You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
@jtaleric points out that Rally scenarios should be autonomous, that is they should not depend of the system state created by any scenarios that have run before them.
This is currently not true for scenarios that Rally ovs plugin provides because they depend on having a running controller sandbox (one with ovn-northd and NBDB/SBDB databases). Also, scenarios that involve port binding assume that there are fake chassis sandboxes running.
Instead of having scenarios for creating/destroying sandboxes that we need to run before/after the actual workload scenario, it would make sense to extend existing context(s) or create a new one that will create sandboxes before the scenario runs and clean them up afterwards.
@jtaleric points out that Rally scenarios should be autonomous, that is they should not depend of the system state created by any scenarios that have run before them.
This is currently not true for scenarios that Rally
ovs
plugin provides because they depend on having a running controller sandbox (one withovn-northd
and NBDB/SBDB databases). Also, scenarios that involve port binding assume that there are fake chassis sandboxes running.Instead of having scenarios for creating/destroying sandboxes that we need to run before/after the actual workload scenario, it would make sense to extend existing context(s) or create a new one that will create sandboxes before the scenario runs and clean them up afterwards.
An example of a such context can be found in Rally
openstack
plugin:https://github.com/openstack/rally/blob/master/rally/plugins/openstack/context/nova/servers.py
The text was updated successfully, but these errors were encountered: