Bug 856611

Summary: libvirt configurable paths
Product: Red Hat Enterprise Linux 6 Reporter: Saveliev Peter <peet>
Component: libvirtAssignee: Libvirt Maintainers <libvirt-maint>
Status: CLOSED WONTFIX QA Contact: Virtualization Bugs <virt-bugs>
Severity: medium Docs Contact:
Priority: unspecified    
Version: 6.3CC: acathrow, bazulay, berrange, dallan, dyasny, dyuan, iheim, lhornyak, michal.skrivanek, mzhan, rwu
Target Milestone: rc   
Target Release: ---   
Hardware: All   
OS: Unspecified   
Whiteboard:
Fixed In Version: Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2012-09-13 08:09:15 UTC Type: Bug
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Embargoed:

Description Saveliev Peter 2012-09-12 12:32:12 UTC
As it seems, there is nothing to prevent us running several libvirt instances — except of hardcoded pathes to the libvirt socket, images directory and so on.

Keeping in mind libvirt driver lock (https://bugzilla.redhat.com/show_bug.cgi?id=856609), it is a serious limitation of libvirt scalability: there is no use of several CPU while creating VMs (and, maybe, not only creating, but we stuck on this with creating).

It would be nice to make such pathes configurable (with default values == what it uses right now). I will allow us to run several libvirtd instances w/o usage of FS namespaces, that is not too trivial.

Comment 1 Daniel Berrangé 2012-09-12 12:38:01 UTC
The libvirt design is such that there is 1 instance of libvirt which runs privilege, and optionally 1 instance per non-root user account which run unprivileged. Running multiple privileged libvirtd instances is a non-goal.

What is your use case for having multiple privileges instances of libvirtd ?

Comment 3 Saveliev Peter 2012-09-13 07:54:36 UTC
There is a scalability issue with libvirt on hosts with many CPUs, caused by the driver lock (see bz#856609). It can be solved either on the driver level, or with several libvirtd instances. The former is the better, but one doesn't exclude another.

Comment 4 Daniel Berrangé 2012-09-13 08:09:15 UTC
We are not going to support running multiple instances of libvirtd just to hack around that bug. That bug should simply be fixed.

Comment 5 Dave Allan 2012-09-13 14:50:53 UTC
Agreed, we should focus our scalability efforts on improving locking.