lp:~vishvananda/nova/clean-compute

Branch merges

Propose for merging

No branches dependent on this one.

Merged into lp:~hudson-openstack/nova/trunk at revision 568

Eric Day (community): Approve on 2011-01-14

Devin Carlen (community): Approve on 2011-01-14

Diff: 101 lines (+13/-28)

3 files modified

nova/api/ec2/cloud.py (+1/-17)
nova/compute/api.py (+10/-9)
nova/network/manager.py (+2/-2)

api

Related source package recipes

No recipes using this branch. (?)

Create packaging recipe

Related snap packages

Related bugs

Bug #702998: attach_volume fails due to missing import.	High	Fix Released
Bug #703014: get_console_output should be in compute.api	Medium	Fix Released

Link a bug report

Related blueprints

570. By Vish Ishaya on 2011-01-14

removed rpc in cloud

569. By Vish Ishaya on 2011-01-14

merged trial fix again

568. By Vish Ishaya on 2011-01-14

fix bad function signature in create_networks

567. By Vish Ishaya on 2011-01-14

merged lp:~vishvananda/nova/lp703012

566. By Vish Ishaya on 2011-01-14

import re, remove extra call in cloud.py. Move get_console_output to compute_api

565. By Ed Leafe on 2011-01-14

Implements the blueprint for enabling the setting of the root/admin password on an instance.

It uses a new xenapi plugin 'agent' that handles communication to/from the agent running on the guest.

564. By Nachi Ueno on 2011-01-14

OpenStack Compute (Nova) IPv4/IPv6 dual stack support
http://wiki.openstack.org/BexarIpv6supportReadme

Tested with
 unit test
 smoke test

No conflict with current branch r 562.
Fixed comment by Soren and Vish

563. By MORITA Kazutaka on 2011-01-14

This change introduces support for Sheepdog (distributed block storage
system) which is proposed in
https://blueprints.launchpad.net/nova/+spec/sheepdog-support

Requirements:

 - libvirt 0.8.7 or later
 - qemu 0.13.0 or later

How to test:

1. install Sheepdog

  The software is available from SourceForge.net:
    https://sourceforge.net/projects/sheepdog/files/

  See also:
    http://wiki.qemu.org/Features/Sheepdog/Getting_Started#Install

2. run the sheepdog daemon on each hosts

    $ sheep /store_dir

  /store_dir is a directory to store sheepdog objects. The directory
  must be on the filesystem with an xattr support.

3. format the sheepdog storage

    $ collie cluster format --copies=3

4. run nova-volume

    $ nova-volume --volume_driver=nova.volume.driver.SheepdogDriver

562. By Vish Ishaya on 2011-01-14

This modifies libvirt to use CoW images instead of raw images. This is much more efficient and allows us to use the snapshotting capabilities available for qcow2 images. It also changes local storage to be a separate drive instead of a separate partition.

I'm proposing this branch for review to get feedback. I may have inadvertently broken a few things. Comments and possible issues:

1. I haven't tested the other hypervisors. I may have broken libvirt xen support and uml support with this patch.
2. Is it useful to have a use_cow_images param, or should it just be automatic for qemu/kvm and turned off for everything else.
3. create_image is a large annoying method. I tried to clean it up a bit, but it could probably use a bit more refactoring.
4. disk.py seems to be only used by the hypervisors, so perhaps it should move into virt dir.
5. disk.py/partition() is unused now. Should we leave it in or throw it away?

Comments welcome

561. By Soren Hansen on 2011-01-13

Add a new method to firewall drivers to tell them to stop filtering a particular instance. Call it when an instance has been destroyed.

Use dict()s (keyed off id's) instead of set()s for keeping track of instances and security groups in the iptables firewall driver. __eq__ for objects from sqlalchemy fetched in different sessions doesn't work as expected, so I needed to explicitly filter on ID.