I tried the following change which just resets the internal status back to uninitialized when the init call fails due to a failed open on the socket file. This has of course the effect that this is tried again, every time someone connects to libvirt via a xen uri. But it seems to be doing the right thing overall with or without a xen hypervisor present.
Of course the other option is to not build the xenfs driver as a module. But it feels a bit dumb to make the kernel bigger just because of this race. Or we could change the xend startup to stop and start libvirtd if present. Which also seems a but wrong. ** Patch added: "proposed.debdiff" https://bugs.launchpad.net/ubuntu/+source/libvirt/+bug/922486/+attachment/2724070/+files/proposed.debdiff -- You received this bug notification because you are a member of Ubuntu Bugs, which is subscribed to Ubuntu. https://bugs.launchpad.net/bugs/922486 Title: libvirt boot race on xen hypervisor To manage notifications about this bug go to: https://bugs.launchpad.net/ubuntu/+source/libvirt/+bug/922486/+subscriptions -- ubuntu-bugs mailing list [email protected] https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
