packman

It'll be interesting to see if your problems are the same as mine (and a lot of others). See 10 topics back from yours. I've managed to get my two Win2K machines working as a peer-to-peer setup using Internet Connection Sharing but, because using a 56K modem, it won't work as slickly as with an 'always connected' modem. Despite what some people say, I think Microsoft's instructions for configuring for ICS is misleading.

Yes, SP3 has been out for, what seems to a die-hard Win2K person like me, a LONG time. I seem to recall it was finally released last June/July. As 'adam' quite rightly points out, the beta versions of SP4 are already in train. It's difficult to know whether this patch is worth adding. According to the MS KB article, ipnat.sys affects a number of network-related features, including Internet Connection Sharing, and that's something I happen to use. The patch is described as a "performance improvement", so I guess it's not going to "correct" a bug. If anyone knows more about network address translation and how this patch might affect it, I'm 'all ears'.

Does anyone visiting this forum truly know how to successfully configure two Win2K machines, connected as peer-to-peer, to connect to the Internet via a 56K dial-up modem using Internet Connection Sharing? I've tried configuring in all sorts of ways but the best I can achieve is a very unstable arrangement where static addresses 168.192, etc are used. Microsoft's publications seem to suggest that if you set up ICS in the Share tab of TCP/IP Properties, it allocates those particular static addresses, anyway. I call my two machines the gateway and the client. In order to send/receive mail from the client, does an ISP account also have to reside on the client? And does that account have to be installed to 'connect via a LAN'? I would have thought so, but nothing along these lines seems to work. I've tried automatic addresses and I've tried static addresses but neither setup works. The ISP connection on the gateway is configured for TCP/IP and is left to detect the addresses automatically. After several weeks work on this, I've practically given up, as at best (using static addressing) the client rapidly times out before the gateway's had a chance to dial out and, at worst (auto addresses), the gateway generates APIPA addresses (169.254, etc) but then cannot seem to find a server and nothing happens. I have to admit I'm totally confused by this. There's nothing wrong with the LAN itself, as file and printer sharing works perfectly. Also, the two machines 'see' one another when I activate My Network Places.

Well, thanks for those comments, Mezron. Since my last reply, I've swapped back to using auto addresses throughout. However, what happens there is that initially ICS causes automatic assignment of address 192.168.0.1 to the gateway (and thence the next address to the client). But when I reboot, I find (by running ipconfig) that those addresses are transformed into APIPA addresses and it's then impossible to dial out to the Internet. This seems to be confirmed by something I read in the UK PC World magazine, where under a section on Windows 2000 local area networking, it stated: "For small home and business networks, APIPA provides hassle-free automatic network configuration most of the time, but it can only be used on networks that aren't connected to the Internet". Today, I've changed back all my settings to static ones. With ICS still confgured, I find I can now use ICS with the DNS address of the client set as 192.168.0.1, which I couldn't do before. Mind you, this is now starting dial-up from the gateway, not from the client. Before, with static addresses, I could only ever download anything from the client machine by having the ISP's DNS addresses entered in the DNS address box of the client. That seems perverse, as why should external addresses ever work like that INSIDE the LAN? There remains the problem of spurious dialing out from the client. The only way I can think of stopping that is to disable Enable On-Demand Dialing, in the Share tab of the ISP connection. I've now done that. It's kinda contrary to the whole point of ICS but then, as I say, I've never been able to get ICS really properly working. I suspect that using a 56K dial-up modem has a lot to do with it. Instead of expecting to log on and log off the Net from the client, I'm resigned to doing that always from the gateway.

Mezron, Suggesting that the "connection is bad or hosed" doesn't really help, I'm afraid. Yup, I can share files, no problem. And I can print files from one machine to the other, no problem. Yes, I do run firewalls. However, I've run my experiments many times with the firewalls completely disabled and it made no difference. In any event, I've got the known range of LAN addresses configured into the trusted zone of my firewalls. Someone not in this forum, who's never personally set up ICS, has suggested that I might be misunderstanding ICS in that I'm expecting the client to be able to initially dial up the Internet connection via the gateway machine. It CAN do this, I've found (you can watch it happening), but the client then times out before it's had a chance to get online and download anything. So, have I been misunderstanding how ICS should be used from Day One? Should the gateway machine always be already online first? That would certainly fit with the more general expectation of employing DSL and cable modems these days (always on modems).

First, thanks DS3, but I've already tried out the info at THAT website and, frankly, it was useless. I've tried many others also but they too turned out to be useless. They're usually cribs of older guides on ICS, referring to Win98/ME. Some of them are just plain wrong. Mezron, The ISP isn't really relevant, just to say that I'm with a fairly reliable major ISP but certainly not one of the likes of AOL, Earthlink or whatever. I wouldn't touch those with a bargepole. I've been with my ISP for nearly three years and, on the whole, the service (subscription) is reasonable. My account, which I configure manually on my machine (the one I now use as the gateway machine or server), works fine and is optimumly configured as a 56K dial-up. I've configured the two machines for ICS precisely as you've described but the setup simply doesn't work. I've also tried an arrangement where I allocate static addresses and where I use the primary and secondary DNS addresses of my ISP but, although I could "kinda" get that to work, the client would still time out and it'd take several launches of IE from the client to finally download anything to it. Also, the whole static setup was basically unstable and the client would spuriously dial out via the gateway. I'm not using a hub, I'm using a crossover cable. It's definitely the correct cable. It took me ages to find an e-retailer with the one I wanted. It's by Belkin and it's even got "crossover" written on it. Also I've checked out its wiring. As I say, file and printer sharing works with no problem at all. Okay, the ICS software creates its own DHCP server, but where exactly should it register that server, ie in which tabs? Also, you say "If the address is 169.254x.x then look to communication between the two machines". What do you mean by that, exactly?

Just want to update something I contributed on 7th Jan. I've subsequently been in touch with Fujitsu and they utterly refute the press report to which I've referred. They sent me a standard letter in which they've stated that they've had no problems whatever with the MPG34xx series of hard drives and that there's no truth at all in the news item I'd read. Well, either they're the victim of a piece of rogue journalism or they're vainly trying to cover up a serious manufacturing fault by total denial. Take your pick.

Curley Boy, That hard disk of yours isn't a Fujitsu, by any chance, is it? I read a short article somewhere recently (Feb edition of PC Pro (UK), I think) where Fujitsu had taken CirrusLogic to court over this last year for being instrumental in the premature failure of thousands of Fujitsu MPG3xxx hard disks. Apparently, Cirrus reformulated the epoxy encapsulation for the drive's chips without telling Fujitsu. It seems the epoxy gradually eats through the dies themselves. This last year Fujitsu have left the ordinary consumer market, where hard disk drives are concerned, and obviously for understandable reasons. I've also heard some pretty horrible stories about Fujitsu hard drives generally, from contacts of mine. Fujitsu now concentrate on the true professional market instead, with much pricier products. Pity, really, as I have an MPG3xxx HDD myself! 40GB, and I've had it about 2 years, has always run fast and extremely quietly (fluid bearings), but I re-formatted the other day and I had real trouble getting it to wipe clean. So, can't be sure whether MINE's on its way out, to be honest! Must start thinking about additional means of backing up my partitions.

Tomay, Normally, I'd be the LAST person to gee up Microsoft but I was just expressing an opinion gained from experience with previous mice. I used to work in IT (the research side of it) and former colleagues who used to use Logitech mice have now switched to the latest MS mice, so I'm not the only one who's been impressed. Just because Logitech have produced some good mice in the past doesn't mean that ALL of theirs are good. An open mind is what's needed on these things.

Well, TheSam, if you've got Win2K, then if you're thinking of upgrading to a better mouse in the near future, I can't recommend the corded Blue Mouse from Microsoft enough (and I'm very critical!). By comparison, my old wheelmouse is so sluggish. In the UK, the corded Blue sells for about £24 including tax. There's a black one available as well. Although wireless mice are catching on, they have two disadvantages: 1) weight (they need batteries). 2) the batteries periodically need replacing and you have to watch for signs of malfunction. 3) you've got the transmitter and receiver to install into the motherboard and configure (BIOS).

Yeh, but that's a bit drastic for TheSam to have to do. He's probably got a tower and it's tucked away under a desk. Be practical!

TheSam, Which OS are you using? I think whether or not the light stays on depends on the OS (but don't quote me on that!). For a brief while I had a MS Wheelmouse Optical, using it with Win98. The ***** light stayed on and was really annoying. I went back to using a standard MS wheelmouse. More recently, I decided I really needed an optical mouse and, egged on by various contacts of mine saying that I'd have no trouble with the built-in light with the OS I now use (Win2K), I took the plunge and got myself a new corded Optical Blue Mouse. Its light does indeed go out! Furthermore, this Blue mouse is so much better, much lighter and faster to use. So, Win98 might be the culprit. Whatever, if you upgrade to a better mouse - the latest Microsoft range are as good, if not better, than Logitech's, in my view - I don't think you'll have that problem.

Two further questions: On the Host machine, is the default gateway supposed to be configured to be the same as the host's own address, ie 192.168.0.1? Should the DNS server address be 192.168.0.1 on BOTH the host and Client machine? Remember, this is for ICS running on the host machine and both machines using manual IPs.

Many thanks, Phalanx. I've followed up all of your suggestions and my setup is now ALMOST working. I've also gone to that website you recommended and have printed out the guides for configuring Win2K ICS as either auto or static addressing. Because it makes it more feasible to configure my firewalls, I've chosen to use static addressing. Printing from client to host definitely works with static, though I've found it'll only do it with 'NetBIOS over TCP/IP' enabled on the LAN. Now, bear in mind my setup - it's just two Win2K machines, one a host and one a client, each with identical Ethernet adaptors. The dial-up modem is attached to the host. What happens now with ICS is that I initiate IE from the client and, hey presto, the host dials up and makes the connection with the ISP. However, the client then times out, saying again that it can't find the server. Meanwhile (and correctly), the host doesn't download any homepage. However, I find that if I just refresh the search for the homepage on the client (by pressing Enter after its URL), it'll download to the client. So, I'm NEARLY there. (I thought that the DNS server address was one that you had to configure in the host and which was the ISP's DNS address. Was I wrong about that?). Also, I can't figure out why my e-mail client on the client machine is disabled for Send/Receive. In a setup like this, does the ISP account have to be installed on the client as well? I thought not. Incidentally, what exactly did you mean by 'uncheck Autodetect proxy server'? I know where that setting resides but, on that tab, there's: automatic detection of settings use a proxy server Which one of these should be unticked? Should the other setting there be ticked, or not? And should this be repeated on the host?

I'm not using mixed OSs. Do take seriously what I said about File & Printer Sharing. On the local network, it needs to ENABLED but, unless you're doing secure PC-to-PC Connection Sharing over the Internet, it needs to be DISABLED on the ISP Connection. Otherwise, you run the risk of someone on the Internet possibly wiping your hard drive clean or wrecking something else. It's easy to miss that setting, especially as Client for MS Networks, File & Pr Sharing, and TCP/IP look exactly the same in the components box in the ISP's properties and the Local Area Connection properties. Microsoft specifically warn about File & Pr Sharing on the Internet side.

Yup, I've had Client for MS Networks on both machines for ages and ages. Incidentally DON'T install or enable Client for MS Networks on the ISP connection, as it'll lay your network completely open to hackers, etc , who could completely wreck the contents of your drives. I'm not sure I've followed exactly what you've said in your final paragraph but if you're saying that I should try using NetBIOS protocol, then again, I've already done that, with no result. Don't assume that the configuration required for XP will be the same as for Win2K because it ain't. And Win95/98 are different again. For some hours of utter confusion, try taking a look at the otherwise very good networking website www.wown.com. For now, all I can say about my own problem is: Aaaaargh!

Your problem, which you say is now resolved, is interesting because I'm ALSO trying to set up a small network, consisting of just two machines, each running Win2K, and I'm blowed if I can get Internet Connection Sharing to work. No matter what form of IP allocation I try to use, the client PC just can't find the server when I try to dial up from it. My IP addresses check out okay. Did you find a crucial setting that did the trick? The LAN itself definitely works because the host machine successfully gives out the IP address across it, and also shared printing works. Any clues would be welcomed.

If your BIOS software is well-written, THAT's the most practical and reliable monitor to use for your PSU's supply voltages and the CPU core voltage. Third-party utilities can instead be unreliable. The manufacturer's data sheets for the motherboard components in question generally state that the respective voltage must not deviate up or down by more than 10% and it's my understanding that the ATX specification for the PSU uses that as the design guideline. In other words, all ATX PSU manufacturers should, in theory, adhere to that rule. Because motherboard components are continually evolving and consuming heavier currents, however, the regulation of those voltages on some PSUs might sometimes not be up to scratch. That's why there are now different total power ratings for ATX PSUs and you need to choose a particular power rating to suit your particular circumstances. As for the CPU core voltage, the ideal is, of course, to have it exactly right and, indeed, some BIOSs will allow it to be adjusted. However, for most Pentium 3s and 4s, it's not recommended to have the core voltage deviate by more than 0.1v (and preferably much less), as the processor might otherwise become unstable or begin to overheat. No doubt, some readers of this forum will cite instances of where a CPU has withstood harsher treatment but nonetheless the CPU manufacturer's recommended voltage should be adhered to, for guaranteed performance.

Dr Suds, Well, exactly which DirectX version did you install? There's been a DX8.1a and a DX8.1b this year, and as is now apparent, a DX9 in beta form. a and b have been quite reliable, from my own experience. The b version is otherwise known as DX81NTeng.exe and you should be able to download it from Microsoft's site. I know how flakey and troublesome DX8.1 derivatives can be, though. I recently installed 'Direct X' off a motherboard installation CD, thinking it was DX81NTeng.exe (I'd previously used that NT version and found it A OK) but the darned version on that CD wasn't at all compatible and now causes my e-mail client and mouse to arbitrarily freeze and I then have no recourse but to do a system reset. I'm planning to re-format shortly, as that's the only guaranteed way of getting rid of a duff DirectX. But when I reinstall everything, I'll make damn sure I use that NT version. Someone on the Net told me about a utility that's been designed to completely remove DirectX from Windows. It was called DX Exterminator and the program can be downloaded as dxe108.zip. Trouble is, I can't remember the name of the website now. Anyway, you could probably find it, if you're interested, by doing a search on Google or similar. Personally, I wouldn't trust a program to be that good, as DirectX is well and truly integrated into Windows.

Well, for a start, Presto Pagemanager is not compatible with Win2K, so why on earth you keep insisting on using it I simply don't know. InCD DOES work with Win2K, but you need to use at least v2.32. Go to Ahead's website or the Nero website and you can probably download a compatible version for free.

You must be mad, going back to Windows 98. It's a truly dreadful OS. I was glad to get shot of it. Windows 2000 is infinitely better. My conclusion is that you either haven't followed some basic rules for installing Windows 2000, chipset drivers, etc or you're trying to use applications with it that are not compatible. Just because your applications seem to work okay in Windows 98 doesn't mean that they'll work in Windows 2000. Frankly, the bundled software that comes with that scanner isn't worth the bother using. Stick to apps that are known to be compatible with 2000. The overriding advantage of 2000 over 98 is that 2000 is rock-solid stable. It won't keep crashing and, even if does, it'll recover easily. 2000 is also more secure.

Stoffer, I don't know whether this is of any use at this late stage but, before SP3 was issued, Win2K didn't support ATA100 per se. However, a patch for it was made available by Microsoft. The Q article on it might still be around at Microsoft's website. It was: Q260233 - "Support for ATA100 (Mode 5) in Windows 2000". The article starts "Windows 2000 does not support ATA 100 (Mode 5) for IDE hard disks. All ATA 100 IDE hard disks that are used with Windows 2000 default to ATA66 (Mode 4)". I didn't start using that Seagate ATA100 drive until SP3 was on the scene and, as I mentioned before, it was necessary for me to run the Seagate utility, anyway. So, if you're truly using SP3 and have also made use of that downloadable Seagate utility, you should be okay.

No, as far as I'm aware, all partitions should run at either the default UDMA speed or the UDMA speed that you choose when you run the manufacturer's utility. I run three partitions myself, at ATA100. Your test programs are no doubt faulty and misleading in the results they give. SiSoft Sandra is notoriously inaccurate for some things, at least the shareware version is. If the post-POST listing shows UDMA5 for that physical drive, you've nothing to worry about. Ensure that the drive is set up to UDMA mode in Device Manager as well. In Device Manager, it should be under: ATA/ATAPI Controller/Primary IDE Channel/Advanced Settings tab/Device 0.

Unless you can see, during the DOS-type bootup (that short sequence between POST and Windows starting to load), that HDD0 is already supposed to be running at ATA100, you DEFINITELY have to run that UDMA utility of Seagate's, as otherwise the drive simply won't operate at that speed. It's very simple indeed to run (provided you've access to an A prompt ouside of Windows; boot from a Win98 startup floppy). Indeed, at Seagate's website for these drives, there's a description of what to do. I had to do that for mine. It was factory-set to run at ATA66. I've also had to do that kind of thing for a Fujitsu drive that I use. Perhaps you don't have access to an A or C prompt outside of Windows because you've set up Win2K to run in NTFS, rather than FAT32? Hmmm. Perhaps you and others will begin to realise some of the advantages of running under FAT32, even though it seems old hat. If you can't get to an A prompt outside of Windows (these UDMA utilities are designed to work only in a DOS environment, usually), I suggest you e-mail Seagate for advice as to an alternative method to run the utility.

Stoffer, I've got that very drive, used with Windows 2000 and Service Pack 3. You need to download to a bootable floppy the special UDMA utility, from Seagate's website. You'll find it among the webpages devoted to that drive and similar IV series. This alters the firmware of the drive to make it run at the appropriate speed.