Created attachment 321707 [details] Screen capture of failed NFS install Description of problem: When installing Fedora 10 snapshot 3 using a NFS repository the install fails with a "metadata error". Version-Release number of selected component (if applicable): Fedora 10 snapshot 3 How reproducible: Everytime Steps to Reproduce: 1. Try to install Fedora 10 snapshot 3 using a NFS repository Actual results: See attachment and above description Expected results: Working installation Additional info: This method of installation has worked for me without problems for multiple Fedora/RHEL releases up to and including F8, I haven't tried F9.
Created attachment 321708 [details] Screen capture of failed NFS install with a manually created loopback mount When I tried manually creating a loopback mount on the NFS server for the ISO image I get the following error.
As an FYI, the NFS directory contains the following files: * Fedora-10-Snap3-x86_64-DVD.iso * install.img * images (symlink to '.')
Can you attach /tmp/anaconda.log to this bug report?
Created attachment 321716 [details] Anaconda Log Requested logfile, /tmp/anaconda.log
Does it work if you pass method=nfsiso:... instead of method=nfs:... ?
When I used "nfsiso" instead of "nfs" the text portion of the installer asks me to pick an install method.
Okay, does keeping method=nfs: but using updates=http://clumens.fedorapeople.org/468885.img fix things? The problem here is we're simply never mounting the ISO images.
Created attachment 321958 [details] Anaconda Log using clumens update from #7 When using the update from comment #7 Anaconda throws an exception and the install has to be aborted.
Oh, sorry about that - I was mixing up two different methods of fixing this problem. Could you please retry with the same updates image?
Created attachment 321983 [details] Anaconda Log using clumens update from #9 Another exception, see the attached log. Are you guys not able to reproduce this problem locally? I'm happy to help verify a fix but it seems like I'm doing the unit testing here ...
Sorry about that, the proper fix should be in the next build of anaconda.
I have verified the above scenario with the latest anaconda (which will hit rawhide tomorrow). Closing this bug.