to this debate by the way, because we're coming at it from two different perspectives

So why is dev being the same as prod important?

Peter Vandivier

seriously?

George.Palacios

11:53 AM

To you

Personally I think that dev should be as flexible as possible to satisfy dev requirements. Integration testing and subsequent test environments should be close to prod.

Peter Vandivier

so that changes made to dev have a higher likelihood of demonstrating behaviour that is closely representative of what will happen in production

you're arguing semantics now

George.Palacios

I don't think I am

Peter Vandivier

the definition of dev/test/qa/uat/bananas

George.Palacios

dev is dev though

it's development

not test

Peter Vandivier

4 mins ago, by Peter Vandivier

you don't have to back up your laptop or local VMs, (although there's no reason not to if you're building your environment the same there as in prod) - but if it's a shared environment then it gets the shared environment treatment

☝️

George.Palacios

11:55 AM

So as an example, it's a principal that shared environments must be recovered to a point in time?

Peter Vandivier

3 mins ago, by Peter Vandivier

okay, so drive your ola hallengren backup jobs from config and have a shorter retention period in dev, but keep the architecture the same

keep shared environments - whatever label you give them - similar

George.Palacios

Yeah we do that anyway. We just disable the log backups

and have DB's set to simple mode

because devs like to fill those logs far too often for my liking

One thing you haven't touched on is operational overhead

there's a bigger overhead to treating everything as prod

Peter Vandivier

there's less

George.Palacios

If you have the money / resource the throw at it, despite it not being a business or architectural requirement - great! A lot of organisations like to run inherently lean processes which for me don't fit in that bucket.

Peter Vandivier

because you don't need to keep track of multiple architectures

George.Palacios

11:58 AM

"keep track" isn't hard though. Devise your 3/4 off the shelf patterns

that's what I've done

Then when a new system comes in, just pick up the appropriate pattern based on SLA

Flexibility is good too

I see the benefits don't get me wrong, but I tend to come at this from a standpoint of what's the MVP for this service, rather than trying to reduce overhead in the testing / dev process

Automated test suites as an example are another way around these issues.

Low overhead, high return in integration test scenarios

There really is no correct answer though. It's a business decision ultimately

The counter point of your argument Peter is that resource consumption increases. More backup space needed, better IOPS on shared storage needed etc

Paul White says GoFundMonica

What's the point of backing up a test system? The only use I have ever seen is taking a database snapshot after the initial restore from a real backup so the test environment can be easily reset.

George.Palacios

@PaulWhitesaysGoFundMonica See this is where my head is too.

Test for me should be build/destroy

at which point backups are meaningless

Paul White says GoFundMonica

@Johnakahot2use What do you mean by "pseudo backup"?

Backup to NUL? Copy only backup?

The main reason I can see to maintain FULL recovery is to ensure test has similar performance characteristics to production and doesn't sneakily take advantage of optimizations not present on prod. That all assumes the hardware is the same too.

Few places I have seen could afford to have prod-spec hardware for test and dev

12:54 PM

@PaulWhitesaysGoFundMonica Have you heard about Phillip Hughes?

Paul White says GoFundMonica

@AJ no?

Phillip Joel Hughes (30 November 1988 – 27 November 2014) was an Australian Test and One Day International (ODI) cricketer who played domestic cricket for South Australia and Worcestershire. He was a left-handed opening batsman who played for two seasons with New South Wales before making his Test debut in 2009 at the age of 20. He made his One Day International Debut in 2013.Hughes scored his first Test century in March 2009, aged 20, in his second Test match for Australia, opening the batting and hitting 115 in the first innings against South Africa in Durban. This made Hughes Australia's youngest...

He was an Australian cricketer who died after getting injured from a bouncer.

Paul White says GoFundMonica

@AJ Well yes obviously I am familiar with that part, but why mention it now?

Did something happen today?

Well, Michael Clarke posted on Instagram to pay tributes.

Today is his death anniversary.

Paul White says GoFundMonica

Ah I hadn't realized

(also it is 28 November here now)

1:04 PM

ohh.

And you're still up?

Paul White says GoFundMonica

Can that really be 5 years ago

sadly yes. I was traveling to my college to get some documents at that time. I came to know about his death from the headline in newspaper.

There were three pages just for him.

1 hour later…

Peter Vandivier

2:21 PM

i find myself right clicking google search results for database docs pages and manually editting out the "version" with "/stable/" or "/current/"

fighting a losing war over here

Peter Vandivier

2:31 PM

is there a default datatype conversion matrix for postgresql? i'm used to this bad boi from sql server

user image

2:57 PM

@PeterVandivier I don't think so.

It's complicated: postgresql.org/docs/current/typeconv.html

Peter Vandivier

🥺

asking a question on main now... keep the hope alive

Tom V - Team Monica

@PeterVandivier closed as shopping list

Peter Vandivier

come at me bro

your downvotes only feed my self-satisfied rage

as Christmas shopping list better

Peter Vandivier

all the good boys and girls get a visit from the Hot Network Santa

0

Q: default permissible datatype conversion matrix

Today I wanted to define a uuid of value 00000000-0000-0000-0000-000000000000. Being a SQL Server man, I'd usually... select cast(0x0 as uniqueidentifier); ...but I'm in postgres world now so I whipped out a sensible... select cast('\x00'::bytea as uuid); ERROR: cannot cast type bytea to u...

postgresql datatypes type-conversion documentation

Paul White says GoFundMonica

3:19 PM

> cast(0x0 as uniqueidentifier);

wat

Peter Vandivier

he who controls the spice, controls the universe

yesterday, by Peter Vandivier

user image

lol @TomV-TeamMonica

i c u

Tom V - Team Monica

@PeterVandivier The down-vote isn't mine

Peter Vandivier

4 tru?

i am wounded

Tom V - Team Monica

It was already down-voted by the time I clicked the link you posted

Peter Vandivier

slings and arrows

Paul White says GoFundMonica

3:25 PM

seriously that's one of the wackiest bits of code I've ever seen

Tom V - Team Monica

user image

@PeterVandivier not me ^^

Peter Vandivier

haha, i believe you ;p

no worries, really

i wasn't offended even when i thought it was you. i rather assumed it was good natured sassing on your part

George.Palacios

@PeterVandivier It's ok I got you bro

Peter Vandivier

lol ty

Paul White says GoFundMonica

@George.Palacios and compensated for the next four downvotes at the same time

George.Palacios

3:28 PM

hahaha

Tom V - Team Monica

Oh that answer is unexpected :D

Peter Vandivier

works though

glad i asked on main d:

i just wish Laurenz was a little more verbose with his answers

he's always very terse

John aka hot2use

@PaulWhitesaysGoFundMonica A backup to disk with a retention policy of 24 hours. Just to keep the TLog at bay.

@TomV-TeamMonica Me neither

only just got back from a meeting with the tax department regarding the 99% ad hoc queries

@PaulWhitesaysGoFundMonica I defined it in my question as: In the process of implementing the SIMPLE Recovery Model in our test environment I thought about implementing a pseudo-backup procedure similar to the production environment instead. The idea would be to keep the user databases in the FULL Recovery Model and perform backups (preferably using Ola Hallengren's backup script) similar to the productive environment, which would retain the backup files for a maximum of 24 to 48 hours .

Paul White says GoFundMonica

3:53 PM

I've never heard that called a pseudo backup before

John aka hot2use

Pseudo-backup -> Clean up the TLog -> Dump backup files after 24 hours not required for restore.

The advantage: Database works and feels like a productive database with backup, but doesn't have to be restored like a productive database.

Well I could backup to NUL, yes. Good idea.

FDA Approved
MDR Approved

I need holidays

John aka hot2use

@McNets go for it.

@Johnakahot2use yes!!

@PeterVandivier interestingly his boss also used to write very short answers

Peter Vandivier

4:08 PM

he's gotten a bit snippy with me before when i've asked him to expand on things so i'm kind of just doing the leg work now to figure out why his answer is right and what links / code samples i can add to a supplementary answer before accepting his answer

like ... it looks like \dCS(+) lists "friendly" datatype names, but the underlying data comes from pg_proc -> pg_type & pg_type doesn't have those names

--echo-all doesn't reveal a SQL query like SMO->Profiler (in SQL Server world) does so i don't know where the word "bigint" is coming from when the underlying typname is "int8"

George.Palacios

@PeterVandivier well it's an int, and it's big

John aka hot2use

4:23 PM

That was fast.

dba.stackexchange.com/questions/254257/…

Peter Vandivier

ahaaaaaaaaa! echo_hidden <> echo-all. TIL

George.Palacios

4:53 PM

dba.stackexchange.com/questions/254263/…

SQL-server-2018

lol

5:07 PM

@PeterVandivier you can run psql with (--echo-hidden on): psql -E

Peter Vandivier

5:18 PM

@ypercubeᵀᴹ on the bus atm but is there a “discard result set” flag or something for when you only want the echo-hidden stream?

5:50 PM

jeez.. you can't be that young.. can you? :)
congrats, great article, lots of useful info!

Ugh, I only look that young. I'm in my thirties, and that's a recent pic :/
Glad you enjoyed the article!

Peter Vandivier

6:09 PM

30's?! geeze... ok boomer...

Isn't it the case that every partial backup is required to include the primary filegroup? Or am I misremembering? — Peter Vandivier 3 mins ago

am i misremembering that?

Paul White says GoFundMonica

6:45 PM

@Johnakahot2use Well they already asked on SO and it was closed stackoverflow.com/q/59070079

Peter Vandivier

🤨trying to decide what to select on the flag dialogue

1 hour later…

John aka hot2use

7:50 PM

@PaulWhitesaysGoFundMonica Ouch

@PaulWhitesaysGoFundMonica Did we get a "Not very welcoming" for that? I thought I had been quite friendly in my comment.

Paul White says GoFundMonica

@Johnakahot2use Sorry I don't understand what you're asking

Peter Vandivier

8:30 PM

@Johnakahot2use the OP was rude to you in a way that isn’t immediately clear to a non native English speaker. You acted properly

John aka hot2use

8:46 PM

@PeterVandivier must have missed that bit

Paul White says GoFundMonica

17

Q: What do the plans for incorporating "new ways to say thanks" look like?

A recent Tweet from @StackOverflow on Twitter said: Announcing our new reactions feature, available today on Stack Overflow for Teams for Basic and Business tiers, and coming to Enterprise in 2020. New ways of saying 'thanks' will be coming to the Stack Exchange network next year, too. ...

discussion support reactions-feature

John aka hot2use

@PaulWhitesaysGoFundMonica i commented his question here and was wondering uf my comment may have resulted in OP deleting his question

@PaulWhitesaysGoFundMonica 🆘

John aka hot2use

sorry crappy tappy

Paul White says GoFundMonica

@Johnakahot2use neither question has been deleted?

John aka hot2use

8:52 PM

Strange. I had a _This question has been deleted_
message after commenting his question. Strange.

Peter Vandivier

@Johnakahot2use OP did delete it for a bit dba.stackexchange.com/posts/254257/revisions

Paul White says GoFundMonica

8 seconds

Peter Vandivier

@PaulWhitesaysGoFundMonica can you see the exact timestamps?

Paul White says GoFundMonica

yes

Peter Vandivier

can i?

ah

yes

tooltips

Paul White says GoFundMonica

8:56 PM

hover over "occurred 4 hours ago"

right

Peter Vandivier

still though, the SO/SE caching layer is known to be aggressive

i feel like it's a fair assumption

Paul White says GoFundMonica

what's a fair assumption?

Peter Vandivier

that @Johnakahot2use may have seen the "deleted" status that was only "true" for 8 seconds because caching

it's when you assume something with a rational basis for that

Peter Vandivier

☝️also yes

Paul White says GoFundMonica

9:00 PM

@Lamak thank you

no problem

Peter Vandivier

user image

2 hours later…

Master Database

10:42 PM

0

Q: The forum needs a tag for Windows 2019 Server!

I have a question about my Database Administrators Stack Exchange post: MySQL Workbench, Windows Server 2019, MySQL not showing Performance, Information of MySQL Schemas! I tried to use the tag windows-server-2019 but it does not exist. I would like to propose creating it. I can't do this o...

discussion feature-request

« first day (3249 days earlier) ← previous day next day → last day (1605 days later) »

Transcript for

Nov '1927

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