Issue with Nexus 1000v iSCSI multipathing

[malammouri2015]

Hello Experts,

We have configured port-profile in Nexus 1000v switch to be used for VMKs that will binded to iSCSI software vmhba adapter.

This iSCSI vmhba adapter is part of EMC  (RecoverPoint for VM) setup. Recoverpoint will have two virtual appliances ( vRPAs ) that will have two interfaces used for iSCSI connections with ESXi host to access Virtual machine storage.

The issue we have is that when we place vRPAs on ESXi host, we cannot ping vRPA iSCSI interface from the same ESXi host using the VMK binded to vmhba. If we do the same ping from other ESXi hosts, it works fine.

for example:

vRPA-1 ( on ESXi-1):  int01: x.x.x.203 / int02: x.x.x.x.205

vRPA-2 ( on ESXi-2): int01: x.x.x.204 / int02: x.x.x.x.206

ESXi-1 VMK4 ( used for iSCSI and inhering iSCSI port-profile): x.x.x.101

ESXi-2 VMK4 ( used for iSCSI and inhering iSCSI port-profile): x.x.x.102

esxi-1#vmkping -I vmk4 x.x.x.203 ( fails )

esxi-1#vmkping -I vmk4 x.x.x.205 ( fails )

esxi-1#vmkping -I vmk4 x.x.x.204 ( pass)

esxi-1#vmkping -I vmk4 x.x.x.206 ( pass)

esxi-2#vmkping -I vmk4 x.x.x.203 ( pass)

esxi-2#vmkping -I vmk4 x.x.x.205 ( pass)

esxi-2#vmkping -I vmk4 x.x.x.204 ( fail)

esxi-2#vmkping -I vmk4 x.x.x.206 ( fail)

This is consistent across all hosts wherever we move the vRPA , we cannot reach vRPA iSCSI interfaces from the same host ( we loose paths to this vRPA, but we still can ping the other vRPA on different host ).

VMKs port-profile has been configured as per Cisco documentation :

http://www.cisco.com/c/en/us/td/docs/switches/datacenter/nexus1000/sw/4_2_1_s_v_1_5_1/system_managem...

as below:

port-profile type vethernet iscsi-profile

vmware port-group

switchport mode access

switchport access vlan x

capability iscsi-mulipathing

state enable

system vlan x

no shut

attached screenshot shows part of troubleshooting steps done:

vRPAs on ESXi01

checking the ARP table of ESXi , I can see the IP addresses of both iSCSI interface of VRPAs:

x.x.x.203 ( vRPA-1 interface-1)

x.x.x.205 ( vRPA-1 interface-2)

x.x.x.204 ( vRPA-1 interface-1)

x.x.x.206 ( vRPA-2 interface-2)

vRPA-1: ( on ESXi-01)

x.x.x.203

x.x.x.205

( Not pingable from ESXi-1 ( same host ) VMK, but pingable from VMKs of all other hosts )

vRPA-2: ( on ESXi-2)

x.x.x.204

x.x.x.206

( Not pingable from ESXi-2 ( same host ) VMK, but pingable from VMKs of all other hosts )

Checking the mac address of vRPA interface on nexus 1000v , I can see that it is local to veth253 ( vethernet port connected to vRPA interface ) on Mod17 ( ESXi-1).

I found the below blog post where we have the same issue , but the suggested fix by removing the (capability iscsi-multipath ) command from port-profile will not work as the vmk won't appear when you try to bind it to the vmhba.

Recoverpoint for VMs (RP4VM) – RAGEinStorage

Any help on this would appreciated.

Thanks,

Mohammad

Message was edited by: Mohammad Alammouri