Recently, I was trying to install a package. And it gave the error telling that, some "other patch" is already installed and hence this will not be installed till the other patch is removed. (I don't remember the exact error message. But the gist of the message is this). So, I removed the "other patch". Before I removed the other patch, it checked the contents of that patch and found that it had many other (>25) packages. So, I didn't want to waste time, trying to find which is necessary or not. I was confident that, if something was "very much necessary" for the system, then it will not be removed. But, when I removed the patch, the system got broke and it doesn't even boot, to login prompt!!
I think, this should be corrected. The system should not allow the user to remove or atleast warn if "very much necessary" package is removed. Regards, Narendra On 12/6/06, Chris <cpj1 at visi.com> wrote: > > On Tue, 5 Dec 2006, Valerie Anne Bubb wrote: > > > On Thu, 30 Nov 2006, Jens Elkner wrote: > > > Unfortunately the running system has no idea wrt. to clusters. And > no-one > > > should expect an user/admin to inspect the .clustertoc, before he > removes > > > a from the running system as unused reported package. > > > Compared with Linux, Solaris seems still to live in the stoneage: > > > The package system provides the opportunity to record dependencies, > > > but it seems to be rarely used (perhaps because the developer think, > > > the initial install program needs to know about the deps, only)... > > > > I have to agree with Jens here - our packaging dependancies could really > > use some work (it's been a nit of mine since before I started working at > > Sun... so it's not a new problem... ) The packaging utilities need to > > prevent our users from hurting themselves this way, and the only way > > is to set up correct packaging dependancies. > > Just to throw my 2 cents in the fire, I think an additional method to > preventing users from "hurting themselves" is better education. > > I've done hundreds of jumpstarts, package installs/removals, and I'll > admit I've also read .clustertoc. I've had Solaris 8 servers running with > a bare minimum of 40 packages (I think, it's been awhile). > > I never really understood the administrator mentallity of just blindly > uninstalling a package from a bare minimal install just because it > doesn't appear to be used by anyone. If I ever had questions about a > single package, I'd check whether or not it was mostly binary files or > shared libraries, whether any obvious programs dynamically linked to that > library. And after all is said and done and I removed it, I'd only end up > saving a couple of megabytes of disk space. > > Maybe it would add a layer of safety if clusters were treated as stump > packages, with directories like SUNWCreq under /var/sadm/pkg/ that could > be referenced by pkgadd/pkgrm or whatever. It would certainly help if > you're trying to push Solaris towards unskilled Linux > administrators. But maybe a stern warning of "don't uninstall this > without really looking at it" can also accomplish the same goals. > > > -------------------- > Christopher Josephes > cpj1 at visi.com > _______________________________________________ > install-discuss mailing list > install-discuss at opensolaris.org > http://opensolaris.org/mailman/listinfo/install-discuss > -- Warm Regards, Narendra Visit my blogs at: http://ssnarendrakumar.blogspot.com/ http://blogs.sun.com/roller/page/ssnkumar ___ ___ __ _ / __/ / __/ / | / / _\ \ _ \ \ / /| |/ / \___/ \___/ /_/ |__/ -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://mail.opensolaris.org/pipermail/install-discuss/attachments/20061207/949ce315/attachment.html>
