Bug 1064947

Summary: nova-compute shouldn't spawn two libguestfs appliances every time an instance is launched
Product: Red Hat OpenStack Reporter: Jim Minter <jminter>
Component: openstack-novaAssignee: Pádraig Brady <pbrady>
Status: CLOSED WONTFIX QA Contact: Ami Jeain <ajeain>
Severity: low Docs Contact:
Priority: low    
Version: 4.0CC: ndipanov, rjones, sgordon, yeylon
Target Milestone: ---   
Target Release: 6.0 (Juno)   
Hardware: Unspecified   
OS: Unspecified   
Whiteboard:
Fixed In Version: Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2015-02-03 12:58:06 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 Jim Minter 2014-02-13 15:43:35 UTC 
Using RHELOSP 4.0 GA bits, I'm finding that when I launch the Cirros 0.3.1 image, separate calls to libguestfs within the nova codebase cause qemu-kvm to be run twice *before* the instance is launched.  This is suboptimal.

One libguestfs call (file injection) can be disabled by setting libvirt_inject_partition=-2, but this does not work for the second one (checking to see if the volume partition/filesystem can be extended).  The codepath for the second call is approximately:

/nova/virt/disk/api.py extend()
/nova/virt/disk/api.py is_image_partitionless()
/nova/virt/disk/vfs/guestfs.py VFSGuestFS.setup()

It would be good if all of this could be done with one libguestfs instance which could also be disabled in the global nova config.
Comment 2 Pádraig Brady 2015-02-03 12:58:06 UTC 
Upstream is moving away from file injection. To disable for now you can remove python-libguestfs from the compute nodes