Bug 1383397 - SR-IOV number of VFs is not persistent after reboot
Summary: SR-IOV number of VFs is not persistent after reboot
Keywords:
Status: CLOSED CURRENTRELEASE
Alias: None
Product: Red Hat OpenStack
Classification: Red Hat
Component: rhosp-director
Version: 10.0 (Newton)
Hardware: x86_64
OS: Linux
medium
high
Target Milestone: rc
: 10.0 (Newton)
Assignee: Angus Thomas
QA Contact: Eyal Dannon
URL:
Whiteboard:
Depends On:
Blocks: 1242903
TreeView+ depends on / blocked
 
Reported: 2016-10-10 14:15 UTC by Eyal Dannon
Modified: 2016-12-16 16:51 UTC (History)
19 users (show)

Fixed In Version:
Doc Type: If docs needed, set a value
Doc Text:
Clone Of:
Environment:
Last Closed: 2016-12-16 16:51:25 UTC
Target Upstream Version:


Attachments (Terms of Use)

Description Eyal Dannon 2016-10-10 14:15:30 UTC
Description of problem:
Number of VFs configured over SR-IOV NIC resets after a reboot.

On OSPD10 + SR-IOV setup the number of VFs should be static and remain after a reboot.


Version-Release number of selected component (if applicable):
OSPD10, build - 2016-10-06.1.


How reproducible:
Always

Steps to Reproduce:
1.Install OSPD
2.Use SR-IOV post install yaml[http://skvs.etherpad.corp.redhat.com/integration-sriov ]
3.cat /sys/class/net/p6p1/device/sriov_numvfs  [returns 0]

Actual results:
numvfs returns 0

Expected results:
Should return the value we configured in the post deploy script

Additional info:

Comment 4 Saravanan KR 2016-10-14 06:33:21 UTC
On reboot, the sriov_numvfs for that interface will be set only when "ifup-local p6p1" is invoked. By default, this interface will be NM controlled, which will fail to call ifup-local. Adding below network config for the p6p1 interface in the compute node, will trigger the ifup-local script for this interface (on reboot)

            -
              type: interface
              name: p6p1
              use_dhcp: false
              defroute: false


Verified in the same setup provided by Maxim.

Comment 5 Saravanan KR 2016-10-14 06:34:49 UTC
(In reply to Saravanan KR from comment #4)
> Verified in the same setup provided by Maxim.
Setup provided by Eyal.

Comment 6 Brent Eagles 2016-10-14 15:54:41 UTC
I've been working on the neutron SR-IOV agent angle to this. Just for information purposes, see https://bugzilla.redhat.com/show_bug.cgi?id=1369768 #c12 for some notes on how I enabled VF allocation on reboot/PF allocation/release.

Comment 7 Eoghan Glynn 2016-10-15 19:23:13 UTC
On the basis of discussion with owalsh at the 2016-10-14 nova triage call, this looks like it's related to SR-IOV integration in OSP-director implemented by skramaja.

Hence reassigning to the NFV DFG. Feel free to assign back to DFG:Compute is the above conclusion is not correct.

Comment 8 Vijay Chundury 2016-10-18 07:21:19 UTC
This change or network configuration template is for the operator (documentation). So if somebody lets me know where this has to be placed will do it. Right request Angus to  move bug status to ON_QA to verify the same.

Comment 10 Eyal Dannon 2016-10-20 08:54:33 UTC
I have verified SR-IOV feature with OSPD10.

I suggest to add Saravanan's Fix(Comment 4) to the documentation.


Note You need to log in before you can comment on or make changes to this bug.