On 03/11/2013 03:48 AM, Alex Jia wrote:
On 03/09/2013 02:44 AM, Chris Evich wrote:
+ # TODO, the '--config' option will affect next boot, and if the guest
+ # is shutoff status, the '--current' option will be equivalent to
+ # '--config', if users give a specified nodeset range is more than
+ # host NUMA nodes, and use virsh numatune with '--config' or
'--current'
+ # option to set the invalid nodeset to a guest with shutoff status,
and
+ # then virsh numatune will return 0 rather than 1, because the
libvirt just
+ # check it when starting the guest, however, the current virsh.start()
+ # can't meet this requirement.
Do you think raising a logging.warning() if either condition happens
is warranted? Should we stick that warning into virsh.start()? What's
your opinion on how/where this could be fixed?
At present, we shouldn't hard code "dargs['ignore_status'] = False" at
least, it should be set by caller, of course, we may give it a default
value like this, if users expect failing(dargs['ignore_status'] = True)
to start a guest then the virsh.start() should do a correct judgement
and process rather than logging error and returning False for any failed
case.
Yes, excellent point. I think a default can be established quite easily
inside virsh.VirshBase.__init__(). Then we can remove all/most the
hard-coded values. IIRC, it was hard-coded for the virsh functions
which returned only True/False. I'd really like to purge all those as
well, maybe it makes sense to change all in one patch. I'll take a look
after the freeze. Thanks for your input.
--
Chris Evich, RHCA, RHCE, RHCDS, RHCSS
Quality Assurance Engineer
e-mail: cevich + `@' + redhat.com o: 1-888-RED-HAT1 x44214
_______________________________________________
Virt-test-devel mailing list
[email protected]
https://www.redhat.com/mailman/listinfo/virt-test-devel
