Etherchannel Seems Flaky (No DHCP)

[daniel.s.hall42.ctr1]

4507R+E will let clients communicate, but if someone needs to renew their lease on DHCP, it will not renew the IP.  All other clients in other areas/switches/subnets are using DHCP with no problem.

The configuration I've seen that appears "flaky" is the following:

Age of the Port-channel   = 818d:05h:55m:34s
Logical slot/port   = 21/1          Number of ports = 2
GC                  = 0x00000000
Port state          = Port-channel Ag-Inuse
Protocol            =    -
Port security       = Disabled

Ports in the Port-channel:

Index   Load   Port     EC state        No of bits
------+------+------+------------------+-----------
  0     00     Te3/1    On                 0
  0     00     Te4/1    On                 0

There are two ports in the aggregation, but the index is zero the load is zero, bits are zero.  Clients are still communicating but cant get an IP.  If you statically assign an IP to these computers they can't communicate either.

Ideas?

[Mark Malone]

Hi

If you run wireshark on a client pc on the 4507 do you see the dhcp requests offers going back and forth or are they one way saying going from the pc but not coming into the pc ?

is the switch layer 2 or 3 just wondering is dhcp server on same vlan or is it going through a vlan svi with a helper address

its odd the load doesn't have a value or a bit value and the index is both 0 all mine are 1 and 0 and with it being up so long

You could change the PO to use LACP or PAGP as a test see if it changes anything rather than On mode

Index   Load      Port          EC state       No of bits
------+------+------------+------------------+-----------
 0      E8        Te2/5/4                 On   4
 1      17        Te2/5/5                 On   4

[daniel.s.hall42.ctr1]

I haven't had a chance to do a packet capture yet.  But I will.

It is a layer 2 trunk to the distribution layer.

The DHCP server is on another VLAN and has the ip-helper configured on the VLAN interface.

[Mark Malone]

so yes in short that's what I would try next anyway the packet capture on client and server side see whats going on with the requests where there being dropped between them or run a debug dhcp detail but teh ws is easier to read sometimes

is this your only 4507 just wondering if its a software issue with it but you could rule that out if its running somewhere else too and the dhcp is working

Could also check the caveat section for your current software on the website see if there are any known dhcp issues with your particular version , whats the ios image running on it now ?

if still cant see what's causing it try change the PO type and then upgrade if possible if nothing obvious as may be a odd bug

[daniel.s.hall42.ctr1]

I have another switch with the same symptoms and it also shows the same abnormal etherchannel detail.

                Ports in the group:
                -------------------
Port: Te3/1
------------

Port state    = Up Mstr In-Bndl
Channel group = 1           Mode = On              Gcchange = -
Port-channel  = Po1         GC   =   -             Pseudo port-channel = Po1
Port index    = 0           Load = 0x00            Protocol =    -

Age of the port in the current state: 74d:23h:39m:47s

Port: Te4/1
------------

Port state    = Up Mstr In-Bndl
Channel group = 1           Mode = On              Gcchange = -
Port-channel  = Po1         GC   =   -             Pseudo port-channel = Po1
Port index    = 0           Load = 0x00            Protocol =    -

Age of the port in the current state: 74d:23h:37m:48s

                Port-channels in the group:
                ---------------------------

Port-channel: Po1
------------

Age of the Port-channel   = 819d:04h:33m:16s
Logical slot/port   = 21/1          Number of ports = 2
GC                  = 0x00000000
Port state          = Port-channel Ag-Inuse
Protocol            =    -
Port security       = Disabled

Ports in the Port-channel:

Index   Load   Port     EC state        No of bits
------+------+------+------------------+-----------
  0     00     Te3/1    On                 0
  0     00     Te4/1    On                 0
         
Time since last port bundled:    74d:23h:37m:48s    Te4/1
Time since last port Un-bundled: 74d:23h:44m:20s    Te4/1

I upgraded the first 4507 from 03.04.03 SG, to 03.06.06 E.

I changed the etherchannel from on to active (lacp).

It is still experiencing the same symptoms.

Although, I have multiple other switches 4507's with the same software version and etherchannel configuration that are working just fine.

We are working to do a packet capture now.

[daniel.s.hall42.ctr1]

We just completed the packet capture on the bad switch and a known good switch.

On the bad switch we see a DHCP Discover go out and DHCP Ack come back.  No Offer or Request. 

We see all packets on the known good switch.

[walfors]

Daniel,

Have you tried bouncing the ports Te3/1 and Te4/1?

After doing that, is the "Port Index" updated? (to 0 & 1)

Best ones - Lennart

[Mark Malone]

when did this start happening as the etherchannel was up 800 days is it going on that long or only started recently

very odd even after software upgrade your still seeing same problem but all other traffic is working fine across the etherhchannel

is there any difference in the clients attached to this switch like that are looking for the dhcp ips , software nics etc, do they work fine if connected to the working switches

when you set them as  static ip they work fine yes ?