Jens Elkner wrote: > On Wed, May 25, 2011 at 07:57:50PM -0400, James Carlson wrote: >> It does not just set -R to point to the zone's root and let fly. > > Ja, than I probably mixed it up with the zadm[d] -R ... > > However, at least for native, not running zones I can't see any reason, > why one should not use -R. IIRC in contrast to pkgadd IPS is not running > any pre/post install scripts anymore. So it basically comes down to a > dep check, SW download and "untar" it in the [virtual] root dir. May be > hard links are a problem, but that's IMHO nothing, what the pkg stuff > can't find out/correct.
The first hit on google for "scratch zone design" is the document I wrote describing how some of this stuff works: http://arc.opensolaris.org/caselog/PSARC/2005/474/inception.materials/scratchzone-design.pdf There are a lot of details to it, but the basic problem is that the packaging scripts are all written with assumptions built on the "usual" use of pkgadd, and using -R to point at a non-global zone root breaks those assumptions. For one very simple example, running "zoneadm" inside a non-global zone definitely doesn't do what you want it to do. Check out the pkginfo(4) man page for Zones-related parameters that are involved. If you run with -R pointing to a non-global zone, then there's no way that pkgadd can use these parameters properly, and the result can be somewhere between trashy and wreaking havoc. If you get lucky, sure, using pkgadd -R this way kinda works, and there's possibly no harm at all. However, there's no way to know whether you're going to get lucky, and when you're unlucky, really bad things can happen. The simple way to know that you shouldn't be doing this, though, is that the man page says not to, and nobody writes man page text for their own amusement. Packaging is one of those things that sure seems on the surface like it ought to be simple -- isn't it just a big cpio ball? -- but it's really not. > [ > Just in case: If one is interested in more numbers, have a look at > http://iws.cs.uni-magdeburg.de/~elkner/osol/ips/ I'm not involved in any way with IPS. Your problems there are better discussed with the IPS folks. > Last but not least: Upgrades with IPS! How should that work without -R? > Rebooting the server after update/upgrade, and than reboot all zones > again (re-attach)? - starting to shiver ;-) I don't know, but given what (little) I know about IPS, I'd expect it to be simpler. IPS doesn't allow scripting, so it's fully in control of all the changes that are made, and it should be able to operate on the alternate roots of the non-global zones without trouble. Of course, there are some complicated things going on in IPS that I don't understand, such as the interactions it has with SMF. Your mileage will undoubtedly vary. But I don't understand your comment about multiple reboots. I don't see how that could be a logical consequence of any reasonable design. >> Actually, I read the that initial reply you got much differently. > ... >> that might concern you. It's a request for information. > > Hmmm, may be I'm a victim of the responses we often got wrt. support > cases and should draw the language joker ;-) Good luck. -- James Carlson 42.703N 71.076W <[email protected]> _______________________________________________ install-discuss mailing list [email protected] http://mail.opensolaris.org/mailman/listinfo/install-discuss
