« first day (2534 days earlier)      last day (2338 days later) » 

1:08 AM
ugz does microsoft even test this code
 
1:33 AM
@sp_BlitzErik you ok? you sound like me
 
@JoeObbish there is something amok with the optimizer today.
 
@sp_BlitzErik "optimizer" as in you? or the query optimizer?
 
hrm. i think I see what's happening
SQL Server doesn't actually work on Linux at all.
Looking at the stack trace, it seems like they must have 0 working working copies of this.
strtol(0x7f991e01868b, 0x7ffc2ac14920, 10, 0) = 0x4088
free(0x7f991e018680) = <void>
fclose(0x7f991e011980) = 0
strstr("TracerPid:\t16520\n", "TracerPid:") = "TracerPid:\t16520\n"
so they're looking for TracerPid: then they find it, then they try to convert 16520\n to a long
and that call returns void, and then they try to free a bad address and it causes a core dump
 
@JoeObbish yes
 
well I'm happy to take a look at whatever it is
 
1:48 AM
@EvanCarroll you should get a job at Microsoft and fix it.
@JoeObbish it's unimportant, really
 
but is it INTERESTING?
 
Actually, I was looking at one of Joe's problems when I decided to fire it up.
I was trying to figure out if I could figure out what the partial index thing did by looking at the instructions executed
This code isn't actually generated by Microsoft =(
i'm barking up the wrong tree here.
unless they're explicitly checking to see if they're being traced.
 
 
1 hour later…
2:55 AM
@dezso That fits him perfectly so, yes, that might actually be him.
 
 
3 hours later…
5:30 AM
@JoeObbish I think it's odd you didn't discover the one below it (which does the opposite thing).
TFs are often arranged that way.
In other news: the weather is hot and Christmas shopping is hard
 
 
1 hour later…
6:41 AM
@EvanCarroll ugly
But then again nearly 7 years old. Would that be still available in current processors?
Good Morning
@AndriyM I'm very sorry it took so long to set your answer as accepted. Must have been dreaming...
 
7:00 AM
@hot2use you mean the Intel ME?
 
@EvanCarroll I meant the broken technology
:-)
The built-in feature
 
You can download the intel tool, compile it and it'll tell you if you're vulnerable
I am SOL.
 
Morning
@hot2use That's perfectly all right. If it was my answer that set you dreaming, I've got only myself to blame.
 
@AndriyM haha
 
@hot2use I thought you'd reply with "Keep dreaming" :)
 
7:14 AM
At the time I was just so happy it worked that I forgot all about going back to accept the answer/solution.
 
8:08 AM
morning and evening
 
8:44 AM
morning
 
9:00 AM
@EvanCarroll please see the edit here (I know you are notified, just want to explain): dba.stackexchange.com/q/174175/6219
in its original form it sounded like a too localized one, furthermore the link was broken
 
9:11 AM
@EvanCarroll ...there you go and Bob's your uncle.
 
 
1 hour later…
10:32 AM
I'm voting to close this question as off-topic because this is a cross-post of a Stack Overflow question. Please do not cross-post. — Andriy M 16 secs ago
Interesting problem, though.
 
11:28 AM
@AndriyM why that 1 + in your answer?
Also the two versions of the question (dba and SO) have different values in the wanted column.
 
@sp_BlitzErik eugdpr.org
 
12:15 PM
Hi . What's LHS ?
 
@ypercubeᵀᴹ It's how the logic works. The current attempt's number is how many non-IN_PROGRESS rows there were before this row plus one.
 
gbn
@RoyiNamir Left Hand Side
 
@ypercubeᵀᴹ The data samples are slightly different.
 
So every function become "uses indexes" when it's on the right side ?
NOT ---- > WHERE datediff(dd, senddate, @RunDate) >  @CalculationInterva
YES -----> WHERE @CalculationInterval < datediff(dd, senddate, @RunDate)
?
 
Of course not. It's just that the column (or the function applied to it) is more commonly placed on the left side.
 
12:19 PM
So how can I make this where clause "uses index" ?
Where..... CAST(lcprl.DateCreated AS DATE) >= @from ?
 
...particularly when the other side is just a value
 
The cast make it not using index . right ?
 
@RoyiNamir Put the function around the variables instead of the columns
 
You typically want to avoid wrapping columns in functions
 
> WHERE senddate < dateadd(dd, @calculationvalue, @RunDate)
 
12:21 PM
@RoyiNamir In this specific case lcprl.DateCreated >= CAST(@from AS DATE) will work just as well.
 
unless of course your query isn't selective enough, but that would at least allow the optimizer to consider an index
 
Thanks. But wouldn't cast(@from as date) will make it NOT use an index ?
(becuase cast is a function)
 
@RoyiNamir no you cast a variable, not a column
 
Oh right
 
so the cast doesn't have to be evaluated for each row
 
12:23 PM
Wait .... Datecreated is datetime . and I cast it to date so it will considerr the whole day.
my bad. that will work also
it has to be greater than cast a date
Thank you all
 
If @from is a datetime that doesn't have a time part, you don't even have to cast it as date
 
^right :-)
 
But if it's unknown, then doing the cast would be playing it safe.
 
C# screen sends my yyyymmdd so I thught to use a date fields ( not datetime) in sql. but now when I think of it - there is no need datetime will be fine with 00:00:00
 
@AndriyM the question in dba.se suggests the logic is slightly different.
 
12:26 PM
@ypercubeᵀᴹ I didn't notice, but I must admit I didn't read the description on SO properly.
 
(that the + 1 should be a + CASE WHEN status = 'IN_PROGRESS' THEN 1 ELSE 0 END
 
@ypercubeᵀᴹ I don't think so. There are two IN_PROGESSes in a row for A and for C. In each pair, each row has the same Attempt as the other row.
 
Yeah, I agree on that.
 
So it's the same attempt. As I said, I didn't read the SO description properly, but I saw that the desired results in both questions are consistent with the description on DBA.SE
 
Ah sorry. You have 1 PRECEDING. I thought you had CURRENT ROW
 
12:31 PM
Ah, that's what confused you. Yes, I'm specifically counting occurrences before the current row, in both versions
 
yeah. So you are correct.
 
Anyone done any quantum programming yet? blogs.microsoft.com/ai/2017/12/11/…
 
12:50 PM
@ypercubeᵀᴹ that was fast
alas, the user has not made it to SO yet
 
Marked as a duplicate in less than a minute, wow.
 
1:03 PM
You need to use mysqli_fetch_assoc instead of mysqli_fetch_row . You could have worked this out for yourself if you'd have done some simple debugging and var_dump()ed $row. Basic debugging is one of the most important aspects of programming, please learn how — Phil 49 mins ago
@Philᵀᴹ giving PHP advice on SO, hell has frozen over
 
He was tricked into making it look like that. The advice was given while the question was still on DBA.SE.
 
@AndriyM I know but still
 
1:35 PM
Hats start tonight (UTC)
3
 
@TomV hatters gonna hat?
 
does anyone know how much stack overflow enterprise costs? @bluefeet perhaps?
 
@sp_BlitzErik I don't know the specifics, but you can submit a request for more info.
 
@bluefeet thanks, but it's not for me. just looking for general pricing guidelines.
 
@sp_BlitzErik unfortunately, I don't know any pricing details
 
1:46 PM
cool, thanks anyway
 
@sp_BlitzErik i just use torbrowser or openvpn to a cheap VM somewhere
ec2 IPs are blocked by a lot of things you might want to circumvent but some of the smaller hosting companies aren't
 
33
Q: What is.... the Chicken? 🐔

Journeyman GeekSo, we just noticed a new, fowl user on the tavern I think he called us chicken. Or called itself a chicken. Or something poultry like that. And interestingly it's a system user. (i.e. with negative ID) So... what's the story? Who knows what fowlness lurks in the heart of the chicken? ...

@TomV are you sure?
 
says 10 hours and 3 minutes from now
 
thnx
It's on meta, too
quite early, this year!
 
they need to get it in before the gdpr shutdown
 
1:59 PM
@ypercubeᵀᴹ hopefully there's a multidimensional hat for you
 
the imge allows to click the lights on / off
 
@ypercubeᵀᴹ Yes I still have to work for the most part
 
gbn
2:13 PM
Anyone used CREATE DATABASE AUDIT SPECIFICATION to monitor any principal?
 
2:25 PM
@gbn nope
 
2:36 PM
@PaulWhite mfw when Paul already knows everything that you know
 
2:53 PM
Yo
 
@JoeObbish that would be my regular face
 
What does that mean? That Joe sometimes looks like Lamak?
 
no, I meant that "the face when paul already knows everything that I know" would be my regular face, implying that he knows everything
 
@Lamak mine too
but I was just trying to celebrate my achievement
then BOOM
 
3:09 PM
I'll just admit that my joke was very poor. I won't even bother explaining it, that wouldn't make it any better.
 
@AndriyM I liked it
 
@JoeObbish Thank you, good sir
 
his photo has a lot of orange in it after all
 
4:09 PM
@JoeObbish 9448 was new to me. Always test flags either side for related, often opposite, behaviour is my point.
 
cya folks
<3 to the World
 
@AndyK end of the day?
Oh joy my company banned imgur so most of the SE network is now useless to me... Any screen shots of things I can't upload or see when someone does. What a great company.
 
4:48 PM
@PaulWhite I'm going to print this out and frame it
 
@Zane what is it with people and banning things?
 
@sp_BlitzErik Well I work for Wells Fargo and they got caught robbing a bunch of people. So now they are trying to look like they are making changes. So they are cracking down on things even if it doesn't make sense to do so.
Although if I recall it wasn't an IT problem that caused a bunch of employees to rob customers but more of greedy CEO issue.
 
now they're robbing you of the joys of imgur
sad, really
 
5:16 PM
hallo my ppl
Is mastlog.ldf the wal?
 
but thanks, that's a good tip
 
Is sa specific to a database, or to the server?
 
the server
 
That's what I thought ;)
 
and yes masterlog.ldf is the WAL for the master database
 
5:28 PM
thanks
@JamesL do you happen to know then where the SA password is stored internally.
I'm assuming the ALTER USER [sa] thing is WAL logged even though I'm confused at why that would be the case.
I've verified that altering the [sa] user triggers numerious temp files but ultimately just writes to the wal.
 
you can see the hash in select * from sys.sql_logins
and the execution plan for that shows it's accessing master.sys.sysxlgns
which is probably only accessible over the DAC
i haven't looked into how it's stored beyond that
 
Yea, there is that mast.ldf and mastlog.ldf
So I guess those commands issue wal-logged txns to that which is just the system database.
 
yes it's just one of the system databases
what are you trying to figure out? just looking at things out of curiosity?
 
yea
total curiosity.
 
running this on the DAC should show you how it's stored in the data file:
DBCC TRACEON(3604);
GO
DBCC IND('master','sys.sysxlgns',-1);
/* For pages with type = 1
309 is page id in example
*/
DBCC PAGE('master',1,309,3) WITH TABLERESULTS;
actually you don't need the DAC for that, just for selecting from sys.sysxlgns
 
5:51 PM
@sp_BlitzErik You should get a job writing for the Daily Mail.
 
6:28 PM
@ConcernedOfTunbridgeWells does that mean i'll never have to deal with a database again? because i'm all on board with that.
 
6:46 PM
i'm not sure that password changes are logged for logins
you can look in the log for a database with select * from fn_dblog(null,null)
though it's more likely that those changes would occur in resource db than master
so if you wanted to change the password and check to see if it's logged at all, you'd need to be on the dac (like james mentioned) and looking at resource db
 
Would any of you happen to know any more details about what this non-yielding scheduler fix for 2014 was for?
I'm looking at a 2014 instance with non-yielding schedulers and lots of contention on the SECURITY_CACHE spinlock, but I don't know if that's what this fix was for
 
> Additionally, the SQL Server may crash.
i've done that a couple times and it's pointed me to kb articles about fixes
 
ah good suggestion
 
7:07 PM
i think the resource db is just for read-only stuff that doesn't change between different instances of the same version
but i haven't looked into that closely
 
you'd be surprised
 
stuff like password hashes seems to be physically stored in master
you can see it in dbcc page output anyhow
 
7:20 PM
This should have been closed as "not a real question" ;) — ypercubeᵀᴹ 11 secs ago
 
Question. We have a pretty large DB 9 TB in size and a system that isn't really capable of doing DBCC checks on that DB in timely fashion. So I'm building a system that will look at the average time frames set for that filegroup and build a list of filegroups to DBCC depending that will fit in that window. So basically and automated system for doing DBCC checks on a system that is too big for it's own good. Do you think folks would be interested in that?
 
@ypercubeᵀᴹ I WISH WE HAD THAT VTC REASON
@Zane that sounds neat -- wanna OS it?
 
@sp_BlitzErik Indeed.
 
@sp_BlitzErik don't we? ;)
 
It will eventually be all SQL and Powershell. Although my initial build will be SSIS because I'm good with that and can prototype more quickly.
 
7:25 PM
YOU HAD TO SAY THE POWERSHELL WORD
@ypercubeᵀᴹ i mean, SO literally has it
 
@sp_BlitzErik I could possibly do it in all SQL. Hard to say since I'm still in the prototype phase.
I was just thinking I don't see much online for that and I can't be the only person with this problem.
However I'd like it to be modular enough that someone could just download the code and implement it themselves and run it in any SQL Server environment.
 
you know, before you go crazy
i wanna say minion does something similar
 
lol ha. Of course minionware would have that. Why didn't I think to check them.
 
no one does
ahem
 
@sp_BlitzErik Really? I haven't checked, was it added recently?
 
7:30 PM
@sp_BlitzErik Unfortunately the fact that my companies only product is money that we use to make more money I'm likely to not be able to use that in spite of their impossibly reasonable pricing.
 
@ypercubeᵀᴹ no, it was a real close reason, but not anymore
 
ah
lame
 
yeah, it came with the whole "be nice" thing
 
Yikes. The features there are much better than the ones I was planning and for 250 bucks that's nothing. Don't get me wrong mine was going to be free but still. Dang.
 
7:35 PM
@Lamak yeah. Thnx Lamak and @sp_BlitzErik
 
Well thanks @sp_BlitzErik for the heads up.
 
check me out being helpful today
the end is truly nigh
 
Well technically you help me to crush my dreams of a new helpful tool so it's not all positive. hahaha
 
The worst SQL client in the world award goes to sqlcmd.exe
 
ohhhhh. You think. Teradata SQL Assist is probably the worst tool I've ever used.
 
7:49 PM
Microsoft develops PowerShell for Windows and then ports sqlcmd.exe to Linux as the "official client" this is the command-line equivalent of Total War.
@Zane I can't imagine worse.
 
do you really have to imagine?
i mean... c'mon
even pgadmin might be better
maybe
 
does it have the sunglasses?
no sunglasses? no wonder it's terrible
 
no, you have to wear your own
 
8:07 PM
PgAdmin 4 is pretty awesome.
This is a steaming pile of poop
I mean, it can't even handle statements.
 
what do you mean?
 
sqlplus is worse imo
 
I mean this GO crap.
No tab completion
not a single word of in-client help.
help, nope... show nope
nothing
And you can't even GO on the same line.
1> SELECT 1; GO;
2> GO
Msg 102, Level 15, State 1, Server x230, Line 1
Incorrect syntax near 'GO'.
^ are you kidding.
 
you don't need a semicolon after go
SELECT 1 GO; would work fine
looks like user error to me
 
No, that's not true.
1> SELECT 1; GO
2> GO;
3> GO
Msg 102, Level 15, State 1, Server x230, Line 1
Incorrect syntax near 'GO'.
1> SELECT 1 GO;
2> GO?
3> GO
SqlState 07002, COUNT field incorrect or syntax error
It has to be on its own line.
 
8:12 PM
what client are you using?
 
sqlcmd.exe
 
aren't you a real programmer? why is that surprising?
 
how is what surprising? I'm not surprised at how horrible it is, I'm just stating it as such.
And I do realize that even Microsoft acknowledges this, which is why the're pushing their new Windows Power Shell SQL Server cmdlit or whatever, which they don't make available in their Linux install.
 
welcome to v1
think of yourself as laika the communist space dog
4
you'll likely die in a fiery crash
but you'll always be remembered on corny tshirts
 
Laika took one for the Fatherland.
You can't fault her.
 
8:16 PM
yeah, i saw the consent forms
 
The dog never said "no". That's more than I can say for the women and girls who report interacting with the American bureaucrats and politicians.
 
don't forget actors and directors
and whatever comedians are
 
@sp_BlitzErik hey, don't talk like that about comedians....I thought that you identified yourself with one....
so, now that I think about it, carry on
 
Speaking of which did Alabama elect the child molester yet.
 
i'm not funny.
i'm not sure where alabama is
can you see it from texas?
 
8:25 PM
Somewhere in the US and A.
 
throw a penny
 
@sp_BlitzErik I don't have one...I can throw 10 pesos
 
then how will you pay rent?
 
I knew you were a comedian
@sp_BlitzErik though I own my apartment....it's small though
 
i'm gonna buy the building your apartment is in
 
8:29 PM
cool
 
then i'm gonna paint a gigantic yosemite same on the front
 
@sp_BlitzErik here it is:
 
do you have a balcony?
 
I do, but my apt is on the other side of the building
 
business in front
 
8:32 PM
I would think so
 
9:18 PM
I'm going to make a trivia game with Microsoft acronyms one day.
WBEM, ready go!
 
gfy?
 
0
Q: The catalog and system-tables tag?

Evan CarrollWe have two tags catalog and system-tables. I do not understand the difference between the two tags -- in part because in PostgreSQL pg_catalog is the schema for the system tables. But, I don't want to assume it's that way in the spec or the other database. However, if it is my suggestion that we...

 
@sp_BlitzErik you hating?
 
what do you mean?
i was just looking at your resource database question
not sure if you need an answer or not
fwiw, it's been in the same place (Binn) forever
though i'm not sure what the exact path to it would be in linux
 
9:27 PM
What do you mean? I'm just trying to learn. I don't need anything. Yes, I would like an answer.
 
it's not compiled directly into the database
 
@sp_BlitzErik is it still in Binn in 2017?
 
yeah
 
It's specific to Linux, but yea I'll look again. I don't think so in 2017. But this is your expertise not mine so I'm prolly overlooking it or something
 
the path looks like this on windows
Microsoft SQL Server\MSSQL13.NADA2016\MSSQL\Binn
 
9:29 PM
If these are questions that I know the answer too, I'll always self-answer them and not waste anyone time. =) I'm usually working on finding an answer when I ask it. And only about half the time I find the right one
Is that SQL 2017?
 
that's my local 16 instance
but it hasn't changed
 
Yea, I'm just not sure if 2017 is different here, but again I'll take your word for it. Scavenger hunt goes on.
Is there a method of getting the physical location name from the the name of the database
 
for "normal" databases, yeah, but resource is hidden in many ways
 
Hrm.
that's really interesting, sys.database_files show d:\dbs\sh\s17o\1116_092427_2\cmd\43\obj\x64retail\sql\mkmastr\databases\mkmastr‌​.proj\mssqlsystemresource.ldf
 
for instance, it has a database id of 32767, but SELECT DB_NAME(32767) returns null, where SELECT DB_NAME(1) returns master.
in which context do you see that?
 
9:36 PM
use mssqlsystemresource
go
select * from sys.database_files
go
 
I see and ldf and mdf to that virtualized directory
 
i wouldn't know how to find their non-virtualized location, if that's a thing
 
Currently, I think it's compiled in.
though it may be cute to try to supply an external copy.
 
perhaps that's the case with linux -- does nothing else come up if you search the file system for those names?
 
9:40 PM
nope
tried that first thing
 
wild
 
@sp_BlitzErik do you see it listed in your master_files
?
 
nope
 
What does your database_files show for t?
it?
 
nothing at all
 
9:47 PM
weird
Are you in the resourcedb?
when you read from sys.database_files
 
no, i'd have to sidetrack too much to do that atm
 
ah cool
Just today of all time.
 
interesting
 
 
2 hours later…
11:45 PM
lol they just forked and copied the postgresql shell
awesome job microsoft
 
11:57 PM
That's what open source is for ... Using OSS as intended.
Well done Microsoft.
2
 
I do agree. I can't think of anyone better to write code for future corporate products than those who did it passionately for their free open source alternatives.
 

« first day (2534 days earlier)      last day (2338 days later) »