Wired Intelligent Edge

Hi there,

I have a four 2510 G switches which were set up 3-4 years ago. They seem to have a very basic config (probably almost factory default) and trunking is configured on port 23/24 on each unit.

; J9279A Configuration Editor; Created on release #Y.11.12

hostname "CW1-0"
trunk 23-24 Trk1 Trunk
snmp-server community "public" Unrestricted
snmp-server host 192.168.1.10 "public"
vlan 1
   name "DEFAULT_VLAN"
   untagged 1-22,Trk1
   ip address 192.168.1.200 255.255.255.0
   exit
stack commander "CW1"
stack member 1 mac-address c091346e42c0
spanning-tree
spanning-tree Trk1 priority 4

When I plug VMware ESX 5 and 5.1 onto the switches the management networks are unavailable. Any other switch or hub I plug into the VMHosts communicate fine so I am inclined to believe this could be an issue with this version of the 2510 firmware or maybe something isn't enabled by default in the config. Both ESX configs are pretty vanilla, nothing much has been added. The strange thing is that to get around this I have taken another NIC on the boxes and assigned another virtual switch to these and the virtual servers that sit on it communicate perfectly. It is only connection to the management network that is lost. I have moved the MGT network to the other virtual switch that the servers are hanging off and still the same problem.

I can see there are two firmware updates for the switches. As my next move I will update these but has anyone else had this problem? The default vlan is untagged and I have told VMware to accept any vlan and tried every configuration I can think of in the limited config environment. Still, there may be something I am missing. People have talked about LACP - but I am only using one NIC at the moment for testing. Any ideas would be much appreciated.

Cheers!

Updating firmware is always a good idea, but it isn't clear that a FW update here will make any difference.

You have only 1 VLAN at the moment. Is that what you expected, and is that what VMware wants? Maybe VMware has more than one VLAN configured?

The "trunk" port aggregating 23+24 goes to where? Do you have a loop or chain of the 4 switches?

The standards-based approach to aggregation is LACP, so a better command is:

trunk 23-24 trk1 LACP

 Each end needs to support LACP.

Thanks for your reply.

I think the one untagged VLAN was there by default and just left there by the engineer who set the switches up . VLANs are not used on this network and none are needed.

Yes, there are four switches trunked together via the last two ports on each switch. I'm not sur eif it a chain or a loop - will check.

VMware isn't aware of any VLANs but even when I add 4096 (which will accept any observed VLAN on the vSwitch) the management network refuses to communicate. As I said - the other vSwitch which has a couple of virtual servers on it works perfect on the switches - no VLAN configuration needed - vanilla. It is purely the management network which has issues when configured in a vSwitch and then attaching to the 2510s. No other switch has a problem.

I will try configuring a port on one of the switches as a trunk and enabling lacp (as per your config) and see what happens. Though, I am only using one NIC for management so aggregation isn't really needed.

Thanks for your help. Let me know if you have any other thoughts.

cheers!