Both our production and failover cluster 2100 were originally supplied
with 2gb disks for the root, swap, tmp and usr domains. They are mirrored
as rz0 and rz1 and are under advfs. They may have been big enough at one
time but the constant addition of system software such as dns, reversible
patches etc etc has meant that only constant housekeeping stops them from
running out of space. This weekend the var logs went over the top with a
vengeance due to a DOS mail attack and we need a permanent solution.
What is the best approach bearing in mind that these are 24/7/365 machines
and all the 'real' storage is controlled by ASE 1.6. (We are on 4.0G P3).
Can we physically add, say, 4gb disks (x2 to keep the mirror) in the main
box? If so how both physically and logically - presumably there are more
physical connectors for an rz3/rz4? And logically - via LSM addvol?
Or do we attempt to build a new 4gb disk on a development system? If so
then can we use the sys file from the live system as the configuration is
not the same? If we do that then would we stop the production server,
back up root, tmp and usr to DLT: swap the disks and then vrestore?
As you can see I think there are several factors to consider and I have
been puzzling all weekend on the best approach so I would appreciate any
thoughts.
Thankyou
stuart mckenzie
Received on Sun Aug 19 2001 - 20:06:07 NZST