GroupWise... working despite your best efforts

Last week I was doing a GroupWise 7 upgrade. No problem, 1 Post Office, in-place upgrade... what do you want me to do with the rest of the morning? Naturally there's some prep work to be done, some post work as well.

For the non-GroupWise engineers out there, GroupWise is Novell's premier e-mail and collaboration suite. Novell has a few recommendations to help keep it running well.

1. Try to keep fewer than 1,000 users on each post office
2. Maintain retention policies so that users will have no more than a year or two of mail on the live system. GroupWise and third parties have archiving options for just that reason.
3. Use the built-in database maintenance utilities (primarily GWCHECK) to maintain and repair database errors. A check should be made of the database structure nightly and the contents weekly.

Looking at this system I saw some things to worry about. Between users and resources (read: non-human mailboxes) there were about 2,200 mailboxes in the post office. There were no retention policies whatsoever, the postoffice was/is approaching 200GB (there are items dating to AT LEAST 1998). There were no GWCHECKs scheduled of any consequence. Twice per week there was a check scheduled to check maibox statistics but that's it.

When I ran a structure check on the database I didn't find anything too alarming (apart from the fact that it took 16 hours). Contents was another story. These checks were taking 70-80 hours to complete.

Looking at a recent contents check that had been run at our behest there were tens of thousands of errors, nearly all minor or irrelevant. They did, however, show that there had been almost no maintenance done.

A HUGE number of outdated execution records (expired notifications or alarms). While these are normal, harmless and to be expected, the date stamps were unexpected. Outdated execution records are directly linked to the date an item was created, when a GWCHECK is run they are removed. I was finding outdated execution records dating back to June of 1998... 8 1/2 years. Had a GWCHECK been run in the past 8 1/2 years they would have been removed.

In 1998 we were just beginning to see GroupWise 5.5... that 5.5 (or 5.2) post office was upgraded to 6.0 and 6.5 with these errors in place the whole time.

GroupWise just works no matter how hard you try to kill it.