Bug 1318060

Summary: Unable to utilize SR-IOV on multiple drivers or after changing drivers
Product: [oVirt] ovirt-engine Reporter: Jay Turner <jturner>
Component: BLL.NetworkAssignee: Dan Kenigsberg <danken>
Status: CLOSED INSUFFICIENT_DATA QA Contact: Meni Yakove <myakove>
Severity: high Docs Contact:
Priority: unspecified    
Version: 3.6.3.3CC: bugs, djorm, jturner, ylavi
Target Milestone: ---Flags: danken: needinfo? (jturner)
ylavi: needinfo? (djorm)
rule-engine: planning_ack?
rule-engine: devel_ack?
rule-engine: testing_ack?
Target Release: ---   
Hardware: x86_64   
OS: Linux   
Whiteboard:
Fixed In Version: Doc Type: Bug Fix
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2016-04-14 07:43:00 UTC Type: Bug
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: Network RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:

Description Jay Turner 2016-03-15 22:26:26 UTC
Description of problem:
Two scenarios which seem to be related:
1) Have multiple SR-IOV enabled devices in a machine and attempt to utilize both device types (say having a Mellanox ConnectX-3 - mlx4 driver, along with an Intel XL710 - i40e driver).  oVirt will only recognize/utilize one of the two drivers.  For example, assigning the XL710 virtual function to a guest will result in the mlx4 driver being loaded in the guest as opposed to the i40e.
2) Once oVirt is installed with one SR-IOV device driver, you are not able to trade out for another device driver.  For example start off with the Mellanox device, then remove that device, clear everything out of oVirt, then put the Intel device in the host and attempt to utilize those VFs on new guests . . . oVirt will still assign/load the mlx4 driver even though the Mellanox card is no longer in the host.

In both scenarios, the only way to remedy the situation is to reinstall oVirt on the host.

Version-Release number of selected component (if applicable):
oVirt Engine Version: 3.6.3.4-1.el7.centos

How reproducible:
Always

Steps to Reproduce:
1. See above
2.
3.

Actual results:


Expected results:


Additional info:

Comment 1 Dan Kenigsberg 2016-03-16 09:34:18 UTC
May I ask for a more detailed report, with smaller steps, as a I do not understand what have you done and which driver is wrongly selected.

Can you supply Vdsm logs with the output of getCapabilities and the logs until all VMs are up? In problem 1, do you run the same VM twice, connected to a different PF each time? Or are you running two different VMs concurrently?

can you provide the output of `lspci -vv` from within the VMs?

Comment 2 Dan Kenigsberg 2016-04-14 07:29:46 UTC
Please reopen when the requested information is available.