putting a packeteer packetshaper into a etherchannel bundle

sethroth1 · ‎10-02-2007

Hi,

I am using two packetshapers in a etherchannel bundle to handle the load traffic (massive traffic)- when I do this only one shaper is reachable via ping, when I do a packet trace I see that one of the shapers cannot recieve arp replies - This must be because of the etherchannel itself. I know that you are not supposed to insert devices into etherchannels, and I do not wish to recieve replies stating "etherchannels are not for that etc etc type of forum replies" But, I want to know why only one of the devices are reachable via ping and other services. I see that on the traces that STP and LOOP protocol are in the packetcaptures on that device, can some one that knows STP and LOOP explain why only one device in the bundle can be reached. Thanks so much, NOTE: This is a lab environment, not a production environment, I can break or do anything config and device wise.

Thanks!

paul.matthews · ‎10-02-2007

First thought is what intellingence do packeteers have for the dot1Q traffic? does their connectivity need to be untagged, and as such is the IP address configured in the native VLAN?

sethroth1 · ‎10-02-2007

The shaper understands dot1q and can classify vlans, but when in the channel test, I see that the first device pinged is the only device reachable, I see alot of stp in the traces and cisco LOOP, I am wondering if the STP packets are replying to the arp being sent to the pshaper (from the host after replying to the pshapers arp broadcast) that the pshaper is never recieving. But to answer your question, the ip address is of vlan 20 and the traffic is uuntagged.

The scenario is (for testing) two switches, two L2 vlan's (1 and 2), no router (lab), one L2 etherchannel between both switches, pshaper 1 is 10.0.20.55 ps2 is 10.0.20.77 , icmp host (pinging from host) is 10.0.20.100.

Thank you so much,

all help is appreciated in advance.

Seth

paul.matthews · ‎10-02-2007

So, it looks like the packeteers are getting "involved" in spanning tree!

STP and ARP are totally separate. I presume you have onw switch forced as root? Do you know if the packeteers get involved in spanning tree? If they do. you may be getting an issue where the packeteers are doung traditional single spanning tree, and the switches are doing PVST and the two are not playing nicely.

I presume by saying the tfafic is untagged, VLAN20 is the native VLAN?

sethroth1 · ‎10-03-2007

Hi,

thank you so much for your help - I believe that the PShaper does not get involved in STP, it is not capable of digesting stp traffic, actually the device does not support trunk links at all, yup the native vlan is 20 on the switch, the ps does not see tagged vlan traffic either. So the big question is "why" - why can only one device be reachable - is it way data travels through the etherchannel? is it because the shaper does not support trunking? I do see many stp packets in the traces and the cisco Loop protocols involved have a 'reply' flag in the packet,

Thanks so much,

paul.matthews · ‎10-03-2007

If it does not understand the trunks, it can't manange the traffic on them can it?

Is the same one responsive to both switches, or can both see the "near" packeteer? Can the switches see each other with CDP? Can they ping each other?

Danilo Dy · ‎10-03-2007

Hi,

How does the two PS connected/configured?

1. HA?

2. You have more ports for etherchannel bundle than one PS can handle so you put two?

If you can't access one of the two PS using the IP in the etherchannel or STP, you can configured the PS Management port to access it.

This assuming that you or anyone else did not configured the second PS ACL for host allowed to access it.

Regards,

Dandy