PXE kickstart install. The following repositories are defined: nfs --server=saga --dir=/export/data1/fedora/core/6/i386/os repo --name=updates --baseurl=http://www.cora.nwra.com/fedora/updates/6/i386/ repo --name=extras --baseurl=http://www.cora.nwra.com/fedora/extras/6/i386/ repo --name=CoRA --baseurl=http://www.cora.nwra.com/fedora/CoRPMS/6/i386/ repo --name=livna --baseurl=http://www.cora.nwra.com/fedora/livna/6/i386/ (note that it complains about non-existant updates and livna) but the install fails when starting to install packages and cannot find cups-1.2.3-3.i386.rpm (which is the first to be installed. anaconda.log shows: 16:16:09 WARNING : Try 10/10 for file:///mnt/source/./Fedora/RPMS/cups-1.2.3-3.i386.rpm failed but in the shell I can see /mnt/source/./Fedora/RPMS/cups-1.2.3-3.i386.rpm just fine. File seem okay too. BTW - it would be nice to specify other repositories with nfs: urls.
Created attachment 136406 [details] anaconda log
Also fails with: url --url http://www.cora.nwra.com/fedora/core/6/i386/os repo --name=updates --baseurl=http://www.cora.nwra.com/fedora/updates/6/i386/ repo --name=extras --baseurl=http://www.cora.nwra.com/fedora/extras/6/i386/ repo --name=CoRA --baseurl=http://www.cora.nwra.com/fedora/CoRPMS/6/i386/ repo --name=livna --baseurl=http://www.cora.nwra.com/fedora/livna/6/i386/ and: nfs --server=saga --dir=/export/data1/fedora/core/6/i386/os Verified the md5sum of cups-1.2.3-3.i386.rpm against the Fedora download server.
Is there a /mnt/sysimage/root/install.log? cups should not be the first package to get installed, so I'm wondering if perhaps something else snuck in before it. If cups is actually the first package to be installed, this looks like some sort of package ordering problem. Does the install work if you don't use any other repositories?
install.log is empty. Same result without the "repo" lines. Install of today's (20060919) rawhide went fine. Maybe my test3 copy is bad, though I'm not quite sure how.
Odd. Well, it's possible we fixed something in anaconda (a quick scan of the spec file doesn't have anything jumping out as obvious, though there are a ton of changes). And of course it's possible there was something funny in the kernel NFS code or utilities that has been sorted out in the meantime. I'm going to close this as RAWHIDE, but feel free to reopen it if the problem comes back.