On Tue, 2019-09-03 at 13:36 +0100, Daniel P. Berrangé wrote: > On Tue, Sep 03, 2019 at 02:26:23PM +0200, Andrea Bolognani wrote: > > On Tue, 2019-09-03 at 14:09 +0200, Michal Privoznik wrote: > > > + The big monolithic libvirtd daemon can now be replaced by > > > smaller > > > + per-driver daemons. Distributions can chose if they want the > > > former > > > + or the latter. The libvirtd is still kept around for backwards > > > + compatibility. > > > > We might want to point out that things like SELinux support and > > identity passing are not implemented yet, so distributions will > > probably want to hold off using the split daemons by default for a > > little while longer... Dan? > > Yep, the traditional libvirtd is stil the default. > > We should say that the split daemons are "experimental" at this > stage.
We could have a section called "Experimental features" and move this entry there. > > In order to allow libvirt to abort on out of memory, we need to stop > > linking... > > I would not mention OOM abort here, since that's not in this version. > > I'd just say that we have simiplified the dependancies of these two > parts to make them self-contained. For the NSS plugin this reduces > amount of code & 3rd party libs loaded into every process For the > login shell this makes the thing simpler to audit for security. That's more accurate, I agree. I have already pushed Michal's patch, but we can fix it with a follow-up before release I think. Care to quickly cook up something? -- Andrea Bolognani / Red Hat / Virtualization -- libvir-list mailing list libvir-list@redhat.com https://www.redhat.com/mailman/listinfo/libvir-list
