Description of problem: When using pungi: Traceback (most recent call last): File "/usr/bin/pungi", line 183, in <module> main() File "/usr/bin/pungi", line 118, in main mypungi.doSplittree() File "/usr/lib/python2.5/site-packages/pypungi/pungi.py", line 226, in doSplit tree output = timber.main() File "/usr/lib/anaconda-runtime/splittree.py", line 395, in main self.createSplitDirs() File "/usr/lib/anaconda-runtime/splittree.py", line 234, in createSplitDirs self.createDiscInfo(i) File "/usr/lib/anaconda-runtime/splittree.py", line 162, in createDiscInfo discinfo_file = open("%s-disc%d/.discinfo" % (self.dist_dir, discnumber), 'w ') IOError: [Errno 2] No such file or directory: '/home/fedora-ia64/development/ia6 4/os-disc2/.discinfo' Version-Release number of selected component (if applicable): anaconda-runtime-11.2.0.37-1 How reproducible: 100% Additional info: Patch attached.
Created attachment 150631 [details] Fix for this issue
Those dirs are supposed to be created during the first disc loop. Unless you've somehow started splittree after your first disc existed, you shouldn't get this error. How did you get to this point
Okay -- I stand corrected. It seems to work if you do: pungi -c /etc/pungi/pungi.conf --destdir=/home/fedora-ia64 but not if you do pungi -c /etc/pungi/pungi.conf --destdir=/home/fedora-ia64 -GBP pungi -c /etc/pungi/pungi.conf --destdir=/home/fedora-ia64 -SI That's odd ... P.
I take back comment #3. I've seen this happen with a clean destdir. P.
(In reply to comment #2) > Those dirs are supposed to be created during the first disc loop. Unless you've > somehow started splittree after your first disc existed, you shouldn't get this > error. How did you get to this point Jesse, this seems like a real problem. If I do a "-GBP" followed by a "-S", I get the error. The os-disc[234..] dirs are definately not be created. Furthermore, if I do "-GBP" followed by: /usr/lib/anaconda-runtime/splittree.py --arch=ia64 --total-discs=5 --bin-discs=5 --src-discs=0 --release-string="Fedora Core development" --pkgorderfile=/home/test/work/ia64/pkgorder-ia64 --distdir=/home/test/development/ia64/os --srcdir="/home/test" --productpath=Fedora I see the same error. Are we using this latest version of anaconda to build CD isos in rawhide? If so, then how does that work? P.
We don't make CDs in rawhide, only boot.iso. This could very well be a splittree issue rather than a pungi one. Will investigate more after test3 goes out (which is DVD only so no splitting going on)
Created attachment 151061 [details] Fix for this issue Jesse, There's definately brokeness in anaconda's splittree.py. Here's the situation: In createSplitDirs (splittree.py), the code creates directories for the 1st disc (os-disc1), but does not create dirs for the additional discs. P.
Created attachment 151121 [details] Fix for this issue Sorry -- wrong patch uploaded. P.
Created attachment 153414 [details] Another possible solution Are all the dirs really needed. In the previous versions of splittree.py the line existed without the "for", and created a directory called RPMS: "os.makedirs("%s-disc%d/%s/RPMS" % (self.dist_dir, i, self.product_path))" AFAIK the RPMS directory needs to go but that does not mean that the product_path directory needs to go as well (right ???) Additionally I have looked at isos from previous FC versions and the CD that are not the first CD only have the Fedora/RPMS directory, the rest are files. This means that the "for" line is not necessary.
Applied Joel's patch