Exalogic Virtual: Repeating VNIC Messages ("fip_vnic_rings_create PKey 0x800X not found. Check configuration in SM/BX") Displayed On vServer Console for Existing vServers

Asset ID:	1-72-2124426.1
Update Date:	2017-09-20
Keywords:

Solution Type Problem Resolution Sure

Solution 2124426.1 : Exalogic Virtual: Repeating VNIC Messages ("fip_vnic_rings_create PKey 0x800X not found. Check configuration in SM/BX") Displayed On vServer Console for Existing vServers

Applies to:

Oracle Exalogic Elastic Cloud Software - Version 2.0.6.0.0 and later
Exalogic Elastic Cloud X3-2 Hardware - Version X4 to X4 [Release X4]
Linux x86-64
Oracle Virtual Server x86-64

Symptoms

On Exalogic Virtual racks, when logging into console for an existing vServer using "xm console <UUID> command, the following messages are seen repeating and will not allow any further commands to be executed.

T10875391076 [mlx4_vnic] vnic-1:fip_vnic_rings_create:1243: fip_vnic_rings_create PKey 0x8009 not found. Check configuration in SM/BX
T10875391082 [mlx4_vnic] vnic-1:fip_vnic_fsm:1595: fip_vnic_rings_create failed, closing vnic rc=-19
T10875394075 [mlx4_vnic] vnic-1:fip_vnic_rings_create:1243: fip_vnic_rings_create PKey 0x8009 not found. Check configuration in SM/BX
T10875394079 [mlx4_vnic] vnic-1:fip_vnic_fsm:1595: fip_vnic_rings_create failed, closing vnic rc=-19

Changes

The NN2-GW switch running the Master was rebooted.

Cause

This issue happens due to missing vnics. This can be confirmed by running the following "mlx4_vnic_info -s" command.

# mlx4_vnic_info -s

NETDEV_NAME eth1286_2.682
MAC 00:14:4f:f8:64:8c
VLAN 0x2aa
IOA_PORT mlx4_0:2
BX_NAME exlgw-01
EPORT_NAME 0A-ETH-1

NETDEV_NAME eth1286_2.684
MAC 00:14:4f:f8:c5:00
VLAN 0x2ac
IOA_PORT mlx4_0:2
BX_NAME exlgw-02
EPORT_NAME 0A-ETH-1

NETDEV_NAME eth646_1.684
MAC 00:14:4f:f9:d3:22
VLAN 0x2ac
IOA_PORT mlx4_0:1
BX_NAME exlgw-01
EPORT_NAME 0A-ETH-1

From "mlx4_vnic_info -s" command output we would expect even number of entries as there are 2 slaves per EoIB interface. The example above shows 2 interfaces, with 3 vnics. We would expect 4 vnics in this scenario.

The "showvnics" command on the gateway switches shows that all 4 vnics are present, however 1 vnic is in WAIT_IOA status, which represents the missing vnic from above "mlx4_vnic_info -s" command output. The following is the output of the "showvnics" command from both gateway switches.

Gateway 1

58 WAIT-IOA N 6ABA7FD7C6AD2E1A exlcn-01 EL-C 192.168.10.1 0000 00:14:4F:F8:5A:8C 682 0x8009 0A-ETH-1
52 UP N 6ABA7FD7C6AD2E1A exlcn-01 EL-C 192.168.10.1 0001 00:14:4F:F9:D3:22 684 0x800a 0A-ETH-1

Gateway 2

58 UP N 6ABA7FD7C6AD2E1B exlcn-01 EL-C 192.168.10.1 0000 00:14:4F:F8:64:8C 682 0x8009 0A-ETH-1
52 UP N 6ABA7FD7C6AD2E1B exlcn-01 EL-C 192.168.10.1 0001 00:14:4F:F8:C5:00 684 0x800a 0A-ETH-1

This issue happens to due following known defect which is still under investigation.

Bug 22805940 - pkey index moved after switch reboot

Solution

As a workaround you can issue a stop vServer and start vServer from EMOC. This will re-create the vnics and should solve the issue.

References

<BUG:22805940> - PKEY INDEX MOVED AFTER SWITCH REBOOT
<NOTE:1512139.1> - Oracle Exalogic Elastic Cloud Known Issues - Virtualization Release

Attachments

This solution has no attachment