Philip Brown wrote:
> Bart Smaalders wrote:
>> Philip Brown wrote:
>>> "This is the New and Improved way of doing postinstall scripts. This 
>>> is why it is an improvement. And oh by the way, here are some tools 
>>> for you, to help you transition your existing postinstall scripts, to 
>>> the new-and-better way".
>> I'm hesitant to call these post-install scripts.  They won't take the same
>> arguments, they don't run at the same time, and the order of their
>> execution isn't determined by pkg dependencies, but (likely)
>> by service dependencies.
>>
>> They're going to be quite different, but should be much easier to
>> write correctly, since your script/action runs in the context of the
>> installed image, rather than in the context of the machine creating the
>> image. You can actually invoke binaries that are part of the pkg you've
>> installed, ....
> 
> 
> Sounds like "a postinstall script, that doesnt have to use
>   chroot $PKG_INSTALL_ROOT
>   any more".
 >
 > Why would that be "quite different"?
 >
 > I could see "quite simplified" :-) in that it doesnt need chroot any 
more,
 > etc. but why should it have to be "quite different"?
 >

chroot $PKG_INSTALL_ROOT

will only work for images that are the exact duplicate of the hostmachine,
and will fail spectacularly on diskless machines, cross-architecture 
images,
etc.


> If you chose to implement them in a compatible fashion, it sounds like from 
> your description, it might theoretically be possible to take a *cleanly 
> written* old-style postinstall script, and drop it inside some kind of SMF 
> wrapper to run it at the most appropriate time.
> It would think it was just running with PKG_INSTALL_ROOT=/, and do what it 
> needed to do.
> 

When should this script run?


> 
>> Because we avoid running unknown scripts as root in installation
>> context, we will able to install pkgs into zones w/o having
>> to boot each one as we do today; instead we can do all the
>> pkgs ops on zone clones... this will greatly speed up zone
>> packaging/patching operations.
> 
> I dont see what the big fuss is about "unknown scripts in installation 
> context", compared to "unknown scripts being run through an SMF harness".
> 

Have you ever patched a zone?

- Bart




-- 
Bart Smaalders                  Solaris Kernel Performance
[EMAIL PROTECTED]               http://blogs.sun.com/barts
"You will contribute more with mercurial than with thunderbird".
_______________________________________________
pkg-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/pkg-discuss

Reply via email to