[OpenIndiana-discuss] copying a reference installation to other hosts

[Udo Grabowski (IMK)]

On 15/03/2013 15:11, Hans J. Albertsson wrote:
> I've found a few more problems with this script:

As said, this script has to be adapted to the local needs. I cannot
imagine how to automate this for all weird stuff that can occur, and
this includes only the stuff I know about (such as NFS mounted file
hierarchies, multiple networks with aggregates and VLANs, ISCSI, IB,
etc.ppp..).
Any attempt to generalize this must ultimately fail. Therefore we
usually keep it as small as possible and sufficiently adapted as
necessary for our current needs.

>
> 1: It doesn't create the /export/home FS bits, but gladly brings the users over.
> There should be a privileged user complete with a home dir, a user that isn't
> root, for doing the next install and config bits. I suppose one just adds a zfs
> create /export/home/user, and then some.

No, in our case /home is automounted, and all users are on a central
NFS server, so we do not handle this here (but the proper automounter
files are copied with this script, so no extra step for us). Remember,
this is a script for an enterprise site with centralized services, not
for single user pc deployment schemes, which we really don't want for
numerous reasons....

>
> 2: For some reason the script won't create the swap and dump in my case, and the
> system comes up w/o any swap or dump conf'ed. I calculate the sizes of swap and
> dump from the output of prtconf -v | grep -i memory, dividing by factors 1.9 and
> 1.7 respectively, and a multiplier char from the mem size unit string. I wonder
> if that could cause a problem or if there's some timing issue??
>

Don't know what you are doing there exactly, for fixed sizes this simply works.

> 3:
> ifconfig e1000g0 unplumb
> ifconfig e1000g0 plumb
>
> will definitely fail if the interface is plumbed to start with. Putting a sleep
> 10 between the two will mostly fix it,...

Might be, there are a few places where a sleep is needed here or there,
depending on machine speed or network factors out of control.

>
> PS: I always config two disks and produce a 2way mirror from those. This
> entirely happens before the rpool is send/recv'd over.
> All my servers are standardised to have two identical 2TB disks in a mirror for
> the system.
> This change might prod some extra problems, of course, but I don't know what
> that might be.

We usally put in the mirror by hand later, as it is only a simple
attach with an installgrub, this will save you from trouble if that
script fails for some reason.

-- 
Dr.Udo Grabowski    Inst.f.Meteorology a.Climate Research IMK-ASF-SAT
www.imk-asf.kit.edu/english/sat.php
KIT - Karlsruhe Institute of Technology            http://www.kit.edu
Postfach 3640,76021 Karlsruhe,Germany  T:(+49)721 608-26026 F:-926026