cluberti

I'll have to try it next week when I'm back in the office, but I'm wondering if you could get a few dumps of it either consuming a lot of VA or using high amounts of CPU (or both) so we could take a look.

You didn't mention OS, but the instructions are basically the same for all - see if you can get a "hang" dump of one of these two apps following these instructions stickied here.

You're running a beta checked, unoptimized build. Expect RAM usage to be high. It's checked (debug build), constantly logging, and it's very likely completely (or at least compile time) unoptimized, so it's going to be really "heavy". If you want to see what I mean, download a checked Windows build and a retail Windows build of the same OS from MSDN, install both, and run perfmon - especially x64, the load can be almost 100% more (in fact, I've seen worse with VS betas) with the checked build than the optimized retail release. Not really - they're the same as forefront uses, and those are updated daily (sometimes more frequently, depending on what happens that day). I don't know if you'll find these available on WSUS in the future as the forefront defs are, but it's not impossible to imagine this may happen in the future after RTM (maybe not, but there are other "consumer" things in WSUS, so I guess anything's possible). However, the defs release at least daily.And to the OP, if you find a bug, don't just say "don't run this beta software!!!". Report the bug, otherwise this could make it into the shipping version. That's why they do betas, remember? Being an SQE you of all people should know this .

. 0 Id: 1a20.1890 Suspend: 1 Teb: 7ffde000 Unfrozen # ChildEBP RetAddr Args to Child 00 00139608 7739bf53 7739610a 0004001c 00000001 ntdll!KiFastSystemCallRet (FPO: [0,0,0]) 01 00139640 7738965e 00030094 0004001c 00000001 user32!NtUserWaitMessage+0xc 02 00139668 7739f762 77380000 06475188 0004001c user32!InternalDialogBox+0xd0 (FPO: [6,1,4]) 03 00139928 7739f047 00139a84 001c1dc4 00000000 user32!SoftModalMessageBox+0x94b (FPO: [1,165,4]) 04 00139a78 773d7e85 00139a84 00000028 0004001c user32!MessageBoxWorker+0x2ba (FPO: [1,78,4]) 05 00139ad0 779f1d6a 00139b58 7fa268b4 77988b88 user32!MessageBoxIndirectW+0x55 (FPO: [1,19,0]) 06 00139b94 779f3ca9 00000000 0013af94 00000000 shdocvw!CDocHostUIHandler::ShowMessage+0x10f (FPO: [3,43,4]) Thread 0 has a dialog box (visible or otherwise) that it's waiting for you to interact with. It looks like jscript just went through garbage collection, and I do see unloaded java .dll's on the stack, so it's possible that the garbage collector cleaned up a dereferenced script object that the Java engine still thought should be valid. The reason I believe this to be the case is this: 10 0013b42c 0207b900 0207b830 00000000 7fa81e95 mshtml!CMarkup::SendNotification+0x7a (FPO: [2,17,4]) WARNING: Frame IP not in any known module. Following frames may be wrong. 11 0013b438 7fa81e95 0013b48c ffffffff ffffffff <Unloaded_Eng.dll>+0x207b8ff 12 0013b4a4 76fa0336 00039800 00039800 025ceb28 mshtml!CElement::Notify+0x686 (FPO: [1,16,4]) 13 0013b4b4 76fa04cd 0013b564 0013b558 00000000 jscript!GcContext::Release+0xd (FPO: [0,0,0]) 14 0013b798 020e17b0 020e1890 00000000 00000000 jscript!NameTbl::Release+0x58 (FPO: [1,0,4]) 15 0013b7c4 7faadefd 001a7358 00000000 00000000 <Unloaded_Eng.dll>+0x20e17af 16 0013b7e8 0013b960 020e1710 00000000 00000000 mshtml!CTExec+0x40 (FPO: [6,0,4]) 17 0013b868 7c827d29 77e61d1e 000005c0 00000000 <Unloaded_dll>+0x1145a0You can see jscript garbage collection on frame 14 as a result of something the unloaded java module did in frame 15, hence the CTExec call in mshtml - since we can see jscript invoking the GC right after the java module does whatever it is it did, it's VERY likely it dereferenced some script object and GC is cleaning up after it, which is expected. After this, mshtml notified the java module in frame 11 of what happened, and it does appear that the java module in frame 11 didn't like what it was told, because on frame 10 you start going into the notification routine to display the dialog box you ultimately see causing the hang. I'm taking a wild guess that you don't see the dialog box, because the process isn't hung, but if you can't see the dialog it would definitely appear so as the UI is hung waiting for the dialog to go away. In layman's terms, something in general appears to have misbehaved (again, likely some java code here), and IE has been directed to present (and ultimately did create) a dialog box from this notification, and is waiting for you to acknowledge/clear it. It's a modal dialog, so it's going to get created on the thread it was called on - and the UI thread in a native Windows process is thread 0, so it's going to block/hang the UI until you interact with it (that's the difference between modal and modeless, amongst other things). I'd need to see the dump to know what's in that message box (it'll be in memory in the dump somewhere), but if it only happens in this one app it's likely the app has a coding error (the callstack seems to indicate that is the case, but again with only a stack to go on this is just educated conjecture).

MUI packs (the full MUI pack you seek, not just the small language pack you've been using) are available to Microsoft customers with an account agreement (volume license or premier), you won't find them publicly (at least legally posted, anyway) on the internet. If you need one to apply a MUI to an English 2003 install, contact your Microsoft account rep and they can provide these to you.

Note that if these are domain-joined machines, you can put a default user profile in the netlogon folder in the domain sysvol and that works as a "profile copy" as well. Has worked since NT4, I think, so it's at least an option.

I know, but it came from the same exact search on the same exact site. I'm pointing out that just because you go to Google and search "msfn" doesn't mean Google's search engine did the exact same thing in all scenarios (in fact, as I pointed out, it doesn't).

The real fun is IE8 vs Firefox from the SAME machine making the SAME SEARCH:

If you want to be sure you're installing as admin, either log on directly with the local administrator account, or right-click on the installer and select "Run as administrator" from the list.

It's ok - we try to take cases on a case-by-case basis and judge the contents of a post against the rules. As to the EULA and licensing, regardless of whether or not the admins and mods here like or dislike the limitations/restrictions/etc of the companies involved, we still respect their rights for operating within the laws of theirr countries and international law. Again, it doesn't reflect our personal opinions or our views, just the laws and us making sure that this forum (which is hosted in the US, also) follows applicable local and international laws as much as possible.

Well, considering the mark of a good rogue app is that the user doesn't know it's there, I'd say any machine seeing the internet (or even other networked machines) should be scanned at least once in a while to make sure it's actually as clean as it feels .

Something to shoot for

Seems so. Please attach your last_session.ini (not copy/paste, but attach) and see if someone can help you here. Moving to vLite.

The error you're getting is the one you'd get from a server or client if someone was already logged on. You shouldn't get that locally on the client itself, although it does beg the question did you use vlite or something else to modify the install?

Yeah, but how expensive are those disks, really? So it's 16¢ a GB vs 8¢... I guess if you're on a tighter budget two drives are definitely a cheaper build than 4, but if you can afford 4, it makes more sense to do that than say RAID6 or RAID5 with a hot spare, or even RAID0 over 4 drives (at least the 10 has some redundancy).

And it's not that hard to get Hyper-V IC's compiled and running on Ubuntu 9.04 either, just fyi. It's not as automagic as it would be on SLES, but the benefit is not having to run SLES .

You mean, perhaps, something like the sticky at the top of the section you just posted in called (oddly enough) "Creating memory dumps"?

Are you trying to log in via RDP to a Windows 7 machine (I'm assuming Pro or Ultimate) from more than one location?

No need. FAT32 can handle drives up to 2TB, but the Windows disk management and format utilites are limited on purpose in formatting removable USB media. Use a 3rd party format tool, like the HP USB formatter tool found on the 'net, and you can copy a source as large as you want to a USB key and it will work, formatted FAT32. Not actually true, it's the media. If you do it from a DVD, you are correct - but if you use a USB key (as I've been harping here lately, for some reason ), there is no size limit. In fact, I used oscdimg to make a 21GB ISO of a source with all sorts of WIMs of differing sizes, 2 of them 5.1 and 5.2GB respectively, and all work fine from a 64GB USB key formatted FAT32 with the nt60 bootsector (obviously I don't have optical media that will hold 20+GB of data). I also have a 18GB WIM (huge OS and software load) that works fine from the key as well, and works fine from PXE boot. It appears to be an optical media install problem only.

If you use a USB key to install, and it's formatted FAT32, there doesn't appear to be a size limit to how large your source can be - it's only a problem for creating an ISO.

Better is a tricky word. Assuming I'm a (regular) home user and I don't live on MSFN and am not technically knowledgeable, what is "better"? The ability to get newer peripherals and hardware with drivers that support my OS natively? The ability to get newer software versions and perhaps newer software in general for the OS because it's still supported by the vendor? Better security from the kind of stuff they'll come across on a regular basis than the previous version(s)? The ability to call the vendor for free if the product breaks for the first 5 years of it's life? Since most people only upgrade their OS when they buy another machine, it's really a moot point anyway, and most people will just learn the new Windows because they got it with their new computer, so they'll have relegated themselves to the fact that it'll be a bit different than their "old" computer - people managed (easily) to switch from the 3.1-style program manager to Windows 95, which I would argue was the largest ever UI shift in Windows to date, so I don't think people will have too much trouble with 2000, XP, or Vista to Win7 as some people predict. Most folks aren't technically savvy, but they're not generally stupid either. If people can go from 3.1 and DOS to Win95 without totally giving up Windows, they'll go just as easily to Win7 without any problems either. For a business, what is "better"? Power savings due to much better power management than previous versions of the software, which can be a 10+% portion of a company's operating budget, depending on size (the larger the company, the more $$$ it costs to run their hardware)? Having one version of a software package (including the OS) to support across all machines, with lots and lots of management options to increase security or make the software easier to control, cutting support costs? Reduced management overhead for deployment, having one master image that can be rolled out in all languages to all hardware platforms versus having a separate image for each machine type difference, and another for different languages, etc? The ability to patch and keep an image updated *offline* without having to install each and every image in the org, update the packages, and re-sysprep it (which can only be done once or twice anyway safely)? Staying within the 5 year support window for bugfixes, keeping support costs down when a vendor issue *does* start costing you money? Those are just a few reasons. I can think of a few in favor of staying with an older OS for each too, but most of those come back to being able to run hardware you already have (although Win7 mitigates that a bit by running on older hardware much, much better than Vista did, sometimes as well as XP does) or you plan on not upgrading software, don't need vendor support, and in general might not like the newer versions from an aesthetic perspective (bloat is genrally only complained about by folks who want the newest and latest to run on older, smaller hardware - smaller HDDs, RAM sizes, slower CPUs, etc). Windows never has been, and likely never will be, smaller with each successive version - because the vast majority of consumers and businesses who buy a new product from Microsoft want more features, more security, or both. Both of these generally require more code, although Windows has always been the "feature" OS, and in general that's what drives sales. People don't get all warm and fuzzy about security, but they do about shiny new eye candy and things like the taskbar dock - so, Windows grows. It's always been that way, and it's likely never to change - it makes money for Microsoft, and they're a for-profit business. As long as it continues to be profitable, and shareholders will approve, it'll continue to be modus operandi. And honestly, I'm ok with that. It's also important to remember that the whole of Microsoft Windows is not just the NT kernel, just like Linux is not just the kernel in most people's eyes. Both are very small, and the "bloat" comes from all the software (including the windowing engine - the windows shell in Windows and Gnome/KDE/xfce/etc for a Linux box) layered on top. In fact, I just installed desktop ubuntu 9.04 with all the defaults, and it took up just a shade over 6GB on my HDD. For reference, Win7 took just under 8GB. Three words - flat memory model. Faster? On relatively modern hardware, most definitely, as it's doing less with less code to run. Stable? Depends - 98SE was very stable for me and I'm guessing most others, but 95 wasn't stable as soon as you introduced newer USB devices, and 98FE was about as stable as WinME for most people, and it took an "SE" to fix it. But, and this is the problem I have - flat memory model. Any running app, anywhere on the system, can write to any other app (or even the kernel)'s process space. There's zero security, at all, in the system. The second problem I consider major is dual and multi-core CPU support (none), and memory limits (2GB at best). I do lots of code work (dev and test) and not being able to test code to see if it's threadsafe, or waiting for things to compile on one core with limited RAM at this point is a dealbreaker in and of itself.

This also works with vb script: http://www.activexperts.com/activmonitor/w...s/#MoveFile.htm

It takes approximately 15-16 seconds from power-on to usable desktop as a Hyper-V VM - it's a single processor 512MB machine, and it has no A/V or any other driver than hardware drivers for the virtual hardware. It's used for running VS 2008 off network, so no domain login either.

Agreed - the behavior seems more network related than server related. I do recommend managed switches, and also using Multicast vs Unicast (no word on which you're using) if you plan on having multiple users imaging at the exact same time.

For display? No, not at all. Unlike the spooler, say, there's no wow64 for running 32bit code in 64bit - it'll fail.