PCI whitelist exception causes the resource tracker to stop and will not allow us to spawn further SR-IOV/PCIPT VMs when SR-IOV PF is assigned to a VM.
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
OpenStack Compute (nova) |
Fix Released
|
High
|
MANJUNATH PATIL |
Bug Description
Encountered an exception in the pci whitelist causes the resource tracker to stop and blocks user/admin to spawn further VMs
we have the following pci_whitelist to support both SRIOV and PCIPT on
pci_passthrough
[{"devname": "eth1", "physical_network": "physnet1"},
{"physical_
{"physical_
Once we boot the PCI passthrough VM on physnet1 using eth1,
the device eth1 no longer available to hypervisor.
So when we try to boot another PCI passthrough VM using eth2,
the current code tries to validate the pci_whitelist and
throws an error saying - device eth1 is not found.
This is because pci_whitelist has devname eth1 and
code tries to get the PCI address of the device which is not available.
We also found that with the above mentioned pci_whitelist,
as soon as we boot a PCI passthrough VM, the periodic resource
tracker also stops. We further analysed and found that any
misconfiguration of pci_whitelist could cause periodic
resource tracker to stop.
We get the following error in the nova compute log if eth1 is not present. But compute still shows up and the periodic hypervisor update stops working.
2016-07-13 09:22:42.146 28800 ERROR nova.compute.
2016-07-13 09:22:42.146 28800 ERROR nova.compute.
2016-07-13 09:22:42.146 28800 ERROR nova.compute.
2016-07-13 09:22:42.146 28800 ERROR nova.compute.
2016-07-13 09:22:42.146 28800 ERROR nova.compute.
2016-07-13 09:22:42.146 28800 ERROR nova.compute.
2016-07-13 09:22:42.146 28800 ERROR nova.compute.
2016-07-13 09:22:42.146 28800 ERROR nova.compute.
2016-07-13 09:22:42.146 28800 ERROR nova.compute.
2016-07-13 09:22:42.146 28800 ERROR nova.compute.
2016-07-13 09:22:42.146 28800 ERROR nova.compute.
2016-07-13 09:22:42.146 28800 ERROR nova.compute.
2016-07-13 09:22:42.146 28800 ERROR nova.compute.
2016-07-13 09:22:42.146 28800 ERROR nova.compute.
2016-07-13 09:22:42.146 28800 ERROR nova.compute.
2016-07-13 09:22:42.146 28800 ERROR nova.compute.
2016-07-13 09:22:42.146 28800 ERROR nova.compute.
2016-07-13 09:22:42.146 28800 ERROR nova.compute.
2016-07-13 09:22:42.146 28800 ERROR nova.compute.
2016-07-13 09:22:42.146 28800 ERROR nova.compute.
2016-07-13 09:22:42.146 28800 ERROR nova.compute.
2016-07-13 09:22:42.146 28800 ERROR nova.compute.
Changed in nova: | |
assignee: | nobody → MANJUNATH PATIL (mpatil) |
Changed in nova: | |
importance: | Undecided → High |
Fix proposed to branch: master /review. openstack. org/345925
Review: https:/