DHCP relay on 2948G-GE-TX (CAT OS)

[kennylls]

I have a 2948G-GE-TX which is a CAT OS based catalyst. I saw product intro that it supports DHCP Relay agent, may I know how to configure that? Because my DHCP server is located next level cascaded switch and they are in same segment, but my workstation hard to obtain an IP from it.

I have been searching around the world but only found out how to enable that on IOS by ip helper-address command.

[vijayasankar]

Hi,

Only 2948G-L3 is a layer 3 router( running Cisco IOS)

Catalyst 2948-G is a layer 2 only switch, which runs the Cat-OS.

Moreover, if the client and DHCP server are in the same vlan, we dont need a dhcp relay agent.

Which devices is doing layer 3 routing in your setup...?

If you have configure the DHCP relay in that device properly to get DHCP ip for client which are located in a different vlan other than that of the DHCP Server.

-VJ

[kennylls]

Thanks for the prompt response.

The product catelogue posted in Cisco.com of this model is stated it has DHCP Relay agent, could it be wrong?

Anyway, my purpose is just want to have DHCP client to get IP from the DHCP server that situated at client' uplink switch(cascaded, no VLAN applied). I have also tried 'portfast' but it doesn't help either.

The symptom is that, when first connected the LAN cable(Cat5E of course), it attempts to obtain IP but eventually pops up 'DHCP server could not be contacted'(something similiar) and if lucky it will get IP after the error msg pops up and the balloon is clicked.

Should there any command or setting could help to achieve better response time if the switch is being expanded by uplink cable connected to one of the port?

[vijayasankar]

Hi,

Thanks for the update.

As stated earlier If there is no vlans involved , then there is no need for DHCP relay.

Lets rule out any layer 2 errors in the involved interfaces.

Check for any errors in all these interfaces.

1) Interface to which the Client PC is connected.

2) Interface on the next switch where the DHCP server is connected

3) Uplink / trunk interface on both the switches.

Also check if the Client PC is not having issues with speed/duplex setting.

If you statically configure a IP in the client, then check if yare you able to successfully ping the DHCP server IP with out any packet losses.

HTH

-VJ

[kennylls]

Hello VJ,

I have verified that there is no error on port interface of PC and uplink as well as DHCP server.

If PC is configured with static IP address, it could be up and running in less than 10 seconds, and there is no packet loss.

Is there any setting can be done, some sort similiar to Service Advertisement in Novell?

If the uplink is a non-Cisco switch (3Com), can the 2948G uplink be configured optimally?

[vijayasankar]

Hi,

Thanks for the update.

Im not aware of any additional settings required in this scenario, for this to work.

Can you try the following.

1) Once you get the DHCP ip properly in a client, then if you issue "ipconfig /release" and "ipconfig /renew", do you get the DHCP ip quickly.

The reason im asking this is to check whether the issue is observed only during the initial DHCP request process in a client.

If so, the issue might be to do with speed/duplex setting on the client NIC / Switch port.

If you observe inconsistent result in this test, then try capturing a ethereal trace in the client PC and in the server, to probe more details. Here in the trace, we can check whether the DHCP discover frame from the client PC is hitting the server or not and trace the whole sequence of operation.

-VJ

[Roberto Salazar]

DHCP Relay Agent is a function of a layer 3 device or commonly known as a router. Is it configured by "ip helper-address" under the interface of where the clients will be sending the DHCP request for IP address. since the IP address request or bootp are sent as UDP broadcast on port 67 ( or maybe it's port 68, anyway)since the router does forward broadcast rx from one subnet to another, it will need to configured with "ip helper-adress". "ip helper-address" takes that IP request sent by the client as UDP broadcast and sent it as unicast (if ip helper is accompanied by the unicast ip address of the dhcp server). Having explained that the 2948-GE-TX is a straight Layer 2 devide therefore it cannot be configured as a Relay Agent for DHCP.

Please rate helpful posts.

[kennylls]

I knew about this, that's why I wonder Cisco.com says this switch has DHCP Relay agent.

Is there anything can do to improve the DHCP response time to client in this switch? The uplink is a 3Com switch running at 100Mbps Full duplex.

[jackyoung]

Did you try to plug the client to the 3com switch and test the DHCP; also try to connect the DHCP server to 2948 to test it too ?

I suggest to isolate the problem between box first.

And what is the loading at 3Com and the uplink ? Is the DHCP server also 100/Full ?

[kennylls]

I have tried plug the client to the 3Com switch and it works perfectly, obtained IP in few seconds.

I have checked the possible errors between uplink, but none of packet had error. DHCP server also 100/full.

And, even after obtained IP at 2948 switch(luckily), I tried to Ping to DHCP server and I got < 1ms. I suspect the initial broadcast did not manage to reach DHCP server timely.

Any setting could resolve it? I even tried 'portfast' but doesn't help either.

[jackyoung]

I believe portfast may help a little to increase the port to be active on the LAN switch, but does not increase the DHCP renew time.

I agreed it may be the broadcast issue, but it is difficult to troubleshoot between two brands.

You may need to use Sniffer type of equipment/software to monitor the DHCP broadcast from client and determine the feedback from server.

Do you mean there is no error at the uplink port from Cisco to 3Com ? It is not the DHCP relay issue. It looks like somewhere limit the boradcast.

Can you confirm is there any broadcast suppression enabled in your 2948 ?

Please provide the config. of 2948G

Hope this helps.

[kennylls]

begin

!

# ***** NON-DEFAULT CONFIGURATION *****

!

!

#time: Mon Aug 14 2006, 17:38:10

!

#version 8.3(2)GLX

!

!

#system web interface version(s)

set password xxxxxxxxxxx

set enablepass xxxxxxxxxxxxxxx

set logout 5

!

#dot1x

set feature dot1x-radius-keepalive disable

!

#system

set system name Router1

set system location xxxxxxxxxxxxxxxxx

set system contact xxxxxxxxxxxxx

!

#frame distribution method

set port channel all distribution mac both

!

#vtp

set vtp domain Domain

set vlan 2 name Sales type ethernet mtu 1500 said 100002 state active

set vlan 3 name Marketing type ethernet mtu 1500 said 100003 state active

set vlan 1002 name fddi-default type fddi mtu 1500 said 101002 state active

set vlan 1004 name fddinet-default type fddinet mtu 1500 said 101004 state active stp ieee

set vlan 1005 name trnet-default type trbrf mtu 1500 said 101005 state active stp ibm

set vlan 1

set vlan 1003 name token-ring-default type trcrf mtu 1500 said 101003 state active mode srb aremaxhop 0 stemaxhop 0 backupcrf off

!

#ip

set interface sc0 1 10.205.40.78/255.255.252.0 10.205.43.255

set interface sl0 down

set interface me1 down

set ip route 0.0.0.0/0.0.0.0 10.205.40.60

set ip alias default 0.0.0.0

!

#dns

set ip dns server 10.205.2.9 primary

set ip dns server 10.205.40.50

set ip dns domain domain.com

!

#syslog

set logging console disable

set logging server enable

set logging server 10.205.40.73

!

#ntp

set timezone 8 0

!

#set boot command

set boot config-register 0x2

set boot system flash bootflash:cat4000-k8.8-3-2-GLX.bin

!

#multicast filter

set igmp filter disable

!

#module 1 : 0-port Switching Supervisor

!

#module 2 : 52-port 10/100/1000 Ethernet

set spantree portfast 2/18 disable

end

Above is the config, I will try network sniffer tomorrow, hopefully could find the root cause.

[jackyoung]

Thanks for that. I have no idea at the moment. If there is nothing can do, you may consider to download another ver. of CatOS and test again.