Re: vNICs and ping connectivity

Kernel-Panic · ‎03-27-2019

Hello all, in my UCS environment I've created two iSCSI vNICs for connecting to our Nimble storage; in the Nimble web console it can see the two initiator addresses are connected, however in the Nimble cmd line console I can't ping the IP addresses associated with these vNICs - is this normal i.e. do vNICs not usually reply to pings?

Kirk J · ‎03-27-2019

Has the Host OS booted up already? The iSCSI optionROM is only active until the OS boots, at which time the software iSCSI driver takes over.

If you are still at boot time, or it isn't booting, see https://www.cisco.com/c/en/us/support/docs/servers-unified-computing/ucs-manager/116003-iscsi-ucs-config-00.html, and run the iscis ping tests from the initiators:

Sample from that doc:

F340-31-13-FI-1-A# connect adapter 1/1/1 
adapter 1/1/1 # connect
No entry for terminal type "vt220";
using dumb terminal settings.
 
adapter 1/1/1 (top):1# attach-mcp
No entry for terminal type "vt220";
using dumb terminal settings.
 
adapter 1/1/1 (mcp):1# iscsi_get_config
 
vnic iSCSI Configuration:
----------------------------
 
 
vnic_id: 5
          link_state: Up
 
       Initiator Cfg:
     initiator_state: ISCSI_INITIATOR_READY
initiator_error_code: ISCSI_BOOT_NIC_NO_ERROR
                vlan: 0
         dhcp status: false
                 IQN: iqn.2013-01.com.myserver124
             IP Addr: 14.17.170.2
         Subnet Mask: 255.255.255.0
             Gateway: 14.17.170.254
 
          Target Cfg:
          Target Idx: 0
               State: ISCSI_TARGET_READY
          Prev State: ISCSI_TARGET_DISABLED
        Target Error: ISCSI_TARGET_NO_ERROR
                 IQN: iqn.1992-08.com.netapp:sn.1111111
             IP Addr: 14.17.10.13
                Port: 3260
            Boot Lun: 0
          Ping Stats: Success (9.990ms)

Kernel-Panic · ‎03-27-2019

Hello Kirk thanks for the reply, the OS is not installed on the blade yet however the problem was down to incorrect router config on my vlan interfaces, I can now ping the iSCSI vNICs from the Nimble

Kernel-Panic · ‎03-27-2019

Scratch last reply, was pinging wrong IP addresses.

I still can't ping the iSCSI vNICs from either the Nimble or the Nexus switch, though the Nexus does at least see the MAC addresses of the vNICs.

Kirk J · ‎03-27-2019

Can you paste the output from your adapter connection, and the iscsi_get_config ?

Please keep in mind that iSCSI boot vnics need their intended vlan to be marked as Native in the service profile for it to work, as the optionROM level objects don't have a concept of tagging vlans (PXE is the same way).

Thanks,

Kirk...

Kernel-Panic · ‎03-27-2019

Thanks:

vnic iSCSI Configuration:
----------------------------


vnic_id: 7
          link_state: Up

       Initiator Cfg:
     initiator_state: ISCSI_INITIATOR_READY
initiator_error_code: ISCSI_BOOT_NIC_NO_ERROR
                vlan: 0
         dhcp status: false
                 IQN: iqn.1992-08.com.cisco:ucs-host:0
             IP Addr: 192.168.80.1
         Subnet Mask: 255.255.255.0
             Gateway: 192.168.80.250

          Target Cfg:
          Target Idx: 0
               State: INVALID
          Prev State: ISCSI_TARGET_GET_LUN_INFO
        Target Error: ISCSI_TARGET_GET_HBT_ERROR
                 IQN: iqn.2007-11.com.nimblestorage:boot-esxi-host-01-v1675a3b2c1eb319c.00000016.d79e922e
             IP Addr: 192.168.80.248
                Port: 3260
            Boot Lun: 1
          Ping Stats: Success (10.826ms)


vnic_id: 8
          link_state: Up

       Initiator Cfg:
     initiator_state: ISCSI_INITIATOR_READY
initiator_error_code: ISCSI_BOOT_NIC_NO_ERROR
                vlan: 0
         dhcp status: false
                 IQN: iqn.1992-08.com.cisco:ucs-host:14
             IP Addr: 192.168.81.1
         Subnet Mask: 255.255.255.0
             Gateway: 192.168.81.250

          Target Cfg:
          Target Idx: 0
               State: INVALID
          Prev State: ISCSI_TARGET_GET_LUN_INFO
        Target Error: ISCSI_TARGET_GET_HBT_ERROR
                 IQN: iqn.2007-11.com.nimblestorage:boot-esxi-host-01-v1675a3b2c1eb319c.00000016.d79e922e
             IP Addr: 192.168.81.248
                Port: 3260
            Boot Lun: 1
          Ping Stats: Success (9.854ms)
adapter 1/1/1 (mcp):2#

On the Nexus switches I have two VLAN interfaces:

Switch 1: (VLAN 80, 192.168.80.250 / 255.255.255.0)

Switch 2: (VLAN 81, 192.168.81.251 / 255.255.255.0)

From either switch I can ping the Nimble iSCSI target addresses (192.168.80.248 and 192.168.81.248) but neither from the switch nor the Nimble can I ping the iSCSI vNIC initiator addresses (192.168.80.1 and 192.168.81.1)

Kirk J · ‎03-27-2019

Hmmm...

Initiators seem to be able to reach the storage controller targets just fine over TCP port 3260.

Are the LUNs configured on the Nimble to map to #1 as the UCS iscsi config appears to show configured (default is usually 0 )

ISCSI_TARGET_GET_HBT_ERROR usually means: the initiator is not registered on the storage controller or the LUN # cannot be found (doesn't match between the storage controller config and the boot target definition).

Kirk...

Kernel-Panic · ‎03-27-2019

Thanks, I've changed LUN ID from 1 to 0, now the output is error free:

vnic iSCSI Configuration:
----------------------------


vnic_id: 7
          link_state: Up

       Initiator Cfg:
     initiator_state: ISCSI_INITIATOR_READY
initiator_error_code: ISCSI_BOOT_NIC_NO_ERROR
                vlan: 0
         dhcp status: false
                 IQN: iqn.1992-08.com.cisco:ucs-host:0
             IP Addr: 192.168.80.1
         Subnet Mask: 255.255.255.0
             Gateway: 192.168.80.250

          Target Cfg:
          Target Idx: 0
               State: ISCSI_TARGET_READY
          Prev State: ISCSI_TARGET_DISABLED
        Target Error: ISCSI_TARGET_NO_ERROR
                 IQN: iqn.2007-11.com.nimblestorage:boot-esxi-host-01-v1675a3b2c1eb319c.00000016.d79e922e
             IP Addr: 192.168.80.248
                Port: 3260
            Boot Lun: 0
          Ping Stats: Success (9.946ms)

        Session Info:
          session_id: 0
         host_number: 0
          bus_number: 0
           target_id: 0


vnic_id: 8
          link_state: Up

       Initiator Cfg:
     initiator_state: ISCSI_INITIATOR_READY
initiator_error_code: ISCSI_BOOT_NIC_NO_ERROR
                vlan: 0
         dhcp status: false
                 IQN: iqn.1992-08.com.cisco:ucs-host:14
             IP Addr: 192.168.81.1
         Subnet Mask: 255.255.255.0
             Gateway: 192.168.81.250

          Target Cfg:
          Target Idx: 0
               State: ISCSI_TARGET_READY
          Prev State: ISCSI_TARGET_DISABLED
        Target Error: ISCSI_TARGET_NO_ERROR
                 IQN: iqn.2007-11.com.nimblestorage:boot-esxi-host-01-v1675a3b2c1eb319c.00000016.d79e922e
             IP Addr: 192.168.81.248
                Port: 3260
            Boot Lun: 0
          Ping Stats: Success (10.342ms)

        Session Info:
          session_id: 0
         host_number: 0
          bus_number: 0
           target_id: 0

Although I can't ping the iSCSI vNICs, the Nimble web console seems to think they're connected:

It's just odd that the IP addresses aren't pingable from either the Nimble or the Nexus switches - perhaps they don't respond to ICMP?

Kirk J · ‎03-27-2019

What Firmware package is applied to the blade/integrated Rack server?

I'll see if I can test that in the lab.

Could be expected behavior,although I don't recall seeing that documented anywhere.

Kirk...

Kernel-Panic · ‎03-27-2019

Thanks, this is old equipment that was used in production but has been wiped and i'm now attempting to build a test environment with it:

Switches:

Cisco Nexus 5010 Chassis ("20x10GE/Supervisor") version 5.2(1)N1(9a)

Fabric Interconnects:

6120XP, version 5.2(3)N2(2.28l)

Blades:

Cisco UCS B200 M2, firmware package 2.2(8l)

Kirk J · ‎03-28-2019

Tested with B200M3 and 3.x firmware, and I did not get OptionROM iSCSI initiator icmp responses until OS(VMware in my case) VMK software initiator took over.

I would say that's probably by design.

Kirk...

Kernel-Panic · ‎03-28-2019

Thanks Kirk, I've just installed the VMware OS and can confirm your findings as I can now ping the iSCSI initiator IP addresses.

I do have another related problem - after installing the VMware OS I configured the management network

VLAN ID: 28

IP: 192.168.28.1

Subnet mask: 255.255.255.0

Default router: 192.168.28.250

On my Nexus 5K switches I created vlan 28 with the IP address of 192.168.28.250; however, similar to the iSCSI issue, I can't ping the VMware management interface address (192.168.28.1) from the switch and in ESXi when I run a management network test it can't ping the default router (192.168.28.250).

Do you have any ideas?

Thanks.

Kirk J · ‎03-28-2019

Which VMK interface is the problem interface you are having issues with pinging in VMware, which VMNIC is it using, and with underlying vnic #.

Just wanted to make sure your iSCSI BFS overlay is not tied to the first vnic.

Kirk...

Kernel-Panic · ‎03-28-2019

To be honest I've probably buggered up the vmnic selection for the management interface, the choices were as below:

vmnic0 - status:connected

vmnic1 - status: vSwitch0

vmnic2 - status: iSCSIvSwitch

vmnic3: - status: vSwitch0

I chose both vmnic1 and vmnic3 for the management interface but I'm guessing I should've chose vmnic0 :)

Kirk J · ‎03-28-2019

Keep in mind the VMK0 mgmt interface inherits the mac address usually from the first vmnic/vnic it is bound with initially.

Also, your vnic that carries the iSCSI traffic, is likely using native vlan . Just wanted to confirm your vnics and VMNICS number is really what you think it is.

Compare MAC address in the service profile vs the physical NIC list in vmware, to make sure vnic0=vmnic0, vmnic1=vmnic1, etc, else you will have difference in vlans ucs is allowing vs what vmware is tagging for.

Are you learning MAC address for VMK0 interface on the right vlan at UCSM level?

ssh UCSM IP

#connect nxos a|b

nxos#show mac address-table | in aaaa (where aaaa is last 4 characters of VMK0 mac address)

Kirk...