Details
-
Bug
-
Resolution: Not a Bug
-
Major
-
None
-
4.12
-
-
-
Important
-
No
-
CNF Network Sprint 237
-
1
-
False
-
Description
Description of problem:
SR-IOV operator isn't initializing MT27710 ConnectX-4 NIC attached on few of their nodes. They see this in the config daemon logs DiscoverSriovDevices(): unable to get device mode 0000:01:00.0 "no such device"
Version-Release number of selected component (if applicable):
SR-IOV on OCP v4.12
How reproducible:
Occuring on customer environment
Steps to Reproduce:
1. 2. 3.
Actual results:
NIC can be seen attached but no VFs are up
03:00.0 Ethernet controller [0200]: Mellanox Technologies MT27710 Family [ConnectX-4 Lx] [15b3:1015]
Subsystem: Mellanox Technologies Device [15b3:0009]
Expected results:
VFs should be up as defined in SriovNetworkNodePolicy object
Additional info:
On thing suspicious here is that I can't see SR-IOV capability for this NIC as seen for Intel one. They have turned required settings on from UEFI/BIOS $ egrep -E "SR-IOV|Ether" lspci 01:00.0 Ethernet controller [0200]: Intel Corporation Ethernet Controller 10-Gigabit X540-AT2 [8086:1528] (rev 01) Capabilities: [160 v1] Single Root I/O Virtualization (SR-IOV) 01:00.1 Ethernet controller [0200]: Intel Corporation Ethernet Controller 10-Gigabit X540-AT2 [8086:1528] (rev 01) Capabilities: [160 v1] Single Root I/O Virtualization (SR-IOV) 03:00.0 Ethernet controller [0200]: Mellanox Technologies MT27710 Family [ConnectX-4 Lx] [15b3:1015] 03:00.1 Ethernet controller [0200]: Mellanox Technologies MT27710 Family [ConnectX-4 Lx] [15b3:1015] 82:00.0 Ethernet controller [0200]: Mellanox Technologies MT27710 Family [ConnectX-4 Lx] [15b3:1015] 82:00.1 Ethernet controller [0200]: Mellanox Technologies MT27710 Family [ConnectX-4 Lx] [15b3:1015]
