On Thu, 25 Jan 2001, John Stoffel wrote:
I've been mulling over migration strategies, and babysitting a migration from a F520 with four shelves of 9gb disks to an F740 with three shelves of 36Gb disks.
I went through similar stuff about a year ago when I migrated a pair of old F230's to a pair of F740's. If you can afford to shut the machine down for a long weekend, I highly recommend just taking the hit and doing a straight level 0 dump/restore. In my case I was not able to do that because the project kept squawking about some idiotic concept called "24/7". What I ended up comprising on was the following: I did a level 0 dump/restore of the live system, within an hour of the level 0 completing I shut down access to the original server and then did a level 1 dump/restore and then powered down the original and came up live on the F740. This limited my perceived downtime to about an hour. The amount of downtime that you will experience is dependent on how active your data is and how long it takes to complete the level 0. It's not a very elegant solution but I was able to script the whole thing and it caused minimal impact to the users and minimal loss of hair on my part. There was need to create a special root volume and waste two disks because I did the copies using qtrees and just created the few required files in the /etc directory by hand.
PS: I did not use ndmpcopy at all because when I first attempted to use it it kept causing the F740 to crash.