Post-Installation Configuration in OpenStack Configuration Reference  - havana

Bug #1251241 reported by Stephen Gordon
6
This bug affects 1 person
Affects Status Importance Assigned to Milestone
openstack-manuals
Fix Released
High
Summer Long

Bug Description

Currently the guide recommends setting dynamic_ownership to 0, speaking with one of the libvirt developers - and coincidentally developer of the Nova libvirt driver - (Dan Berrange) it appears there is no reason to do this in OpenStack. In fact setting this option to 0 causes permissions issues when attempting migration.

Since the guide provides no justification for this choice, nor any workaround for the permissions issues, I recommend removing this step and leaving dynamic_ownership as the default (1).

-----------------------------------
Built: 2013-11-14T05:18:26 00:00
git SHA: ed950f1acdf5f007fadbe541ba6a1aac393d1574
URL: http://docs.openstack.org/trunk/config-reference/content/configuring-openstack-compute-basics.html

Stephen Gordon (sgordon)
Changed in openstack-manuals:
assignee: nobody → Summer Long (slong-g)
importance: Undecided → High
status: New → Confirmed
milestone: none → icehouse
Revision history for this message
Alan Pevec (apevec) wrote :

> the guide provides no justification for this choice

Probably docs team picked up advice from the list http://openstack.markmail.org/message/2a2t4al3ivdyb6q6 ?

Revision history for this message
Dafna Ron (dron-3) wrote :

This is how I configured my setup to be able to live migrate instances with gluster storage:

1. create a shared mount (I used the existing /var/lib/nova/instances).
2. add to exports : echo '/var/lib/nova/instances *(rw,sync,no_root_squash)' > /etc/exports
3. add/change the following to nova.conf file:
vncserver_listen = 0.0.0.0
live_migration_flag=VIR_MIGRATE_UNDEFINE_SOURCE,VIR_MIGRATE_PEER2PEER,VIR_MIGRATE_LIVE
instances_path = /var/lib/nova/instances/
live_migration_bandwidth=0
live_migration_retry_count=30
live_migration_uri=qemu+tcp://%s/system
image_cache_manager_interval=0
glusterfs_mount_point_base=/var/lib/nova/instances/

4. mount the storage:

mount <Server-Name>:/<share_on_storage> /var/lib/nova/instances (mount should work now!)

5. Run: openstack-config --set /etc/nova/nova.conf DEFAULT novncproxy_base_url "http://$SERVICE_HOST:6080/vnc_auto.html
6. change to the following in /etc/libvirt/libvirtd.conf
listen_tls = 0
isten_tcp = 1
auth_tcp = "none"

7. change to the following in /etc/sysconfig/libvirtd :
LIBVIRTD_ARGS="--listen"

8. change the following in /etc/libvirt/qemu.conf :
user = "root"
group = "root"

9. selinux and iptables should be configured correctly

10. restart all relevant services on all of the hosts

Revision history for this message
OpenStack Infra (hudson-openstack) wrote : Fix proposed to openstack-manuals (master)

Fix proposed to branch: master
Review: https://review.openstack.org/76744

Changed in openstack-manuals:
status: Confirmed → In Progress
Revision history for this message
Summer Long (slong-g) wrote :

This change is no longer valid; anne removed the CRG page, updated (and in the process, got rid of this issue), and placed back into the Cloud Admin Guide. A search in the doc suite no longer calls up the dynamic_ownership variable. Set to Invalid, since there's no 'notABug' status.

Changed in openstack-manuals:
status: In Progress → Invalid
assignee: Summer Long (slong-g) → nobody
Summer Long (slong-g)
Changed in openstack-manuals:
status: Invalid → Fix Committed
Tom Fifield (fifieldt)
Changed in openstack-manuals:
status: Fix Committed → Fix Released
Summer Long (slong-g)
Changed in openstack-manuals:
assignee: nobody → Summer Long (slong-g)
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.