Mysterious IPX Problems
Articles and Tips: qna
01 Nov 2002
Q.
Over the past week I've been fighting a major issue with our IPX protocol and I'm not sure what the cause of the problem is. I was wondering if you have any ideas, as I've pulled out so much hair that I have very little hair left.
We have 45 different servers across the United States and Canada. We run dual stacks, primarily still IPX as we have not yet started going down the path of Pure IP yet. Just over three to four weeks ago, we converted from AT&T to Sprint for our WAN links. We had very few problems with the conversion and everything ran fine.
We also have a legacy manufacturing system which uses Btrieve 6.15 as its engine and we have one last region using this manufacturing system. We are going to phase out this manufacturing system and replace it with an Oracle-based manufacturing system (homegrown). However we have to get a new power generator and UPS for the new Solaris system to house the last region using the Btrieve manufacturing. In the meantime, I have to hold the old system together with spit and bailing wire until the conversion is completed.
My problem started one week ago. We started receiving reports that people in this region were unable to access the Btrieve manufacturing system. So we started checking and during these checks, I typed display servers and it only showed about 45 IPX Addresses.
Normally we see between 1200 to 1300 IPX addresses on the network (and by-the-way, we do not filter any SAP broadcasts). Once one server only sees about 45 IPX addresses, the other servers in my network start to lose IPX addresses out of their memory until each server only sees about 45 different IPX addresses. To resolve the problem, I either randomly (and luckily) hit a server until I issue a random server that fixes the problem or the problem solves itself.
My problem is two-fold. First, this particular problem has come back every day for the past week. Second, I have yet to find the root cause of this problem. Once I know a root cause, I can resolve it. Any advise on what I should be looking for would be greatly appreciated.
IPX'ed in Ipshire
A.
Dear IPX'ed: You're in luck, as I was revisiting IPX just this past weekend. From what you have told me, most likely the problem is packet flooding--too many packets too quickly. Unless steps are taken to throttle back the rate at which IPX RIP and SAP broadcasts are emitted, then receivers (servers and routers, etc.) may easily lose many packets.This material is sent as datagrams, and losses are not kept track of.
The solutions are two fold: First, set emission pacing to be typically 50ms between packets.This is a control on current NetWare servers and many Cisco routers have it as well. Second, filter your IPX RIP/SAP traffic to reduce the burden on your network.
* Originally published in Novell AppNotes
Disclaimer
The origin of this information may be internal or external to Novell. While Novell makes all reasonable efforts to verify this information, Novell does not make explicit or implied claims to its validity.