Hello Devs. It seems the "/etc/init.d/libvirtd reload" function is gone. Did that happen on purpose? If so may i ask way? It is really nice to reload the libvirtd to get the configfile reread without stopping and starting all running VMs. thanks and cheers t. Reproducible: Always
What libvirt version are you running?
Sorry. Forgot to mention. app-emulation/libvirt-1.3.1. thanks t.
> Did that happen on purpose? If so may i ask way? Yes. From /usr/share/doc/libvirt-1.3.1/README.gentoo.bz2 (that should have been printed as an elog message on upgrade): Important: The openrc libvirtd init script is now broken up into two separate services: libvirtd, that solely handles the daemon, and libvirt-guests, that takes care of clients during shutdown/restart of the host. In order to reenable client handling, edit /etc/conf.d/libvirt-guests and enable the service and start it: $ rc-update add libvirt-guests $ service libvirt-guests start
Hello Matthias. Thanks for your answer. I noticed the init file split. But how do i reload / restart libvirt without shuting down the VMs? When i restart libvirtd all VMs restart too. When i restart libvirt-guests all Vms start too. But i wanna reload libvirt to reread its configuration files. How do i do that? thanks and cheers t.
Try tweaking /etc/init.d/libvirt-guests and change 'need libvirtd' to 'after libvirtd'
Cool. That worked. hn0 ~ # /etc/init.d/libvirtd restart * Stopping libvirtd ... [ ok ] * Starting libvirtd ... [ ok ] hn0 ~ # VMs still running. Thanks Doug. I think thats what we need. cheers t.
(In reply to Thomas Stein from comment #6) > Cool. That worked. > > hn0 ~ # /etc/init.d/libvirtd restart > * Stopping libvirtd ... > [ ok ] > * Starting libvirtd ... > [ ok ] > hn0 ~ # > > VMs still running. Thanks Doug. I think thats what we need. > > cheers > t. Just wondering if 'use' gives similar results? That's what I was thinking of switching things over to.
>Just wondering if 'use' gives similar results? Yes, that works too.
Doug, I suggest that we treat this as a regression for 1.3.1 and prepare a revision bump. (We have a bit of leeway here - our original announcement for the init script split was fortunately a bit vague about the concrete behavior.) But I think this is the best course of action [1]. [1] I hope I haven't missed any regressions we would introduce with a "use" dependency.
Thanks for the report. Fixed in http://gitweb.gentoo.org/repo/gentoo.git/commit/?id=8d5b9e4b342a1a8b1e750373c882260a4fd34a9d