@MDMarra Finally.

nothing critical no, just a few gotcha's that you can easily avoid

some group policy settings are removed from 2012 (most notably the IE maintance stuff). What we've done is keep the 2008 r2 group policy templates, and have a dedicated 2008 r2 server to do GPMC work on

Now people can stop feeling like sysadmins just because they can open a command prompt and run dcpromo...

also, the KMS host on 2012 does not support Windows 7.. Best choice is to keep a 2008 r2 server for KMS purposes (and just add new packs to it for win8/office2013)

lol. Now they get a gui. O.O

Good to know about KMS. Seems odd that would be the case

at launch it didn't even support office 2010

@MDMarra Valley - "When you get a chance can you please send me the printer set-up instructions. I have another HP LaserJet 4200n at 192.1.2.123 that needs to be set-up."

192.1? That's not rfc1918 space...

@DennisKaarsemaker I know :(

produce

why are all your customers so fucked up

I know i've asked before

no IT people.

but you're like flies on shit.. where do you find these

they all came from a job I worked from 2002-2008.

I managed the Linux application server that ran their businesses...

but now I'm trying to fix issues beyond that.

and undo the damage from lots of bad consultants.

man, the 2012 GPMC is golden

it's wonderful to force gpupdate at a OU by just a right click..

@pauska That's in RSAT8?

@jscott yep

Got check that out.

oh ncie

the only reason I run win8 at my desktop at work is the new server manager...

(and of course all the rsat tools under it)

you just launch it with your admin account, and then you have all the tools listed + a nice dashboard

I have my test Win8 b, but haven't transitioned yet.

@MDMarra No $$ until this is fixed!

It begins

@ewwhite whats fixed?

Should a non-technical user know how to configure printers in an AD environment?

@ewwhite Non-tech? no.

agreed..

in this crazy setup @mdmarra came up with, print server is on one VM... but we're deploying printers via GPO...

They'd need to be an [delegated] admin on the print server to create the port/queue

that is not crazy at all

so the client admin would need to add the printer to AD.

we're doing it the same way

(I know it's not crazy)

As are we.

but this is the type of place where I catch them trying to load print driver software from the CD.

you don't want non-tech people to start editing group policies

tell them that this is billable work.. and that it has to be done by you

Limited users won't be able to install drivers unless the device class is approved in GP.

in the past, they installed their own printers...

but it also led to a messed up setup

@ewwhite That's not surprising. Was everyone also a local admin?

so is it fair to put a barrier there and say, "you need to call me for every printer addition"

@jscott it was a windows 2003 terminal server with ~85 users...

so the local guy just added printers directly to the server

(and used the default names, too!)

Ha! HP LaserJet 4250, HP LaserJet 4250 (1), HP LaserJet 4250 (2)...

@StackExchange stop f-ing around @mdmarra and write my documentation!

So you guys almost have all the requirements hammered out?

@jscott I improved names slightly...

but it's still pretty terrible...

@ewwhite I wrote that last night. I actually have 2 DBAs here trying to figure out Oracle BI right now

trust me, I'm not having fun

"Oh, just send the jobs to Konica 222"

@MDMarra ugh, okay...

so should I try to teach someone how to do this, or just say, "it's GPO, you need ME to do it"?

Just say they need you

Is that normal?

GPO is dangerous in the hands of a noob

I've never had non-tech people manage a print server or GPO

/me goes 'Duh!'

I only just realised that you wrote computer accounts

It is a good day not be be logged in as an admin. (for me)

Not sure how I missed that before. You even BOLDed it

@Hennes I won't say who put our SCCM server in the Domain Admins group when they were "Testing".

Testing is different. I have done insane things just to see if it would work.

But I make a clear distinction between testing on test servers, or live production

@Hennes The quotes implied a testing SCCM server on the production domain, I may have omitted that part.

I guessed as much.

Right, shopping, then fixing VMs with wrong partitioning.

(insane large root, oh, do not worry. Everything will fit in the root)

Except that is doesn't.

@Hennes BRB, filling /tmp

Remember to flush

@ewwhite didn't I tell you that earlier? :P

@pauska maybe... but this is one area where I feel awkward telling a customer NO

really

so if you sold cars to them

you'd be ok giving them instructions on how to change the settings inside the cars CPU?

cause this is nearly the same thing.. GP's are dangerous

unlink the wrong GP and you've suddenly lost your domain (like removing the link to the default domain policy and losing encryption keys)

I mean, for me... windows printers mean, identifying the device, giving a proper IP address or hostname... (many of these printers now have DHCP excludes because they were never given static IPs, or their IP's are right in the middle of the DHCP range)

downloading the right (non-universal) driver...

installing 64-bit and 32-bit drivers.

associating the port...

then going to add the GPO printer entry

huh, a sql server log file can have the .mdf extension? weird shit

@pauska you're right.

how do you phrase it?

@ewwhite you forgot to customize the default print settings (duplex, b/w by default) and setting security permissions if you need one

Do you want the performance issues of the last terminal server? Because using it also as a print server will definitely degrade performance

yeah, on their copiers, I have to disable duplexing.

just because they were doing to before doesn't mean it's worth continuing to do

Chat on an iPad sucks. It inserts weird stuff into the next message if you start typing right after you send one

You're spelling like a monkey..

that explains it

so, "Sorry... with this new server setup, the process of managing printers is more complex and involves several additional steps. To ensure consistency and performance, I'd prefer to handle printer management."

I'd say ...involves several additional steps that will ensure that we avoid the performance issues of the last server.

Let them know its not complexity for complexity's sake and that there's an actual benefit

True :(

How do you teach people proper rack unit numbering/spacing?

You see the issue there, right?

hah, someone made and hung up a HAPPY 1 YEAR ANNIVERSARY DAVID! sign above my desk @ work

@ewwhite "Ok Jimmy, stay within the lines. Just like when I give you the coloring mat at Friendly's while we wait for our meal!"

Like, um... things don't fit right...

@ewwhite You don't :) We had a network upgrade throughout our district a couple years ago. I can't tell you how many things are 1/3 or 2/3 a U off.

some people actually think that leaving a gap between improves cooling

that servers get less hot if they're not lying on top of each other

I put gaps in at my colo... but full 1U gaps...

(for cabling purposes)

Anyone know what happened to that fancy Microsoft Group Policy search site? gps.cloudapp.net

@ewwhite whats wrong w/ this picture? not knowing anything about racks/racking

@David 1/3 U spacing...

@David every device is in the wrong holes

i have no idea what that means

Everything should be screwed in at 1U intervals. If you're going to leave a space, leave an entire U

@David 1U = 3 holes

1U = 1.75" vertical

for what purpose, if you're not going to completely fill the rack

at one point in time you are in fact going to need to insert <something> between two other things

Things stop fitting right...and also when your company says "woah, we just landed a huge customer" and starts buying more gear, you don't want to kill yourself

@David What's the point of doing something wrong when it's just as easy to do right?

also, americans needs to stop using threaded hole racks

it makes jesus cry

@pauska Drill 'em out!

drop gallons and miles and am/pm when you're at it

@ewwhite HISS

@pauska in telco racks, it's okay

@MDMarra if anything, it looks nicer @ 1/3U?

threaded hole racks are Satan's sweaty nutsack.

@ewwhite no, it's not. all it takes is one monkey to use the wrong sized bolts and you're fucked

@ewwhite UNACCEPTABLE. Use cage nuts like God and Nature intended.

@David the point is that the holes aren't spaced evenly...

@voretaq7 I've only seen threaded 2-post relay racks.

@ewwhite I'll take a picture of our rack the next time I'm in the datacenter

4-post threaded-hole rack, no U markings, and the previous shithead couldn't count to 3 so theres a machine stuck between 2 U boundaries (with cross-threaded screws)

@ewwhite gotcha

@voretaq7 well, there's no excuse for threaded 4-post racks ;)

@David first of all, it's not spaced the same. every 3 holes is 1u. If you do it off-center you wind up not being able to screw things in properly. Second of all, pretty doesn't matter. Functional and correct matters.

There's a reason things are standardized, like the size of a rack unit

@ewwhite Punishment

FedEx just arrived with a box of 1ft power cables!!

Pretty only matters after you've taken care of functional and correct.

@ewwhite YAAAAAAAAY!

Sting them together to make a 6 foot cable!

Yep. It's time to clean up the "functionally-correct" mess I made at my colo.

at least you're not using cable management arms

did you buy cables with locks on them?

@pauska I will on the server I'm moving today...

but it's like the one acceptable case for using a cable management arm.

@ewwhite An IBM BladeCenter?

(those actually have GOOD cable management arms - they don't block airflow)

The Sun x4540.

top-loading disks...

so the mgmt arm actually has to work.

Don't spill your soda on that.

@ewwhite and the rails need to be long enough to clear the front of the machine above it (including locked bezel cover and pull rings)

yeah, totally... this was purpose-built...

Nice enough that I've walked smart hands through drive replacements

(although, that server was designed to have several spares... and for you to just leave failed disks in place)

@ewwhite How long do you leave them?

I had 4 hot-spares in that setup... and replaced once two failed.

global spares

this is going to be the foundation for a backup service I'm starting... so I need to go retrieve the server and move it to new colo

and meter power VERY closely...

@jscott I like that GAM thing. Nice.

@Tanner I wish there was a gAPI module for PowerShell.

Honestly you probably just saved me hours of troubleshooting.

Haha surprised it helped someone so quickly. When I Googled it all I found was this poor fella: productforums.google.com/forum/#!topic/apps/uhw4kl01EH4

Oh, that reminds me. It might only affect 2003.

Regardless, thanks. It's one more thing to keep in mind if/when we start testing GAPS.

who here knows Hadoop?

@ewwhite I know it sounds stupid!

client spec....

@ewwhite want a great business idea?

@ewwhite "1. No technology with a stupid name"

4x 8 core CPU's
1~1.2T Memory
17x 1T NL-SAS Drives (non-raided)

for hadoop

what would you use?

@pauska sure.

@ewwhite Cheap, disposable compute nodes

@voretaq7 look closer at spec...

@ewwhite convert your customers to hyper-v 2012, and sell them DR space at your colo, by using the built-in replication.

quad-socket...

1TB+ RAM.

just have vpn tunnels to all of them so that you can actually run stuff for them if they need the DR one day

I'd use cheap, disposable compute nodes :-)

@pauska I do that now... with VMware...

the whole point of Hadoop is distributed computing isn't it? :)

@ewwhite vmware costs money! :)

if the individual nodes need to be beefy I'd get Dell systems though

1TB on hadoop nodes...

decent warranty, pretty cheap in quantity...

are you taking the piss?

@voretaq7 right... so the issue is 17 disks + lot of RAM + 4-socket.

@DennisKaarsemaker If he's doing protein folding or complex fluid dynamics models...

a DL580, for instance, only has 8 drive bays...

even then, with hadoop you go for more boxes, not bigger.

@ewwhite well the issue is they seem to be specing the whole damn cluster as if it's one machine

Distributed Computing - they're Doing It Wrong

right... so the design is messed up, right?

@ewwhite unless they're doing something where the individual data chunks are INSANELY huge, yes

find out what you need more, disks or cpu. Then buy nice cheap supermicros with either more disk or more cpu

We have a file server running Windows Server 2008 Standard, every few hours the server is inaccessible. (i.e. map drives don't work, you can't browse to it) but you can ping it and it resolves by DNS. Any ideas what it could be? We've checked all the basics like DNS, duplicate IPs, etc Rebooting the box allows the server to work for 2-3 hours and then the same thing happens again. Banging my head against the wall here.

(also the whole "non-raided" thing is bogus: Even in a Hadoop cluster you'd want RAID so you don't lose a node if a drive fails)

devopsreactions.tumblr.com

@DennisKaarsemaker you know about my Supermicro issues.

no raid in a hadoop cluster. Just make your boxes trivial to build and make hdfs store multiple copies

While transmitting or receiving data, the server encountered a network error. Occassional errors are expected, but large amounts of these indicate a possible error in your network configuration. The error status code is contained within the returned data (formatted as Words) and may point you towards the problem.

^ My lizard tried to do that once

The Hadoop design should be like ZFS... SAS HBA's.

he was young, and didn't realize that he's a desert gecko and doesn't have sticky feet.

it was hilarious watching him slide 4 feet down a wall kicking his feet trying to climb it though

@Cole physical or virtual?

HP has a "big data" server now... SL4500

@voretaq7 upside down like the .gif? :)

@Cole NBT being retarded?

@ewwhite they're doing it wrong. Scale out, not up.

@DennisKaarsemaker or would you use a bunch of these?

We're currently building a hadoop cluster using smaller supermicro nodes

@DennisKaarsemaker no

hp dl160 was the right size, but not cost effective

I was trying to make him eat antibiotics -- he was, shall we say, less than happy about it

@pauska physical

@DennisKaarsemaker if they could live with lower CPU or RAM, I'd use a bunch of DL180's.

but it's the RAM+CPU+disk count that seems off

and I'd hate to approve any more Supermicro gear.

@Tanner you may be onto something

@ewwhite Why?

Just buy the mid-line stuff with IPMI

Oh, forgot to mention: This W2K8 server is on an NT Domain it has a trust to our domain.

@voretaq7 because I also have ~12 DVR solutions at produce sites that are Supermicro...

and even my wife comes home and says, "Supermicro is crap!"

(she works in produce)

@ewwhite mmhmm, and is it the hardware or the software she hates?

failed backplanes and constant drive failures due to poor chassis cooling

across 3 nodes.

@ewwhite hmm, I've not had a single failure in the last 4 years

maybe we're doing it wrong...

I've never had a supermicro fail on me

@Cole any errors on the switch port statistics?

No wait, one failed raid controller

@pauska no errors

@DennisKaarsemaker a have a couple of thousand of them at work...

they have some issues.

however, I don't doubt that people can get them running well.

@ewwhite We had a couple of thousand Dells at $job[-1], they had "some" issues too

well yeah, with a couple thousand of anything you're bound to have some issues

we have the same with hp

(like "guaranteed hard drive failure within 30 days")

early g7's were a complete disaster for instance

Just odd we can still ping it - but we can't connect to it

supermicros were only in the dozens here, mostly for storage and faxing

@Cole tried restarting the "server" service and see if that fixes it?

@DennisKaarsemaker wut? I don't remember that..

2k supermicro server with 50k worth of diva cards :)

@pauska we've only restarted the server completely - I'll try the server service next time it pukes. Any idea what that would point to?

@Cole well mate.. NT Domain in the year 2013.......

@DennisKaarsemaker the benefit is the flexibility to build shit the way you want

@pauska Oh trust me, I'm with you there.

@pauska oh yeah, they kept dying without a possibility to recover. Had to be sent back. The may firmware update fixed it.

@Cole I am pretty sure Microsoft would laugh in your face if you called..

@pauska lol exactly, just told a manager that. They were like "fuck"

File Server Resource Manager failed to enumerate share paths or DFS paths.  Mappings from local file paths to share and DFS paths may be incomplete or temporarily unavailable.  FSRM will retry the operation at a later time.

Error-specific details:
   Error: NetShareEnum, 0x80070842, The Server service is not started.

@Cole It's such an easy upgrade too! Do it man

Best IT policy: Just rawdog it

FSRM and DFS on a NT domain

dude, I'm not sure if this is even supposed to be working...

@pauska lol right?

But notice, the Server service is not started error.

Wonder if the server service is just puking.

ok, I need some advice here

beer or scotch?

@DennisKaarsemaker Budget?

@Cole check if it's set to restart on recover

@Tanner it's in the fridge (hertog jan) or on the shelf (aberlour 10) already :)

@pauska restart for the first two failures

@DennisKaarsemaker Oh, then both. =D

Interesting enough - the dependencies for the service are "srv" - which keeps bombing out on the NT DC's every so often

@Tanner Then I'll make even more of an ass of myself in here :)

But if it has to be one definitely the scotch.

@pauska are you going to help be setup a Hyper-V cluster?

@ewwhite help you how

What kind of jobs did you guys have while you were in college? I'll be moving to either Chico or Monterey in August for school, and kinda worried about not being able to find a legit job in cities so much smaller than Sacramento (where I live now).

@David full time desktop support/jr sys admin role

@David sysadmin at ISP, club DJ, orchestral clarinetist...

@Cole full-time? how

@David work all day, class all night/weekends

@pauska I don't know a thing about it

@Cole eck, not in the plans

worked 8-5PM M-F and class 6-10PM M-F

That's why I dropped out

lol makes sense

currently I work 3 days a week, 5 hours a day

and just work around regular classes

gives me enough cash flow to have fun

but that'll run out quick once i have to quit here and move

HI all!

@ewwhite sure mate, just ask.