ipx problems with dial in access

[mark.weaver]

i have two c3640 access servers with internal mica modems, on one i am getting intermittent failure of lcp to negotiate ipx with the dial in client, sometimes it is ok sometimes not, without making any changes in config to pc or router.

The other router is configured the same with all 30 lines configured as group-async1 and then made an ipx ppp-client of a loopback interface that has an ipx network assigned to it.

The only difference is that the router that is having problems is more heavily utilised than the one that does not have a problem.

any ideas would help.

regards

mark weaver

[makchitale]

1)You see intermittent LCP negotation failure with IPX clients, do the IP client connect fine at the same time?

2)During the LCP nego do you see the router sending CONREQ out but does not get a response?

Clear the interface counters & see if we see any inputs drops on "int async x" corresponding to the call ...if we see no input packets / drops/ errors then need to see if modem retrains / speedshifts.

Modem outputs such as 'sh mod op x/y" or modem call record terse will have info on number of tx/rx characters & ppp characters sent /received.

debug modem / deb modem csm / deb ppp nego will be useful for a failed & good call.

Thanks, Mak.

[mark.weaver]

Ip clients connect fine. I have a sniffer installed on my pc which i dial in for test purposes with and i have do a packet capture which shows that when ipxcp fails to negotiate it sends one ipxcp confreq packet and then nothing no acknowledgement.

when ipxcp is successful it sends the same confreq and then receives a confack.

Doing de-bugs is difficult at the moment becuse the router is always busy receiving calls, but i am working on it.

As a point of interest this problem did not appear during testing before the router went "live".

[makchitale]

During the call that fails ipxcp: When the PC send a ipxcp packet, on the Access server end do we see any ipxcp IN/ OUT packets? Ideally the PC too should send more than one ipxcp request & then timeout if there is no response.

Does this happen with multiple users or just with your PC?

Can you please capture the debug ppp nego / deb ppp packet for the failed call (when you have less traffic on the access server)

Clear the interface counters & see if we see any drops on "int async x" corresponding to the call ...do we see input packets / drops/ errors ?

Thanks, Mak.

[mark.weaver]

Problem solved - i am an idiot.

On this router i have two e1 controllers and two banks of modems, because at the moment i only have one e1 line i assumed that it would only use the first modem bank (group-range 33 62) which i configured as group-async1.

I configured group-async2 (group-range 65 94) with all the necessary config ready for when we had a second e1 line installed apart from assigning it an ipx network - it was only while investigating this morning that i noticed my call was coming in on lines in the 65-94 range and then the fog lifted. I did not have this problem on the first router because that only had one bank of modems and when i tested the second one it was not tested with the heavy volume of calls that it is now getting.

I have now assigned group-async2 an ipx network and things are now working as they should be.

sorry to waste your time but thanks for your help anyway.

regards

mark