> up things my moving volumes to the new server only to realize that I > did something wrong in the process. I though that "copying" ( if such a > thing exists) would be safer. But from the response of Russ only one > copy should exist on a cell so I am not really sure what to do.
I'm not sure, if this will work, but it SHOULD. Someone with more experience and knowledge may want to confirm whether I'm right or not. Anyway, if you're really paranoid about it, you might just do this for all your volumes: 1) vos addsite -server old -partition /wherever_it_is_NOW -id volume_name 2) vos addsite -server new -partition /wherever_it_fits -id volume_name 3) vos release volume_name 4) vos move -id volume_name -fromserver old -frompartition /wherever_it_is_NOW -toserver new -topartition /wherever_it_fits This way you'll have two copies at all times, but still end up having moved the read-write volume to the new server. The addsite commands just set up two read-only replicas. Next, you'll replicate the DB server, too. Easiest way to do this is to create an upserver process on the old machine and an upclient process on the new host. (Read the AFSLore or man bos_create for details.) Adding DB servers to an AFS cell is rather tricky for the *clients*, so you'll need to be sure you do it right; otherwise your client will lose access to their files (nothing serious, just inconvenient: it can be fixed easily). At this point you can test how things work by unplugging the network from the old server - if things go wrong, you just plug it back. If everything works, just use "vos remsite" to remove the RO replicas from the old server and "bos delete" to remove the upclient and upserver processes. After that, just scrap the old server. -Juha -- ----------------------------------------------- | Juha Jäykkä, [EMAIL PROTECTED] | | home: http://www.utu.fi/~juolja/ | -----------------------------------------------
signature.asc
Description: PGP signature