Hide Forgot
Description of problem: error: unable to open stream for '(null)': Bad address -> Unable to install pv guest (sl 6.2) on fedora domain0 (fedora 17 beta rc2) Version-Release number of selected component (if applicable): python-virtinst 0.600.1-1.fc17 kernel-PAE 3.3.0-5.fc17 xen 4.1.2-13.fc17 How reproducible: - installing current fedora 17 (alpha, beta) - installing xen - rebooting in xen environment - ensure that xen domain 0 is running (xl list) 0.5 from 2 GB reserved for dom0 - trying to install guest (domU) from network location -> virt-install --connect xen:/// -n server_xen_232 -f /virtual/xen/server_xen_232.img -s 6.0 -r 512 --vcpus=1 --paravirt --uuid=00000192-0168-0000-0206-001921680232 -l http://192.168.0.250/el6-32 --virt-type=xen --os-type=linux --os-variant=rhel6 -x "ks=http://192.168.0.250/el6-xenoss-srv232.cfg i686 ip=192.168.0.232 netmask=255.255.255.0 gateway=192.168.0.1 dns=192.168.0.1 console=hvc0 earlyprintk=xen" --graphics none --debug --> Connected to domain server_xen_232 --> Escape character is ... --> error: unable to open stream for '(null)': Bad address Steps to Reproduce: 1. 2. 3. see above Actual results: Installation stops (xml file is created) and reports errormessage as listed above. Domain installation is reported to be in progress... but furthermore there's no console connection possible. Expected results: Started guest should continue download network (xen)bootimage and installing rpms. Additional info: Same procedure works fine with fedora 15+16. Not sure if this is a python-virtinst package problem only or a xen error in general!
Created attachment 572814 [details] virt-install logfile
Created attachment 572816 [details] logfile 1 attaching some more xen related logfiles.
Created attachment 572817 [details] logfile 2 attaching some more xen related logfiles.
Hmm, looks like libvirt is generating console XML like: <console type='pty'> <target type='xen' port='0'/> </console> There should be a pty path in there. That error message: error: Unable to open stream for '(null)': Bad address Is coming from virsh console. So this is a libvirt issue
Thanks for additional info and bug redirection. Today, I installed 17-Beta.RC4 which includes latest "libvirt-0.9.11-1.fc17.i686" and still faced the same problem. Seems that from first Fedora 17-TC libvirt does not play together with python-virtinst tool. Meanwhile I used xen old deployment style (xl create -c domu.cfg) with creates a working domU image but does not include 'smart configuration switches' as offered by python-virtinst (xml). Hopefully we'll see a working solution when Fedora 17 Final is released.
If current 17.RC4 will become GOLD/FINAL release the reported problem still exists... No sure if fedora distribution is a suitable choice when it comes to xen as bare metal hypervisor. B.t.w., after installing xen packages, fedora's bootloader configuration (grub2.cfg / grub-install) ist not modified to boot in xen-environment.
'Problem' solved, xenconsoled was not running... daemon had to be started by hand