When to cut your losses and reinstall 2000 server?
Posted on 2003-11-02
Here's the scoop - I'm the IT manager for a small mfg co. Until 4 years ago, our IT was focused mainly on IBM midrange environment - System 36, AS/400's. As more people needed to share pc resources (printers, files, internet connections, etc.) we decided to install a windows 2000 server (dell with intel hubs). My stepson installed it, and he did a good job - at least it worked well. At some point after the installation, the logs started showing 3 errors being thrown every 5 minutes. I'm at home now, but I believe the errors are the userenv and scecli (there are two occurences of one of them.) Additionally, I'm occasionally and sporadically losing connections to the AS/400 when using a telnet 5250 client or IBM Client Access. My environment is: Pulling fractional T (4 channels) through the channel bank/router to Intel hubs (2 daisy chained together, each w/24 ports) to Windows 2000 Server (or advanced server - can't remember). Clients include Windows 95, 98, XP, 2000, an AS/400 and a Linux box.
We are in the process of adding office space and users, and therefore wanted to get some of these issues cleaned up, and set myself up properly to move forward.........So, I decided I needed some outside expert help (seems logical !). In the last 3 months, I have had 9 people with 4 different companies (windows 2000 experts) come in to take a look. The first company came in, poked around, said they'd get back to me on the error messages - never did. The second company came in, poked around some more, didn't have a clue. I asked when it made sense to reload. They said they'd call the next day, never did. The 3rd company came in, we talked about the network structure, he heard we had a AS/400 attached and thought the network problems were AS/400 related - see ya, have a nice day. The 4th company was just in. They seemed the most knowledgeable. and were able to resolve the errors that I mentioned above.
HOWEVER..... When the 1st company was in, they updated the server to SP3. After they left, I started having problems with DNS - server and clients could not resolve names to internet addresses, but could access via their ip address - no errors logged though. I called them, and they walked me through checking services - found that Stopping and Starting DNS server service got things going again - for a while. I could actually make it fail, it I went out to a bunch of sites that I have never been to before, but still not error thrown. They said they'd get back to me - never did - I think they were waiting for me to sign a service contract - why would I do that if they didn't fix anything, and in fact introduced another problem. ANYWAY..........As mentioned, the 4th (last) company seemed to be more knowledgable, and were able to eradicate the scecli and userenv messages. The night before they were in, I was poking around the internet and found reference to DNS failing with no error, and it referenced a known bad DNS.EXE version - the one I had. They upgraded to SP4, and it got the new DNS.EXE. SO......They were able to fix several things, I couldn't force a DNS failure - things looked good. THEN.......soon after they left, clients computers would load IE and just appear to freeze when trying to go out to the internet. Stopping and Starting DNS resolved the issue, and I have even done a reboot, but as of Friday PM it still failed.
When do you finally said the hell with it and do a fresh install of Windows 2000 Server? I'm even willing to buy a new server - I need more capacity anyway, and need to buy more hubs/switches. Doesn't it seem odd to keep throwing man hours after it? What is the likelyhood of these issues going away with a clean install? Does anyone else wish Windows 2000 environments ran as reliably as AS/400 envireonments?
Any help would be greatly appreciated.