[OmniOS-discuss] Internal pkg error during a test r151010 to r151014 upgrade

[Volker A. Brandt]

Chris Siebenmann writes:
> > History lesson: until people could afford to purchase more than
> > one disk and before Sun invented the diskless workstation (with
> > shared /usr), everything was under /.
> 
>  As Richard knows but other people may not, this is ahistorical on
> Unix. 

While history discussions are fun, this one distracts from the
problem at hand a bit. :-)

Your problem (/opt is expected to be a part of the BE and things break
when it isn't) is a case of "works as designed".  Oracle has had the 
same problem, albeit with /var.  In recent 11.2 versions, they have  
documented that behavior, and introduced a new separate dataset for
all those things you *don't* want to have under BE management.  The
dataset is called rpool/VARSHARE and mounted under /var/share.

There is a surprising amount of stuff in there, and most of it makes
sense, too.  If your boot hangs because of some failing binary, you
will still have the core dump in /var/share/cores when you go back
to the last BE.  The audit logs are still there, etc. etc.  Even the
mail spool now is in /var/share/mail.  Symlinks are provided for the
legacy locations (like /var/mail).

So one way would be to follow that example and create an OPTSHARE
dataset, mount it under /opt/share, and symlink all the stuff you
need to be outside the BE into it.  We have considered this idea,
as we build our software for the /opt/local tree, for precisely
the same reasons you cite for your separate /opt dataset.  However,
in the end it is not worth it, due to the amount of extra work
it causes.

A better way is to IPS package everything properly, and add proper
metadata to your packages, so that those packages that should go
into a new BE do ask for one in their manifest.  That way, there is
no distinction between any "system" package living in /usr (or wherever)
and your package living in /opt.  You can leverage the tight integration
of IPS, BEs, and ZFS for your stuff.  Updating and rolling back becomes
very easy.

Disk space is not really an issue; ZFS snapshots are relatively cheap.


Regards -- Volker
-- 
------------------------------------------------------------------------
Volker A. Brandt               Consulting and Support for Oracle Solaris
Brandt & Brandt Computer GmbH                   WWW: http://www.bb-c.de/
Am Wiesenpfad 6, 53340 Meckenheim, GERMANY            Email: vab at bb-c.de
Handelsregister: Amtsgericht Bonn, HRB 10513              Schuhgröße: 46
Geschäftsführer: Rainer J.H. Brandt und Volker A. Brandt

"When logic and proportion have fallen sloppy dead"