« first day (2936 days earlier)      last day (1924 days later) » 

1:10 AM
cotd. Let's talk about Paul White (@SQL_Kiwi). Dude has 50,000 EXP. Gave away 33k. Actually he gave 33,450. The difference between 33,000 and what he actually gave away, over 4 JON SKEETS. https://dba.stackexchange.com/users/1192/paul-white?tab=bounties #StackExchange #StackOverflow
 
 
6 hours later…
7:31 AM
Morning!
@EvanCarroll He can edit your post and remove the downvote
Nice pic btw :D
 
8:12 AM
Morning
 
Morning
 
Morning
 
8:35 AM
Morning
I need an adult - can someone help? Some of our new SQL Server Azure VM's have been built in the wrong collation - would anyone recommend a rebuild of the system DB's or is a clean install better?
 
@George.Palacios sounds like... youtu.be/6Ls5j5iz2eA
 
@PeterVandivier The Azure gallery images come pre-installed with SQL_Latin1_General_CI_AS with no option to change on build. It's very irritating
But yeah I agree
 
8:55 AM
out of curiosity, is the collation difference causing observable problems for you? or just a splinter in your mind that it was deployed off-spec?
 
9:06 AM
@PeterVandivier It's causing a stored procedure to fail due to use of Temp tables
 
womp womp
so - not _quite_ what I was asking about a few weeks ago, but in case anyone missed the Brent O post, some highlights from PASS for the lame-o's who missed it (like me)
https://www.pass.org/marathon/2019/dataexpertseries/DataManagement.aspx
 
9:53 AM
@George.Palacios I did several rebuilds in the past.. nothing too complicated tbh, just follow msdn instructions. I would do it if the instance is not in use yet.
 
@Marian Yeah that's what I'm thinking too - the instance is preprod so I've just taken it offline for users
@Marian Never had to do it before though so hopefully it goes okay!
 
save your settings (in case you have any custom ones) and permissions, in case it's a system with many logins, and then give it a go
 
10:13 AM
@Marian Luckily it's an AlwaysOn cluster so I'm just going to use DBATools to copy all the logins, jobs etc from the secondary
 
10:57 AM
@George.Palacios you're aware of the slack channel, right?
suuuuuuuper friendly / helpful
and afaik - always happy to hear about production implementations of the module even if everything goes smooth
 
@PeterVandivier Yeah - spoken to Chrissy Lemaire a few times and a few others have been very helpful. The tools are just outstanding
 
She's like... maximum friendly
i'd be afraid meeting her in person would be like encountering the Care Bear Stare
 
@PeterVandivier Anyone who is a Powershell MVP scares me a little
 
new task for the weekend: photoshop the powershell logo onto the blue care bear
3
 
11:29 AM
@PeterVandivier Called Chrissy LeBear?
2
 
 
2 hours later…
1:24 PM
Hello! Greetings!
Any tips on how to alias table name through out the session? (Not throughout a query)?
For sql server in SSMS
 
Does it need to be session bound? (IE Not affect other sessions)
 
IF yes better, but otherwise also okay. But for the later the original table name should also be accessible.
 
I don't know a way to make it session bound, but you could look at using a synonym
 
maybe a synonym?
 
Ah thanks. I saw/read in a video somewhere but I was searching it with 'ALIAS' in google and getting all the query related results.
 
1:27 PM
Ah yeah I can see that
 
That reminds me:
ALIAS is an alias for SYNONYM – hmm, I guess you could also say ALIAS is a synonym of SYNONYM... — Andriy M Jan 14 at 21:58
 
:D
 
@AndriyM But is synonym an alias for alias?
 
Perfect. Made my life much easier now. I'm dealing with some really mean and horrible table names.
 
2:25 PM
so... i just said a really bold thing in another chat...
> once you go multi-master, you're kind of accepting eventual consistency somewhere in the chain
...and... i feel like there's probably important exceptions to that.
someone wanna call BS on me?
or is that fair?
 
@PeterVandivier did you mean eventual consistency issues?
 
i mean multi-master needs conflict resolution rules
and if you return success to a closed loop, only to have it be a "valid" path to later pry that loop open...
or retroactively invalidate it
it's violating isolation, right?
 
I'm missing some context I think
I think I see what you're getting at though
 
not really missing context, we weren't discussing a particular platform - i was just kind of throwing my hands up as to why multi-master kind of derailed the conversation
like - you have to start with the requirement that your system is either ACID or BASE
 
No I mean I don't understand what the "closed loop" is pertaining to
 
2:33 PM
and i'm asserting that multi-master is fundamentally BASE
closed loop as in the application
 
Hmm
Is SQLite BASE?
Because that is kind of multi-master
in a weird way
 
Peter (New York) deposits $1
Peter (Hong Kong) withdraws $1
what's Peter(Global) balance?
one of those ATM's is gonna show a wrong value
 
By Multi-Master I assume you're talking specifically about a Server-Client architecture?
 
multiple replicas writeable
 
RAC for Oracle does that and Oracle is ACID
 
2:37 PM
I've been trying to make headway in my understanding of this topic. I think I've found this most helpful so far: aphyr.com/posts/313-strong-consistency-models
 
Who knows what rules they've has to break to do that though
 
I was told there's a central lock manager for RAC
 
It's not my area of expertise by a long shot
 
@Forrest i think of quorum lock release as not partition-tolerant
 
@PeterVandivier So I guess your argument would be that it breaks the A in ACID?
 
2:42 PM
was focusing on the "C" or the "I", really
 
@PeterVandivier I think I meant the "I" actually
Well yeah I suppose logically one replica has to either have an idea of what the others are doing, OR rules which change the way a query is handled based on what another replica has already done
 
@George.Palacios 😝 then yes
 
So I agree on at least a surface level
@PeterVandivier On top of that though, are you judging that ACID needs to apply to a singular replica, or the system as a whole?
 
i kind of think i might be, yea
vis-a-vis writes
 
@PeterVandivier That would be where my main question lies in that case
Does a transaction that hits a replica always need to apply ACID, if the system as a whole is ACIDic?
Said ACID too many times now. Lost all meaning
 
2:51 PM
chat follows the SILLY consistency model
4
 
hehehehe
 
i try not to get SALTY about it
 

« first day (2936 days earlier)      last day (1924 days later) »