This is a message from pci stats, it means scheduler select a host actually not meeting the request. But make sure you do enable the PCI filter in the scheduler first.
Yongli He -----Original Message----- From: yujie [mailto:[email protected]] Sent: Friday, January 8, 2016 12:43 PM To: OpenStack Development Mailing List (not for usage questions) <[email protected]> Subject: [openstack-dev] [sriov] Could create sriov vm on one host successfully but failed on another host Hi all, I have two hosts. Host1 run services of controller/compute/network, and host2 run services of compute. The network type is vlan. When creating port with direct type and booting a vm using this port on host2, the vm works well. But when create the vm on host1 using same way, the vm state is error. And the nova-compute log says: ERROR nova.pci.stats [req-4af9b36d-8a0c-43d5-8535-fbf9e8d1b857 06172a48a7254dda934a729f4a58d2ec 894880b2f31046d7b8a83bc3783df070 - - -] Failed to allocate PCI devices for instance. Unassigning devices back to pools. This should not happen, since the scheduler should have accurate information, and allocation during claims is controlled via a hold on the compute node semaphore Any suggestion will be grateful. Thanks. Yu __________________________________________________________________________ OpenStack Development Mailing List (not for usage questions) Unsubscribe: [email protected]?subject:unsubscribe http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev __________________________________________________________________________ OpenStack Development Mailing List (not for usage questions) Unsubscribe: [email protected]?subject:unsubscribe http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
