IBM HS22, 1st NIC cannot get IP. 2nd NIC ok.

[fajarpri]

Hi all,

I encounter strange thing.

ESXi 4.1 on IBM HS22 blade on Chassis-H.

If I install Windows guest, the NIC cannot gets IP from our DHCP. Even though I have installed Vmware tools on the guest.

I found out the workaround is:

- Add a 2nd NIC

- Boot  the guest, 2nd NIC gets IP

- Shutdown guest

- Remove 1st NiC

Has anyone seen this too? What cause it?

Thank you.

[fajarpri]

Ok strange thing....

I found the cause.

For Windows 2008 I must use VMXNET 3 driver

For Windows 2003, use FLEXIBLE driver

I don't know if this is bug or not.

[fajarpri]

Hmm I cancel that.

The network is not consistent. Sometimes if rebooted, it lost connection again.

And this happens even though I set static IP on the network adapter.

Does anyone have this too?

Thank you.

[alefestaedist]

May you give me more info on your conf.? do you have a blade H with brocade,cisco switch? do you have  qlogic, netextreme? expansion cards?

Alex

[fajarpri]

Hi Alex,

Thanks for you reply.

The Chassis has:

Slot1: Nortel Ethernet Switch Module

Slot2: IBM Server Connection Module (unused)

No any other expansion module/card.

The HS22 Blade run on Bay 1-6 on the H chassis.

Do you have similar situation?

Pls let me know if you need more info.

[alefestaedist]

yes I have a couple of customer in this kind of config.. now what I need to know is:

nortel is at the latest firmware?

do you have  specific vlan for your LAN, if it's vlan 1 you found the problem.

In the slot 2 you have a server connection module...is it a passthrough module or a multiswitch module?

Alex

[fajarpri]

1. Latest firmware. I don't think it has the latest firmware if any. But how to update the firmware? The advance management module said that the Nortel doesn't support flashing.

2. VLAN, I don't setup any vlan at all. It's all normal network.

3. I don't know if it's passthrough or SM.

Property	Value
Product Name	IBM Ethernet Server Connectivity Module, 6 GB Ethernet ports
Description	Server Conn Mod
Part Number	39Y9326
FRU Number	39Y9327
FRU Serial No.	ZK11MV68M101
Hardware Revision	1
Manuf. Date	0632
UUID	0017 EF19 F500 0000 0000 0000 0000 0000
Manufacturer	IBM (IBM)
Manuf. ID	20301
Product ID	75

This is the nortel:

Property	Value
Product Name	BNT 6x Layer 2/3 Copper and Fiber GB Ethernet Switch Module
Description	Ethernet SM
Part Number	32R1866
FRU Number	32R1869
FRU Serial No.	YJ1JXF6C7582
Hardware Revision	1
Manuf. Date	5006
UUID	0017 EF73 3100 0000 0000 0000 0000 0000
Manufacturer	NT ()
Manuf. ID	304
Product ID	5

[alefestaedist · Accepted solution]

I would say that is a standard conf but seems you have problem passing through one of the swtich. I would try this, remove (if you can) the  IBM Ethernet Server Connectivity Module thant check wich nic on the esx host are still connected.

Reconfigure the VM nic if needed and try again.

Second thing I am sure you did or you would not have connectivitry at all.. but double check that you have enable external ports from the AAM.

If you set a static ip on the VM does it works? I am wondering if you dont'have some kind of filter ..on the switch.

Alex

[fajarpri]

Thanks Alex.

Things look rather OK now.

Silly me, I didn't connect any network cable on the 2nd network module. I thought all traffic will go through the 1st network module only if the 2nd one is empty.

Now it looks fine. VM can get IP from DHCP.

Although there is one instance when a W2K8 can get IP but cannot do any network traffic. I don't know why. The workaround is to install 2nd NIC in the VM.

I'll check VMware for any ESXi patches on this.

[fajarpri]

I confirm there are cases where the VM can get IP from DHCP but cannot ping other machine.

- The first workaround is to ping that trouble VM from other machine, and they both can ping each other. But this is not always working, and also intermitten problem exists.

- The 2nd workaround which is more reliable is to install a second NIC on that VM.

I still don't know what's wrong. Snuffing on the network, I don't see any obvious problem.

Pls someone if you have encountered similar case, pls let me know.

[fajarpri]

After further investigation, I found out that the Cisco Switch disabled and block 2 ports from the Nortel Switch.

The documentation says that the switch may do that if there are some condition that it thinks a port is behaving that may risk the whole switch.

However, monitoring the Cisco log from command line doesn't reveal what is the reason for this particular case.

When I enable the ports, they would become disable again right away. I check the cable, they are fine. Check the port, it's fine too with other computer.

I'm not sure why the EXT5 and EXT6 port from the Nortel Switch are behaving like that. So, currently what I do is disable those two ports from the Nortel management web in hope that they won't be used for any traffic.

I'll keep monitoring if this solves the case. Otherwise I would replace the whole Nortel module.

[alefestaedist]

Hi there, I suppose that your case is when you got a Cisco Switch directly attached to another switch, since the switch-2-switch connection is not enable by deafult by Cisco, it try to protect the port shutting down. Another case could be if you use as VLAN the same VLAN id that the blade use as internal (1 or 3 it depends on the BLADE version).

Glad you fix the problem.

Alex