yahoo-eng-team team mailing list archive
-
yahoo-eng-team team
-
Mailing list archive
-
Message #04927
[Bug 1219761] Re: The instanceType of the requestSpec is set to be the current instanceType of the instance instead of the one that is going to resize to
** Changed in: nova
Status: Fix Committed => Fix Released
--
You received this bug notification because you are a member of Yahoo!
Engineering Team, which is subscribed to OpenStack Compute (nova).
https://bugs.launchpad.net/bugs/1219761
Title:
The instanceType of the requestSpec is set to be the current
instanceType of the instance instead of the one that is going to
resize to
Status in OpenStack Compute (Nova):
Fix Released
Bug description:
The scenario where this bug was found is:
Based on HyperV hypervisor, while resizing disk size value 3T that is bigger than available value ,vm status is not correct and vm missed in the hyperv cloud.
Steps to re-create this issue:
1. Create a flavor Disk_3T with disk storage size 3T.
2. Resize one vm ,either win2012 or redhat64 vm with the flavor.
Actual Result:
1. But the VM status is error. (see screen shot)
2. Vm lost in the hyper-v cloud.
The solution we proposed previously is to add the
scheduler_default_filters = RetryFilter,AvailabilityZoneFilter,RamFilter,ComputeFilter,ComputeCapabilitiesFilter,ImagePropertiesFilter,DiskFilter
to the nova.conf and restart the scheduler service. The DiskFilter detects that there was no host meets the instanceType that was going to resize, so a NoValidHost was raised and the VM instance was marked as ACTIVE.
But that worked almost 2 month ago.
During the regression test, this case failed again, and I found there
is a great code change there and is still undergoing according to the
review: https://review.openstack.org/#/c/39936/
I found the root cause is that the instanceType the Filter uses to
filter the host is the one the instance currently belong to, instead
of the one the instance is going to resize to.
The following files indicates the clue of this issue
https://github.com/openstack/nova/blob/master/nova/conductor/manager.py#L646
https://github.com/openstack/nova/blob/master/nova/scheduler/utils.py#L38
https://github.com/openstack/nova/blob/master/nova/scheduler/filter_scheduler.py#L283
https://github.com/openstack/nova/blob/master/nova/scheduler/filter_scheduler.py#L310
https://github.com/openstack/nova/blob/master/nova/scheduler/filters/disk_filter.py##L36
To manage notifications about this bug go to:
https://bugs.launchpad.net/nova/+bug/1219761/+subscriptions