With this patch, all test cases but one are fixed. The one involves a quirk when destroying a vm resource pool, where the application incorrectly tries to unassociate hosts from the pool. This is most likely due to the hack we had to do where we associated hosts w/ with the base pools class instead of vm_resource_pools to fix something or an other. Additionally the test database user had to be changed to the 'postgres' database superuser as postgresql does not allow any other users to enable / disable foreign key constraints, which is required during the initial test fixture load. Finally I didn't create this patch with git format-patch as I've been developing on my ovirt appliance install (slow but needed due to ldap), specifically in the /usr/share/ovirt-wui dir which is not a git repo. I simply generated this via diff. -Mo -------------- next part -------------- A non-text attachment was scrubbed... Name: tests-fix1.patch Type: text/x-patch Size: 9171 bytes Desc: not available URL: <http://listman.redhat.com/archives/ovirt-devel/attachments/20080630/1f91015e/attachment.bin>
Mohammed Morsi wrote:> With this patch, all test cases but one are fixed. The one involves a > quirk when destroying a vm resource pool, where the application > incorrectly tries to unassociate hosts from the pool. This is most > likely due to the hack we had to do where we associated hosts w/ with > the base pools class instead of vm_resource_pools to fix something or > an other.Hmm -- what was your exact error here? Although hosts are listed on the base pools class now, there will never be hosts assigned here, so the "unassociation" should be a no-op. BTW, the reason we had to move the associations to the base class was in order to get :include directives working -- so queries on the base class (i.e. Pool.children, etc.) could automatically include hosts for child Hardware pools -- but hopefully this won't keep tests from working here. Scott