logging pci_devices from the resource tracker is kind of terrible

[Matt Riedemann]

The _report_hypervisor_resource_view method in the resource tracker on the compute node is logging pci devices (if set in the resources dict from the virt driver).

I have a compute node with libvirt 1.2.2 with several hundred devices:

http://paste.openstack.org/show/542185/

Those get logged every TWICE every 60 seconds (by default) because of the update_available_resource periodic task in the compute manager.

We should at the very least only log the giant dict of pci devices once in _report_hypervisor_resource_view, or maybe not at all.

[Matt Riedemann (mriedem)]

For example you can see the duplicate logging here:

http://logs.openstack.org/31/342231/1/check/gate-tempest-dsvm-full/bbb67ff/logs/screen-n-cpu.txt.gz#_2016-07-14_16_26_28_524

[Matt Riedemann (mriedem)]

Patch: https://review.openstack.org/#/c/347576/

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/347576
Committed: https://git.openstack.org/cgit/openstack/nova/commit/?id=0f790181f40cb3e3ca0ae10f2293777b8c32bd9a
Submitter: Jenkins
Branch: master

commit 0f790181f40cb3e3ca0ae10f2293777b8c32bd9a
Author: Matt Riedemann <email address hidden>
Date: Tue Jul 26 19:02:23 2016 -0400

    rt: don't log pci_devices twice when updating resources

    By default we update available resources on the compute every
    60 seconds. The _report_hypervisor_resource_view method is
    logging pci devices twice. On a compute with hundreds of pci
    devices this blows up the logs in a short amount of time.

    This change removes at least the duplicate pci device logging
    but we might want to even consider just not logging these in
    each iteration. I've opted to at least push the simple fix for
    now.

    Change-Id: Id05bfef44b1108dec286486d42516649dd0683e9
    Closes-Bug: #1606718

[OpenStack Infra (hudson-openstack)]

This issue was fixed in the openstack/nova 14.0.0.0b3 development milestone.