« first day (1863 days earlier)      last day (2990 days later) » 

12:12 AM
How do I post pics from my phone
 
12:54 AM
I noticed that for my version of android, doing a long press to paste for chat is broken
I'll click the chat area for focus and when I long press to bring up the paste window, it brings focus back out of the the chat pane
 
 
7 hours later…
7:59 AM
@MDCCL ^^^
 
8:44 AM
@MDCCL Thanks, though it didn't precisely answer his question I wanted to steer his thinking a bit using other wording than the theory he just learned and was confused about
 
 
1 hour later…
9:57 AM
@swasheck "World" champs
4
 
@Phil exactly
 
Day 2 of Exchange Server downtime here :D
 
@Phil You guys need more cloud, that never fails
 
Death by virtualisation
 
@Phil I got my first real IT job because the company needed something to replace an Exchange instance that run out of licence (too many users)
so you can also set up a postfix ;)
 
10:34 AM
-3
A: Can SQL Server Audit be used to track hostnames or IP adresses of the users performing DML on a table?

Ruggiero LauriaAt that point I decided to use DDL triggers, which unlike the DML, have not a significant impact in terms of performances. First of all we have to understand which events intercept, using this query you can have a hierarchical view of available events: WITH EventsReports (name, PARENT_TYPE, type...

I'm not going mad and this IS answering a completely different question...right?
 
@MarkSinkinson Right, you are not going mad.
 
First time I notice that on MSDN (the stackoverflow link)
 
 
3 hours later…
2:12 PM
Am I the only person who doesn't understand the 'compute top and bottom cumulative values' window frame optimization? sqlmag.com/sql-server-2012/…
Worth asking a question on the site?
 
2:30 PM
@JamesLupolt I think so
I look forward to Paul's essay :)
 
> A NoSQL programmer walks into a bar. He sees two cute ladies at a table. He can't join them.
 
@JamesLupolt I think it means something like: For:
SUM(val) OVER(PARTITION BY actid ORDER BY tranid
                        ROWS BETWEEN 24 PRECEDING AND CURRENT ROW)
it does instead:
SUM(val) OVER(PARTITION BY actid ORDER BY tranid
                        ROWS BETWEEN UNBOUNDED PRECEDING AND CURRENT ROW)
and then calculates the diff between the SUM for this row and the row with -24 offset
 
A tester walks into a bar. He orders 1 beer, 0 beers, 99999999999 beers, -1 beers, A£"%"$TS beers ...
666;drop table beers...
 
3:00 PM
@ypercubeᵀᴹ Thanks, I think I get it nowᵀᴹ.
 
3:13 PM
More from the Interwebs...
Java is like Alzheimers
It starts slowly, and then takes all of your memory
3
I'm here 'til Friday. Try the fish.
 
4:09 PM
It's March 1st and the first day of RDBMS school. The teacher starts off with a role call.

Teacher: Oracle?
"Present sir"
Teacher: DB2?
"Present sir"
Teacher: SQL Server?
"Present sir"
Teacher: MySQL?
[Silence]
Teacher: MySQL?
[Silence]
Teacher: Where the hell is MySQL?!
[In rushes MySQL, unshaven, hair a mess]
Teacher: Where have you been MySQL?
"Sorry sir I thought it was 29:99:-1, February 50th"
 
4:19 PM
@Phil yeah. i've never quite understood that
 
The other countries aren't even trying
 
@swasheck I don't use mobile chat so I'm not sure. You can ask on MSE?
 
@bluefeet why are you punishing @swasheck? ;-)
 
@bluefeet yeah. what'd i ever do to you?
 
4:43 PM
@AaronBertrand are you around?
 
Busy arguing about what is a catch and what is the definition of is (what is a performance issue). What's up
 
I'm wondering if there is any place I should look to speed up my sql sentry app.
It chugs like crazy. I'm pretty sure it's because it live updates and my company encryption is blocking it but I have no clue.
Yup. Pausing sync fixed it nvm.
No wait it's still happening.
 
No idea, would have to get support looking and possibly tracing the information that is coming down to your client and what might be slowing it down. Sorry, no magic answer for you.
 
Gotcha. I'll have to do that then.
 
If your network forces encryption on every byte then yeah I suspect that would slow things down and unfortunately I don't think there's anything we can do about that.
Support would suggest either (a) turn off encryption (maybe give you a special dedicated VPN without it) or (b) put a client on the other end, if a server in the data center can talk to the monitoring service without encryption, then the lag would be in the video / mouse / click delays going over the encrypted connection. Pick your poison.
 
4:54 PM
That's why I'm trying to turn off auto sync
 
@AaronBertrand still?
when even the "experts" dont know what a catch is or isnt
i'm not sure you can settle it with anyone :)
 
Plus a lengthy discussion in twitter DMs with Kenneth
 
5:13 PM
@AaronBertrand sorry. i'm a bit dense :)
@Kin ... sorry if it felt like i was contradicting you. i was trying to support you with additional evidence here :)
@swasheck absolutely agree with you. Valid points .. Thanks ! — Kin 42 mins ago
 
Well, is a horrible tag
 
wow. a few questions regarding facebook-terminated service
 
Kin
5:37 PM
@swasheck All good and I agree with what you wrote.
 
<poof>
 
Kin
@AaronBertrand thanks
 
> An intoxicated passenger, who has reportedly brought his own alcohol on the plane, became disruptive -- prompting the flight crew to tell him to calm down.
sweet
"SQL Operating System" — Aaron Bertrand ♦ 1 min ago
Save Our SQL
 
5:53 PM
Oh no a leadership survey. I'm swimming in corporate jargon.
 
Nope now I appear to be drowning.
 
6:38 PM
/sigh
Ok, so if you already know your priorities, why are you asking us (a community of peers with different priorities) an opinion-based question? — Aaron Bertrand ♦ 5 mins ago
 
The best design is to go home and think about what you did. And stick with MySQL until you're ready to accept your design is terribad
 
@AaronBertrand because I wanted you to agree with me duh!
 
6:53 PM
 
Signing off. Nincompoopery is off the charts today.
I mean, there have been worse days, but the combination of stress from deliverables and not wanting to address stupid questions is peaking
 
7:09 PM
> A SQL programmer walks into a bar. When asked what his order is, he selects every drink in the place with "star" in the name.
 
7:27 PM
Should anybody have a strong SQL Server opinion about this.. justsayin
3
Q: DL380 G9 which processor and what impact on NUMA

Tom VWe are about to build a new server and I was wondering what would the impact on NUMA with the options available. The system will be built using a Proliant DL380 G9 8SFF which has 2 sockets, and 12 memory banks per socket. The question I have is what would be better configuration 2 CPU's with 6 ...

I pondered if I should post that there and what the SQL community there is, but it would have been off topic here I guess
 
@TomV i think it fits either location
 
@swasheck I figured it was a bit of a shopping list question and hardware advice is explicitly on topic there
 
@TomV yeah, but depending on how the answer is fashioned, it could certainly be useful
@TomV anyhoo ... what's your workload going to be?
 
To be honest, I think the box is oversized, 100 users for an ERP system, RAM should be more than enough and cores will be too much is my guesstimate, but it's a new implementation so there are no figures about current workload
So I figured less initial complexity, plus the added flexibility of a free socket should calculations be wrong sounded like a good idea
@swasheck current workload is a custom program running on intersystems caché, new workload will be an off the shelve ERP so apples and oranges I suppose
 
7:47 PM
primarily OLTP? OLAP? ferraris vs. trucks
 
OLTP, only one database on the instance backing an ERP system (dynamics AX)
 
i'd probably stick with a single-socket fastest conforming cpu (without having researched the exact model in your question)
foreign memory penalty should be fairly low, at this point ... but i'm not 100% sure on that
 
The idea is high clock speeds yes, and I wanted to avoid things like cache floods on one node or things like that leading to variable performance
 
 
1 hour later…
8:55 PM
man. i thought mongo was a silver bullet
2
Q: Sudden Mongodb high connections/queues, db stops responding

Camilo López A.The issue We have a strange issue on our mongodb setup. Sometimes we get peaks of high connections and high queues and the mongodb process stops responding if we let the queues and connections increase. We need to restart the instance using sigkill using htop. About the software environment Th...

 
9:33 PM
@swasheck Right to the face.
If reliability is your werewolf.
 
10:10 PM
@AaronBertrand is it bad that this pisses me off so much?
> "Improvements in cardinality estimation – Clearly the SQL server team has some work to do to further tune the new cardinality estimator and they are all over that."
"clearly the VS team has some SLDC work to do"
 
10:44 PM
@swasheck No, that pissed me off too. If they had a massive query regression on a vanilla query, ok, maybe they have a point (putting aside the whole "test before you deploy" thing). But they had 5 or 6 things in this query that are bad ideas in general, never mind combining them all.
Blaming the cardinality estimator for a bunch of shit you spackled together because you knew it only worked while touching your nose on a Tuesday when it is raining outside...
 
If it were vital the plan always had a certain shape, instead of using every hint they could find, they could have used a plan guide. I very much doubt the desired plan shape has a sort at all.
 
I very much doubt that team knows the desired plan shape. :-)
 
i was at a loss for words as to how to respond
so i never did.
 
Now, I do want to stress that I give them a lot of kudos for being so transparent. They could have made up any old nonsense to excuse the outage, and everyone would have bought it.
 
but, in all honesty, it feeds so many of the "bad developer" stereotypes that exist
@AaronBertrand good point. but it does seem a bit disingenuous.
 
10:55 PM
@AaronBertrand Well exactly. The mess-o-hints just screams "panicked newbie" to me.
 
"our processes are broken, but the underlying tech was what really failed us"
 
@AaronBertrand Sure. It just the code! Reminds me of the AdventureWorks design too much perhaps.
@swasheck Default Blame Assignment
 
heh. fair enough. and i'm used to it. but it's just so weird watching it happen publicly
 
I suppose it's not really that surprising that expert users write better SQL than the people that write the product.
 
Definitely some bad things in that messaging, but some good things too.
It's the right direction, if not perfect yet.
And I've said publicly multiple times that there's no excuse for the irresponsibility of not regression testing before hitting that switch in production. That's just so many levels of wrong.,
So I'm certainly not trying to defend the whole thing. I'm just a little nervous now that all of the backlash - and I've definitely contributed to it - might make them a little gun-shy about being so transparent next time.
 
11:10 PM
One would hope they would be able to distinguish between the two separate things: happiness over the transparency; shock at the quality of the code.
 
@AaronBertrand i get your point ... but is it truly transparent if you're not able to correctly address the facts? i dont disagree that it's a step in the correct direction, but it'd be nice to know that they're not just pretending :)
 
@swasheck well I do think they were led to believe that this regression shouldn't have happened regardless of the noise they added to that query. So while many other things could have prevented this in the first place, I don't think they're being dishonest that they blame the estimator, and I even agree that the estimator could be better at this.
 
Really they should have just asked on Database Administrators :)
 
@AaronBertrand ah. fair enough.
@PaulWhite screw those guys. they're jerks.
@AaronBertrand definitely could be better. just wait until 2016 ;)
 
@swasheck Yeah I guess.
 
11:14 PM
I think they put a little too much blame on the estimator, but...
 
@PaulWhite except you, of course
 
@swasheck well, that's where it gets funny, because that system actually runs on a much more recent version of SQL Server than sites like the Register etc. reported. They all said it was SQL Server 2014 but they were wrong.
 
@swasheck No I'm one of the lead jerks on dba.se
 
@PaulWhite yes, #2
 
Ha ha ha
 
11:15 PM
@AaronBertrand sorry. i should have put that in quotes, but that's interesting
 
The Register "article" was the one that irked me. Stupid click bait title, inaccurate content. The other stuff was great, just lacking detail (plans!).
 
@PaulWhite should have gone straight to The Heap for shenanigans, mocking, ridicule, and (finally), an answer
 
RAM Gobbling Bug!!!111!!! wasn't it?
@swasheck Correct.
 
I wanted to call that out in my blog post, because I didn't want panic for people on 2014 in production and using any or all of those hints. But I can't say for certain 2014 wouldn't have been equally affected, so I held back.
 
the should have just used mongo. duh.
 
11:19 PM
OMG
0
Q: Call UDF on linked server with parameters

user3735855I am using SQL Server 2012. I have a @variable that I can need to populate with data from a linked server. This @variable is then used inside a Table Valued Function. My code works as it should if I execute it actually on the linked server itself, but I really need the script to run on a dif...

 
@AaronBertrand stackoverflow question: expected behavior.
 
Ok I need to eat before I look at that again, else I'll lose my appetite
 

« first day (1863 days earlier)      last day (2990 days later) »