yahoo-eng-team team mailing list archive

[Preethi Dsilva <preethi.dsilva@xxxxxxx>]

Public bug reported:

VM's go into error state when booted with SRIOV nic

STeps to reproduce:
================
1.enable sriov in the bios in my case I have mellanox card with dual port nic which shows up in the OS as eth4 and eth5
2.provide PCI whitelist in nova.conf
pci_passthrough_whitelist = {"address":"*:04:00.*","physical_network":"physnet1"}
3.the mlx4_core file is set as options mlx4_core port_type_array=2,2 num_vfs=3,3,0 probe_vf=3,3,0 enable_64b_cqe_eqe=0 log_num_mgm_entry_size=-1
4.Its observed that 3 vm's went into eth4 vf's and 3 vm's went into eth5 vf's
the sequence is first vm landed on eth4 vf2 then 2nd on eth4 vf1  both were up with ip assigned.3rd vm landed on eth5 vf5 bt state of VF remained in auto state(if we manually set the state to enable then vm gets IP but nova fails to do so hence vm goes into error state)
5.4th vm landed into eth5 again however nova was able to make state to enable hence the vm got IP
5th vm landed on eth4 vf0 and it gt ip

This pattern is not certain.Every time vm goes into error the logs show the below error 
VirtualInterfaceCreateException: Virtual Interface creation failed

** Affects: nova
     Importance: Undecided
         Status: New

-- 
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/1572831

Title:
  VM's go into error state when booted with SRIOV nic

Status in OpenStack Compute (nova):
  New

Bug description:
  VM's go into error state when booted with SRIOV nic

  STeps to reproduce:
  ================
  1.enable sriov in the bios in my case I have mellanox card with dual port nic which shows up in the OS as eth4 and eth5
  2.provide PCI whitelist in nova.conf
  pci_passthrough_whitelist = {"address":"*:04:00.*","physical_network":"physnet1"}
  3.the mlx4_core file is set as options mlx4_core port_type_array=2,2 num_vfs=3,3,0 probe_vf=3,3,0 enable_64b_cqe_eqe=0 log_num_mgm_entry_size=-1
  4.Its observed that 3 vm's went into eth4 vf's and 3 vm's went into eth5 vf's
  the sequence is first vm landed on eth4 vf2 then 2nd on eth4 vf1  both were up with ip assigned.3rd vm landed on eth5 vf5 bt state of VF remained in auto state(if we manually set the state to enable then vm gets IP but nova fails to do so hence vm goes into error state)
  5.4th vm landed into eth5 again however nova was able to make state to enable hence the vm got IP
  5th vm landed on eth4 vf0 and it gt ip

  This pattern is not certain.Every time vm goes into error the logs show the below error 
  VirtualInterfaceCreateException: Virtual Interface creation failed

To manage notifications about this bug go to:
https://bugs.launchpad.net/nova/+bug/1572831/+subscriptions