VMware Cloud Community
CKF1028
Enthusiast
Enthusiast

Nested ESXi 5.1 Duplicate ICMP Response

Hello,

My infrastructure of nested ESXi 5.1 is as below and everything is working fine.(vMotion, HA/DRS, iSCSI/NFS connect to physical storage and so on)

But there's a strange problem between x1 pESXi and x2 vESXi about ICMP response, please give me some suggestions to solve it, thanks !!!

x1 pESXi (Management IP=172.16.10.10)

     Standard vSwitch2

     Uplinks x4 which are teaming together with IP-hash to Cisco switch (etherchannel and trunk)

     Portgroups x5 (MGMT, STOR, vMOT, FT, PROD) with it's own vlan id.

x2 vESXi (Management IP=172.16.10.111 & 172.16.10.222)

There're x5 Standard vSwitch which are configured as follows.

     Name: vSwitch0

     Uplinks x2 which are teaming together with IP-hash to pESXi vSwitch.

     Portgroups: MGMT

When pESXi ping vESXI vmkernel IP address with each other, the duplicate ICMP response will be occurred.

~ # ping 172.16.10.111

PING 172.16.10.111 (172.16.10.111): 56 data bytes

64 bytes from 172.16.10.111: icmp_seq=0 ttl=64 time=0.960 ms

64 bytes from 172.16.10.111: icmp_seq=1 ttl=64 time=0.525 ms (DUP!)

64 bytes from 172.16.10.111: icmp_seq=1 ttl=64 time=0.609 ms

64 bytes from 172.16.10.111: icmp_seq=2 ttl=64 time=0.531 ms (DUP!)


Reply
0 Kudos
12 Replies
a_p_
Leadership
Leadership

How did you configure the EtherChannel? Take a look at http://kb.vmware.com/kb/1004048 for restrictions (i.e. "mode on"). Unless there's an absolute/special requirement for EtherChannel, I'd suggest you remove it an go with default network settings.

André

Reply
0 Kudos
CKF1028
Enthusiast
Enthusiast

Thanks for reply from André

The switch EtherChannel configuration isn't be misconfigured, 'cause pESXi can communicate with LAN / WAN or storage devices normally.

This strange problem can be solved by setting x2 uplinks to x1 uplink of vESXi, then pESXi and vESXi can ping normally with each other.

But, I'm really want to know whether the vSwitch of vESXi can be configured NIC teaming (IP-Hash or portID) to pESXi vSwitch or not.

Reply
0 Kudos
a_p_
Leadership
Leadership

Seems I missed this in the original post. Afaik vSwitches do not support teaming on the port groups, i.e. for virtual machines. Setting the nested ESXi host's network back to defaults (port based) should solve the issue.

André

CKF1028
Enthusiast
Enthusiast

This problem had been solved, appreciate for André answers !!!  :smileygrin:


vESXi vSwitch do not support teaming on the port groups, it should be setting back to "Routed based on the originating virtual port ID"

Reply
0 Kudos
CKF1028
Enthusiast
Enthusiast

After changing both pESXi and vESXi vSwitch to default setting, the duplicate ICMP response between pESXi and vESXi be solved.

But when I try to ping between vESXi, the duplicate ICMP response occur.

vESXi-01

     Name: vSwitch0

     Uplinks x2 which load balancing are setting "port ID" to pESXi vSwitch2-portgroup-MGMT

     Portgroups: MGMT (IP=172.16.10.111)

     Name: vSwitch1

     Uplinks x2 which load balancing are setting "port ID" to pESXi vSwitch2-portgroup-STOR

     Portgroups: STOR (IP=172.16.20.111)

vESXi-02

     Name: vSwitch0

     Uplinks x2 which load balancing are setting "port ID" to pESXi vSwitch2-portgroup-MGMT

     Portgroups: MGMT (IP=172.16.10.222)

     Name: vSwitch1

     Uplinks x2 which load balancing are setting "port ID" to pESXi vSwitch2-portgroup-STOR

     Portgroups: STOR (IP=172.16.20.222)


From vESXi-01 to ESXi-02

~ # ping 172.16.10.111

PING 172.16.10.111 (172.16.10.111): 56 data bytes

64 bytes from 172.16.10.111: icmp_seq=0 ttl=64 time=0.403 ms

64 bytes from 172.16.10.111: icmp_seq=1 ttl=64 time=0.435 ms

64 bytes from 172.16.10.111: icmp_seq=2 ttl=64 time=0.461 ms


~ # ping 172.16.20.111

PING 172.16.20.111 (172.16.20.111): 56 data bytes

64 bytes from 172.16.20.111: icmp_seq=0 ttl=64 time=1.047 ms

64 bytes from 172.16.20.111: icmp_seq=0 ttl=64 time=1.141 ms (DUP!)

64 bytes from 172.16.20.111: icmp_seq=0 ttl=64 time=1.165 ms (DUP!)

64 bytes from 172.16.20.111: icmp_seq=0 ttl=64 time=1.186 ms (DUP!)


Reply
0 Kudos
a_p_
Leadership
Leadership

What subnet mask do you use for the 172.16.x.x networks?

André

Reply
0 Kudos
CKF1028
Enthusiast
Enthusiast

172.16.10.0/24 for MGMT

172.16.20.0/24 for STOR

Reply
0 Kudos
a_p_
Leadership
Leadership

Can you please temporarily disconnect one of the vSwitch0 uplinks for each of the vHosts to see whether this makes a difference?

André

Reply
0 Kudos
CKF1028
Enthusiast
Enthusiast

Both vESXi-01 and vESXi-02 have changed to x1 uplink as below, but the duplicate ICMP issue still exist.

vESXi.jpg

Reply
0 Kudos
a_p_
Leadership
Leadership

"10000 Full"!? Which virtual adapter do you use? Not sure whether this causes the issue, but afaik only the E1000 adapter is supported for virtual ESXi hosts.

André

Reply
0 Kudos
CKF1028
Enthusiast
Enthusiast

Thanks André remind me this issue, I'm using vmxnet3 not E1000 adapter.


But, why vESXi-02 ping vESXi-01 management IP=172.16.10.111 didn't appear duplicate ICMP response ?!

Why vESXi-02 ping vESXi-01 vmkernel IP=172.16.20.111 will appear duplicate ICMP respnse ?!


I'll go to change vmxnet3 to E1000 for testing this again, and report the result here.


Reply
0 Kudos
CKF1028
Enthusiast
Enthusiast

The result is the same !!! :S

I've tried x1 or x2 uplinks (E1000) within vESXi-01&02 vSwitch, the result is being the same.

From vESXi-02 ping to vESXi-01

ping 172.16.10.111 ---> it's ok !!!

ping 172.16.20.111 ---> it occurs duplicate ICMP response.

vESXi.jpg

Reply
0 Kudos