Weird Routing Problem ?

[Mike McWethy]

Ok, I have been trying to sort out an issue that we have been experiencing recently, and I am not having much luck. Here is the basis of the problem:

At our DR site, we have two 3750G switches that have several vlans including 200 and 207. Vlan 207 is used for our Dell Equallogic storage array and Vlan 200 is our primary Vlan. I am using EIGRP and static routes. We have four ESX hosts. We have three VM's running on Vlan 200. One VM on ESX host 2 can ping and traceroute successfully with no dropped packets. One VM on ESX host 3 can traceroute fine but drops every two packets so I get two successful, two time outs, two successful, two time outs, and so on. The third VM on ESX host 2 cannot ping at all UNLESS I do a traceroute first, then the pings are 100% successful. The individual switchports for Vlan 207 have all been configured exactly the same as suggested by the Dell Equallogic best practice guide.

I know you will probably need some configurations and will have plenty of requests for more information that I can kindly provide. Does anyone have ANY thoughts as to what may be causing these problems and inconsistencies?

Mike

[gfcisco31]

Hi Mike,

You do have a weird problem indeed. Let me see if i can help, and if i understood exactly what you have.

FIrst:

"I am using EIGRP and static routes"

WHy do you do you need these static entries ? What is the correlation with the VM's communication ?

Second:

" We have three VM's running on Vlan 200. One VM on ESX host 2 can ping and traceroute successfully with no dropped packets. One VM on ESX host 3 can traceroute fine but drops every two packets so I get two successful, two time outs, two successful, two time outs, and so on. The third VM on ESX host 2 cannot ping at all UNLESS I do a traceroute first, then the pings are 100% succesful"

Looks like you have each VM on a particular host, (each connected to a different port all in vlan 200.  RIght ?      Unless some static entries in one of the servers are meesed up, you shouldn't have this behaviour.  IF you have this scenario, please check the hosts entries.

ON the traceroute behavior, i would think as a bug on the VM or maybe on the switch, i never seen something like this to be honest!   CAuse... ARP table on the switch should be populated during the ICMP proccess and the traceroute proccess

PLease, explain more if you can about your scenario, this way i can understand better what is going on, thus help you better ;)

Cheers

[Mike McWethy]

The problem has been resolved. I was missing a route in one of my routers. Silly me