On Fri, Oct 11, 2013 at 09:01:24AM +0200, Philipp Hahn wrote:
Hello,
while trying to reproduce a performance problem I had to fake 95 hosts with 300 VMs. My idea was to setup 95 users with each one running a session libvirtd. When adding them to our management system I had was troubled with finding the right libvirt URI, as the default session UNIX socket is in the "abstract ns", which has no file system representation. I solved that by explicitly starting the session libvirtds with a custom config only specifying a physical socket path: for ((h=0;h<95;h+=1));do su -c "/usr/sbin/libvirtd -f <(echo "unix_sock_dir=\"/home/host$h/.libvirt\"") -d" "host$h";done
1. is there a way to specify an libvirt URI using the abstract ns?
No, its not possible - that's one of the reasons why we stopped using the abstract namespace for session sockets.
2. Does anybody know another/better way to simulate many hosts with many VMs?
Containers ? eg create a container for each VM host you want to simulate. Daniel -- |: http://berrange.com -o- http://www.flickr.com/photos/dberrange/ :| |: http://libvirt.org -o- http://virt-manager.org :| |: http://autobuild.org -o- http://search.cpan.org/~danberr/ :| |: http://entangle-photo.org -o- http://live.gnome.org/gtk-vnc :|