>THE FOLLOWING IS VERY DANGROUS AND COULD LEAD TO DATA LOSS, HARDWARE FAILURE
>or SYSTEM CRASH OR ONE OR MORE OF THE ABOVE DO NOT DO THIS!!!!
>If you ignored the above below is how do to this.
>All hardware should be added when the machine is off, how ever if you really
>really really must do this simply daisy chain the new disk in to the chain
>making sure the scsi bus is terminated and the new disk has a unique scsi id
>when you have finished.
I run a Solaris x86 server, and 12 out of its 14 disks
have been added to the machine and brought into service
with the server fully running, without any reboots.
Actually, same goes for the tape drives and CDROM drive.
I allowed for them all on the SCSI buses, so I can plug
them in and out without effecting the integrity of the
SCSI termination. Rules are to always ensure all slices
on the disk are unmounted first (if pulling one off the
bus or changing its SCSI id), and always ensure the
drive is unpowered when [dis]connecting from the SCSI
bus. I also type 'sync' just before each risky bit, so
that if I did crash the machine, damage is limited.
>If the machine is still working at this stage login as root and run
which creates the /devices entries, and then
to create the /dev/[r]dsk/* links (or 'tapes' for tape drives).
Quote:>my advice to you is
>DO NOT DO THIS POWER DOWN THE MACHINE FIRST THEN BOOT IN RECONFIGURATION MODE.
I won't argue; it's clearly risky doing what I do, but
I've had no problems yet.
Consultant Software Engineer