In-Depth

Exchange 2000 Upgrade, Times Two

It took three separate attempts to finally upgrade my company to Exchange 2000—this after doing the upgrade in a lab environment and upgrading a pilot box with the IT department on it...

After extensive testing, we started the upgrade.

During the first upgrade attempt last November, I received an error message stating that there wasn’t enough space for the database to upgrade, despite the fact that I had about 25 percent free space available. Strangely, it gave me the option to continue anyway. I called tech support (which was no help, as they told me that they wouldn’t do it if they were me.) We decided to delete the information for some people who were no longer with the company, defrag the database and retry. No go. We quit at midnight.

In December we bought a new server, installed Exchange 2000, and moved the 20 people with the largest mailboxes off the first server and onto the new server.

Also in this issue:

 Get Active Directory Replication Right!
by Andrew Lindley

 Wireless Meets Mother Nature
by Justin Melot

 The Expiration Date That Did Us In
by Jeremy Dillinger

 Troubleshooting Under Pressure
by James D. Pollock

 Hard Drive Fall Down, Go Boom!
by Christopher M. Roscoe

(Back to introduction.)

In January, with plenty of free space on the server, I ran the upgrade. With no explanation, the upgrade died in the middle, leaving the database in an inconsistent state, and with some pieces upgraded to Exchange 2000 and others not. We had to completely uninstall Exchange, then restore the database. I got to work at 10 a.m. that day. We started the restore process at 7 p.m. I took a nap at 4 a.m. while the database was restoring. I got locked out of my office at 5 a.m. after going to the bathroom and had to explain my presence to a security guard I’d never seen before (who was finally convinced with, “It’s Friday. It’s 5 a.m. I’m not wearing any shoes. Why would I be here?”) I got home at 9 a.m. the following day—the first full day of Exchange fun.

Around this time, we started having problems with an application used by a team in our company. The application’s producer blamed our mixed-mode Exchange environment.

In February I decided to kill the “upgrade” process and move everyone off of the old server onto the new (much bigger) server. Because of the random application issue, we couldn’t leave until we were in native mode. Also during this time, we decided to upgrade another non-Microsoft application. I got to work again at 10 a.m. and we started the process at 7 p.m. We moved all the databases, uninstalled the former Exchange server, and switched to native mode at about 5 a.m. However, the other upgrade wasn’t going so smoothly. At 6 a.m., we had breakfast tacos. I got home at 10 a.m. the next day, yet another 24-hour shift.

None of us had a “fun, smooth” Exchange 2000 upgrade. One important lesson learned: Remember your badge when you go to the bathroom, because you probably don’t know the security guard who works at 5 a.m.

About the Author

Cynthia Balusek, MCSE, MCT, is a systems analyst for a software company. She swears everything worked wonderfully in testing.

comments powered by Disqus

Reader Comments:

Mon, Oct 7, 2002 Anonymous Anonymous

Serves you right for trying this without testing it in the lab first.

Thu, Sep 26, 2002 Jim Washington, DC

I understand the reluctance on the 24-hours days. My exchange 2k Upgrade went without error, but had a hardware failure on the server before the first regular backup. This caused me to have to go back to my 5.5 backup, and then each upgrade failed from there. MS had to clean up AD for me. Go figure. Lesson here: Backup system; upgrade system; backup system - I guess.

Wed, Sep 25, 2002 Craig Lindstrom, MN

Yep, cheap hardware is often a problem.

Wed, Sep 25, 2002 Anonymous Anonymous

Fun Article. I'm glad I'm not the only one up all night. All that and a 20-30% reduction in pay over the last year for all us IT people! Weee!!

Tue, Sep 24, 2002 Anonymous Anonymous

Same thing has happened to me. Yesterday was my second attempt and it failed, having to roll back to 5.5 again. AD screwed up my upgrade.

Sat, Sep 21, 2002 MikeW Anonymous

I just migrated a client to 2000 from NT4 & exchange 5.5
Migration using the Exmerge utility is the way to go.
Check for viruses first.
A virus wacked the sysvol on the first attempt. Found over 4900 infect emails.
After removal everything went smooth.
Took about 6 hours from format to up and running.

Thu, Sep 19, 2002 Anonymous Anonymous

As a cute anecdote, this article scored. As for being helpful in getting through an Exchange 5.5 to 2000 upgrade, it gets a ZERO.

Wed, Sep 18, 2002 Anonymous Anonymous

I understand your pain. We had a similar experience (65 hours on site over a 72 hour period) when we tried and in place upgrade from Exchange 5.5 to Exchange 2000. Big mistake. Migration is the way to go. Unfortunatly, we moved to Exchange 2000 about 1 month after it was released and the pain of upgrading was not wildly known (in the lab it worked great).

Wed, Sep 18, 2002 Webproxy MCSE, CCNA in the middle of a 26 site e2k migratio

Whatever possesed you to do an in place upgrade? Even with native tools a migration is safer and less prone to disaster. If cost of a new server was an issue you should could have used a desktop with extra HD space as a temp machine, as a previous commentor pointed out. As for the all nighters, I'ved found these are usually brought on by poor planning and lack of experience.

Wed, Sep 18, 2002 NightKiller there abouts

You bought a new server for this? Why was migration not considered? Did you need the security groups? It is FAR less painful and the migration can be done in stages. Even if the mail ultimately needed to be on the same box, it's easier to migrate to Exchange 2000, decommission 5.5, remove 5.5 from a site, install a new instance of Exchange 2000 on the old machine, migrate the mailboxes back and decommision the temporary Exchange Server. Most of this can be done during regular business hours. I just completed a 50 user site exactly this way in 12 hours including a 5.5 online backup and an ADC Connector install. This was done on a Mixed Mode domain as well.
While I like the granularity of AD, replication is a pain in the backside, even on a well connected network.
This is the approach an MCT would take.

Wed, Sep 18, 2002 Anonymous Anonymous

I went through a failed attempt to upgade as well. I was greeted by the blue screen of death while the mailboxes were being upgraded to 200. Ended up putting 26 hours in total to restore from backup etc. I feel your pain.

Wed, Sep 18, 2002 Anonymous Anonymous

Not enough technical explantion, too much whining about an all-nighter. If you can't stand the heat ...

Wed, Sep 18, 2002 Xena IT warrior Omaha

I shared your pain, twice, and did lock myself out of the building on one of the upgrades. Thank goodness for McD's breakfast tacos!

Wed, Sep 18, 2002 bigbillyt a data protection corp

I have had to bail out many people who have faced the same pain. AD, ain't it great?!

Add Your Comment Now:

Your Name:(optional)
Your Email:(optional)
Your Location:(optional)
Comment:
Please type the letters/numbers you see above

Redmond Tech Watch

Sign up for our newsletter.

I agree to this site's Privacy Policy.