https://blueprints.launchpad.net/fuel/+spec/refactor-cinder-vmware-role
Currently, cinder-vmware [0] (cinder-volume) installed with default cinder-volume service. This leads to many problems, since such behavior is not considered. Moreover, it is wrong from a logical point of view, because we have an extra unaccounted default cinder-volume service (not for VMware). It is also necessary to alter startstop scripts for cinder-vmware.
Deploying cinder-vmware occurs abnormal scenario. At first configure and run all services cinder (cinder-api, cinder-scheduler,cinder-volume) configuration for a given environment (LVM or Ceph). Once configured and started cinder-volume with vmware-backend.
This behavior leads to some problems:
Also, it is necessary to rework the start/stop script that is copied when we install cinder-volume (MOS), for support systemd and sysv-rc.
To implement this blueprint, do the following:
None.
None.
None.
It is planned to make changes to openstack-cinder and cinder-vmware manifests.
There are no alternatives, because if leave as is, we need to prepare for a large number of bugs and problems.
None.
None.
None.
None.
None.
None.
None.
We remove custom upstart cinder-volume-vmware.conf from cinder-volume package.
None.
Primary assignee |
|
Developers |
|
QA engineers |
|
Mandatory design review |
|
None.
User is able to deploy cluster with vCenter and cinder-vmware role. After deploy user can use create volume, create volume from image, etc for vCenter availability zone.