[Libvirt-cim] [PATCH] [TEST] Fixing 06_paused_active_suspend.py tc of ComputerSystem

[Kaitlin Rupert]

Deepti B. Kalakeri wrote:
> # HG changeset patch
> # User Deepti B. Kalakeri <deeptik at linux.vnet.ibm.com>
> # Date 1213968570 25200
> # Node ID 92675efb0941a25500820ccfde0a7d73d6058acc
> # Parent  1837fcafef1281396deb7d03f6754da320af8a36
> [TEST] Fixing 06_paused_active_suspend.py tc of ComputerSystem.
> 
> 1) Included the poll method since the enabledState was not getting set for XenFV appropriately.
> 2) Included the bugno["0001"], since RequestStateChange() method was not able to suspend a XenFV domain even when it
>    was in a expected running state.

Bug #0001 is related to a problem where a <bootloader /> was being 
appended to a guest's XML.  If you do a dumpxml of the guest created by 
the test (do a dumpxml after the guest is running), you won't see such a 
tag.

What's happening here is that the FV guest will sometimes return "no 
state" instead of running of blocked.  The provider won't pause the 
guest unless it's in the running or the blocked state.

I modified state_change_pause() so that virDomainSuspend() is called if 
the state is VIR_DOMAIN_NOSTATE.  virDomainSuspend() doesn't return an 
error if the state is VIR_DOMAIN_NOSTATE.  I'm wondering if the guest is 
running by the time virDomainSuspend().  We do have a check there so if 
virDomainSuspend() does fail, we return an error.  So this might be a 
reasonable approach here.

Dan - any thoughts on this?

-- 
Kaitlin Rupert
IBM Linux Technology Center
kaitlin at linux.vnet.ibm.com