This post was edited - but the editor replaced the "sql" tag with "sql-server". It wasn't clear to me that the OP meant Microsoft SQL Server - so I rejected it. I'm just wondering if that was what I should have done - I'm 85% sure that it was the right call - any advice/opinions?

I'd have done a similar thing and probably asked the OP via comments whether they were looking for an RDBMS specific answer or a general.

good morning

phone calls you don't want to receive

"hi, our sql server had no disk space left and we couldn't start it, we removed some files and now our db is inaccessible, oh yeah, we shift-deleted the ldf files and our backups have been failing for weeks"

@Vérace that was the good call

there is absolutely no sign of any specific RDBMS

Morning all

@dezso I put the why of my rejection in my comment to the edit - so hopefully the OP will see it and respond if necessary. It was my first time rejecting outright - if it's a SQL Server issue, I normally accept (or skip if I'm not sure). It's the one major server that I don't run.

@PaulWhite I've tried to explain myself in the link on my SO profile.

zdnet.com/article/…

^^^ worth knowing about

even if they are over-egging it

@JackDouglas I think this is a known issue with the type of flash they use on high-volume drives like the Intel S3700. The flash can cope with a lot more cycles than standard NAND flash but the data is volatile if the power is left off. For drives using ordinary NAND flash (e.g. the S3500) this should be a non-issue.

Hello everyone!

Does someone here know a good mongodb tutorial that covers security features? The official documentation feels too fragmented to get a good grasp of if :/

@Gemtastic unfortunately there aren't too many mongodb users who are regular in chat, but it is the right place to ask :)

Yeah, I prefer postgres to be honest, but this is what I'm being paid to learn now so I'm putting down all my effort into it ^_^

@ConcernedOfTunbridgeWells I feel like I should have known that, and I'm guessing it applies to the P3700s in our servers too - though that's what I mean by over-egging - if it's just enterprise drives then they aren't likely to be in storage.

If you are using enterprise flash for archival you are doing something wrong :S

@Gemtastic I hope the compensation is high

@ypercube I think I'm good with getting paid at all :P

I mean; I could be sitting at home learning this for free too :P

Learning at least one NoSQL database is something I had my sights on anyway

I Can't Wait for NoSQL to Die

VtC if you've got a vote burning a hole in your pocket

@billinkc you're welcome

@ypercube Hehe

I'm too new to this to have any opinion though.

@ypercube or just settle back down into the niche where it is useful (and where it's been for decades)

@bluefeet You're the best SE employee in this room and in all of Arizona

@billinkc That's oddly specific

Kinda like "You're the best mother that is my biological mother in the whole world!" ;P

I'm rather odd. I blame it on my fascination with placing my head in my father's vise and seeing how far I could turn the handle

O.o

Well, whatever floats your goat I suppose :P

I'm quite odd myself ;P

Everyone is uniquely odd.

@billinkc I try

(:

@JackDouglas Yes, that's the point. Anything that's supposed to be busy with write traffic is likely to be powered.

Anyone have experience with Read_Committed_Snapshot?

Is this post for DBA.SE or SO. I would have thought that PL/SQL programming was within the remit of a DBA site? Opinions?

@Vérace pl/sql is good for dba but the question is very poor.

@ypercube But I don't think PHP is! I'm voting to close, only the poor guy was told on SO that he should come here...

@bluefeet what do you think?

They first posted at SO: stackoverflow.com/questions/31141366/…

What's the best fit, Superuser maybe?

@ypercube it's a pretty crappy question overall. I wouldn't migrate that anywhere

@Zane I love reading committed snapshots

No-one commits snapshots where I work

Not readable ones anyway

@ypercube best fit would probably be freelancer.com where someone gets paid to debugg for him ;P

In all seriousness: We use RCSI on several databases where I work now. We've been happy with it so far. We're about to enable it on another where it's more likely to cause pain, but less pain than the current situation with blocking.

We have another database where devs enabled snapshot (not RCSI) in a release without adding the necessary error handling to their code. Probably partly my fault for not forcing them to explain what they had tested. They're about to switch to RCSI instead soon.

Would it be fair to generalise that enabling RCSI will inevitably increase load on tempdb?

I really wanted to enable RCSI on the data warehouse at the last place I worked, but I didn't think TempDB could take it.

@AndriyM I think so. Although it doesn't always matter.

@JamesLupolt How more likely is it to matter if we don't have tempdb on a separate drive?

@AndriyM I wouldn't think a separate drive would matter as much as having fast disks. I don't think I have a good general answer to that, though. I like to use distributed replay to test such scenarios.

@JamesLupolt And I'm probably not asking the right questions either. I guess I'll need to read more about the matter to try and get a clue.

dba.stackexchange.com/review/suggested-edits/48930

What to do with this? The 'Internet directory' sounds a bit underdefined (and, at least to me, the wikipedia definition sounds much better).

@dezso Wasn't sure about this so I elected to skip - I think I'm just seeing these now because I just passed 5,000 points. Would be open to guidance as to what to do - I thought it was a bit light.

@Vérace most tag descriptions are not exactly PhD theses

@dezso Not even a basic degree by the looks of it! :-) I'll take a look round the tabs - haven't really done much of that (edited 1 so far) and see what the generally accepted standard is. Thanks.

@dezso Seems to be from MSDN

Okay so I've got this query that is producing a poor plan and is taking forever to run.

It's using date as the predicate.

Here's the weird part. If i send the date back it returns faster.

6/28/2015 <---- runs well over a half an hour.

6/2/2015 < under 30 seconds.

@AndriyM then the source has to be shown

Does anyone know how this could possibly be the case.

With an index hint it runs in 11 seconds

Is it using an index that it shouldn't? Any chance stats are stale?

Well they were updated the 28th.

OHHHHHHHHH shit!

Our Clustered index is set up in such a way that it only get's 4 Histogram steps.

That may be the problem right there :)

Is the 6/2/2015 at the top/bottom of the hist range?

ohhhh man.

That's no bueno.

Just killed a SPID from SSIS and then SSIS just went on to the next step....

@Zane SSIS gives zero f*cks, it don't care.

Apparently.

Things are getting crazy go nuts.

@Zane do you have a diversity of data? 4 steps isnt alarming if you have an even distribution of data. it's called "histogram consolidation"

sqlmag.com/blog/how-many-steps-should-histogram-have