On Nov 19, 2007 11:20 AM, UNIX admin <[EMAIL PROTECTED]> wrote: > - if SVM is used, metadevices in question must be `metadetach`ed, and any > meta databases on these metadevices destroyed > - the metadevice(s) in question must not house /, /usr or swap on them > - the controller must be capable of hot remove/hot plug support > - the operating system must be notified which action will be taken, prior to > actually performing the action > - the port(s) must support hot remove/hot plug operations.
Yes, so a Netra, as dclake pointed out he's using or what i was discussing doesn't have to be in a cluster and does not require failover to another node for disk replacement. There are advantages of using a cluster but they're not needed for disk replacement on a Netra.. for other non-hotswap systems your point of having a cluster is valid, but then again, one might argue that if they don't spend 5K euro on a Netra with hotswap capabilities they won't spend 3K on a Sun cluster validation and/or Sun Cluster support contract either. So basicly the comments about running 'cfgadm -c unconfigure' in one of the first emails plus the comments from nachox were already establishing what you say. About the "> - the metadevice(s) in question must not house /, /usr or swap on them", I don't know where you read that but this is again in contradiction with my experience and what Sun engineers tell me and how they handle our systems. A Netra v240 only has two internal bays. We slice c1t0d0 into /, swap, /var, /opt and /metadb (last one is used for metadb later) and we have no problems whatsoever with replacing one of the internal disks. Where did you read this statement exactly? A Sun authentic resource? Patrick _______________________________________________ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org