Monday, February 14, 2005

Clustering NetWare - Today's Lesson

Hello Folks

I thought I would share with you an experience that my customer had with Novell Clustering Services. The outcome was that my customer lost all the data on the SAN supporting 12 Cluster nodes running GroupWise post offices. Fortunately the post offices were all empty post offices, they are planning on migrating users over in early March. But if the users had been migrated to these post offices, the effect would have been devastating!

1. The customer plugged a server box into the SAN, and proceeded to do an installation of NetWare 6.5.

2. The customer installed NetWare 6.5, and just took all of the defaults, clicking the "Yes" button all the way through.

NOTE: At the very beginning of the NetWare 6.5 installation, the installation assesses all of the available disk space and asks if the operator would like to have NetWare reformat it and make it accessible for the NetWare 6.5 server. The customer said "Yes".

3. Nothing adverse seemed to happen!

4. The next day the customer attempted to add the new NetWare 6.5 server to the cluster, which caused a re-read of the Partition tables on the SAN, and WHAM! All cluster nodes went comatose, and all NSS volumes were gone.

It was the NetWare 6.5 installation from the day before that actually caused this problem, but none of the nodes were aware of the problem until they re-read the partition tables.

In NetWare 6.5 SP3 there will be a warning box so that customers are less likely to do this. More than one customer has experienced this, development was aware of the issue when my customer's issue was reported to them.


No comments: