Monday, October 11, 2004

To Rebuild, or Not To Rebuild.....

Problem: Once in awhile you come across a machine that is SO screwed up that it doesn't pay to figure out what's wrong. Unfortunately, you might work on it for a long time with that I-can't-let-this-damn-thing-beat-me! attitude before you realize you're past the point of diminishing returns. Not very cost effective.

Solution: Set a time beyond which you will wipe the system and rebuild it from scratch. Easy, eh?

It's wrong to call this the "solution" because it might not be the solution for you. It's certainly not the only solution, but it could be one that would work if the conditions are right. Anyhow, you have to know enough about your operating environment to make the call on this. That is, you must know enough about the machines you support, the applications they run and the network they are in to know if/when/how to stop recovery efforts and start rebuilding it from scratch.

For purposes of this example, let's assume that the machines in question are all workstations (as opposed to servers), and that the workstations run no critical or peer-to-peer shared applications. Let's also assume that all the workstations run roughly the same set of standard software, like an antivirus package, MS Office, etc. One further assumption: each workstation has at least one network drive where the user can/should keep all his or her working data, even if s/he also keeps a lot of the current data on the C: drive. Using these assumptions, you could say I will spend no more than 1 hour solving a problem before I begin the rebuild process. After all, the rebuild process under the given circumstances could reasonably be done within 2-3 hours (depending on hardware and software configurations). The process could be something like this:

  1. Save all the user's data on the network drive, making sure to check the desktop, My Documents and folders and files at the root of the C: drive.
  2. Save the user's local email data, such as mail stored on the C: drive and the address book. You'll need to know enough about the email client in use, and the location of the local mail data.
  3. Save the user's internet bookmarks.
  4. Save the machine's network configuration (write it down).
  5. Save the machine's security configuration (write down users and groups that need administrative access).
  6. Save the machine's printer configuration (note all the defined printers and make sure you can physically find them).
  7. IMPORTANT - If the machine is relatively new, record the device information, especially for the network device, in case the rebuild process doesn't find the drivers you need.
  8. Restart the system from the OS installation CD, delete the existing partition and rebuild the OS from the bottom up.
  9. Resolve/install any missing drivers that the OS installation didn't find.
  10. Reinstall all the standard software, beginning with the antivirus package.
  11. Re-enter the network information and download any current upgrades from the LAN or the internet.
  12. Redefine the users and groups that need administrative access.
  13. Have the user logon, then restore the saved bookmarks.
  14. Add the printers.
  15. Restore the user's data from the network drive.
  16. Restore/reconfigure the user's email settings.

It sounds like a drastic measure if you're not used to doing this, but it's really not. It can take a lot less time than working on some nasty problem for too long and then having to rebuild the system anyhow. You just need to be ready for it by having all the installation CDs in your bag of tricks. Even more time can be saved if the users are informed about keeping their data on the network drive(s) at all times.

It works for me....


No comments: