Gate occasionally failed with nova libvirt error

[hongbin]

There is an occasionally failure on the gate. Here is a sample: http://logs.openstack.org/06/251206/1/check/gate-functional-dsvm-magnum-k8s/8c67af8/

The root cause is from nova/libvirt. In particular, below is the error (full log http://logs.openstack.org/06/251206/1/check/gate-functional-dsvm-magnum-k8s/8c67af8/logs/screen-n-cpu.txt ).

2015-11-30 12:59:51.920 12631 ERROR nova.compute.manager [instance: 3cdf2482-5c32-4ab2-9400-774537ad50d4] libvirtError: internal error: process exited while connecting to monitor: Cannot set up guest memory 'pc.ram': Cannot allocate memory

[Vilobh Meshram (vilobhmm)]

Changing it to Nova for now for someone in Nova to have a look. Feel free to move back to Magnum if it seems to be a Magnum issue.

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/255378
Committed: https://git.openstack.org/cgit/openstack/magnum/commit/?id=958e63341a3a9dbf9a10d5c0bf5dcf6145a78bd5
Submitter: Jenkins
Branch: master

commit 958e63341a3a9dbf9a10d5c0bf5dcf6145a78bd5
Author: Hongbin Lu <email address hidden>
Date: Wed Dec 9 11:50:06 2015 -0500

    Gate: Attempt to fix a memory allocation error

    This random error is due to the different swap space at different
    test hosts. In particular, there is a provider who provides VMs with
    0 swap space, while other providers provide VMs with 8G swap space.
    As a result, Magnum jobs failed at VMs with 0 swap space.

    This commit reduces memory consumption of Magnum jobs by reducing the
    bay node flavor from 2G to 1G. I consider this as a temporary fix.
    A permanent fix should be from infra side to make the swap space
    consistent across test nodes.

    Partial-Bug: #1521237
    Change-Id: Ia043fab1f7665f7299e4094634e02c42b43dcce1

[Spyros Trigazis (strigazi)]

I don't think this bug is valid any more.