Does ANYONE have a clue how to fix this? Novell says it's MS's crappy protocols. MS admits they are not to standard but nobody has a fix. We've been fighting this battle for months.
Conflict of IP Address on Novell 5 network w/ Win98 clients
- Thread starter 0vercast
- Start date
Yeah what's up with that? I never had any problems with address conflicts and I installed nw5 networks with over 1000 win98pc's.
I have seen weird problems reporting address conflicts, but it turned out to be layer2 problems.
Something like 8 hubs with multiple ethernet loops (hub1<-->hub2<-->hub1). The machines would see their own arp and responde to it!!!!
Wondering if there isn't some other problem here
Something like 8 hubs with multiple ethernet loops (hub1<-->hub2<-->hub1). The machines would see their own arp and responde to it!!!!
Wondering if there isn't some other problem here
Given the wealth of supplied information, I'd throw a dart at the use of RIP between the two. Check the Novell server and make sure that RIPII is diabled, or at least only listening. MS Winderz don't do RIP II. So, the Novell box should only be blabbing RIP I, and listening for RIP I and II (or only RIP I). There might be a "compatibility mode," for RIP II I don't remember
This may screw you up if you're using Variable Length Subnet Masking (VLSM), and (if that's the case) you may have to reconfigure one/some/all of your routers to make the acommodation.
What may be happening is whatever address you have subnetted is being passed through a RIP I - only segment (which will reset the mask to the default for the address class you're using), and comes out to the next segment as two chunks of the same network (the two subnets are now two seperate chunks of the same network) and/or The Winderz box is getting RIP II information only, and doesn't understand the subnetting (like maybe you have two or more NICs in this server?)
I'm not sure about how Windows handles RIP II, how it sees it, or what it does with it...but like I said, with the given information, that's my first best guess.
My second best guess (supplied by one of the guys I work with) is that perhaps there are address assignments being issued by DHCP that are overlapping with hard-coded addresses (like maybe you forgot to exclude them from the DHCP pool). For example, you have a router at 172.16.1.254/24...and you have a DHCP address pool scoped for 172.16.1.(1-254)...when DHCP issues the ".254" address... the conflict pops up. The fact that there's Novell and MS on the same net is just a Red Herring.
SO, there ya go.
FWIW /.02
Scott
This may screw you up if you're using Variable Length Subnet Masking (VLSM), and (if that's the case) you may have to reconfigure one/some/all of your routers to make the acommodation.
What may be happening is whatever address you have subnetted is being passed through a RIP I - only segment (which will reset the mask to the default for the address class you're using), and comes out to the next segment as two chunks of the same network (the two subnets are now two seperate chunks of the same network) and/or The Winderz box is getting RIP II information only, and doesn't understand the subnetting (like maybe you have two or more NICs in this server?)
I'm not sure about how Windows handles RIP II, how it sees it, or what it does with it...but like I said, with the given information, that's my first best guess.
My second best guess (supplied by one of the guys I work with) is that perhaps there are address assignments being issued by DHCP that are overlapping with hard-coded addresses (like maybe you forgot to exclude them from the DHCP pool). For example, you have a router at 172.16.1.254/24...and you have a DHCP address pool scoped for 172.16.1.(1-254)...when DHCP issues the ".254" address... the conflict pops up. The fact that there's Novell and MS on the same net is just a Red Herring.
SO, there ya go.
FWIW /.02
Scott
<< what kind of error message are you getting? symptoms of the problem? >>
"Windows has detected a conflict of IP for address XXX.XXX.XXX.XXX" and it tells you the mac address... don't remember the exact wording.
on the server you will see this:
"Received an unexpected decline packet identifying address XXX.XXX.XXX.XXX or the address failed a ping test."
our current fix is simply to release and renew in WINIPCFG, but that only takes care of that particular conflict and only at that particular moment in time.
<< My second best guess (supplied by one of the guys I work with) is that perhaps there are address assignments being issued by DHCP that are overlapping with hard-coded addresses (like maybe you forgot to exclude them from the DHCP pool). For example, you have a router at 172.16.1.254/24...and you have a DHCP address pool scoped for 172.16.1.(1-254)...when DHCP issues the ".254" address... the conflict pops up. The fact that there's Novell and MS on the same net is just a Red Herring.
>>
that was our best guess at one time also but we've found no problems with our DHCP setup.
<< Oh, I've got a fix: get rid of IP and use IPX. Goodbye addressing conflicts!
>>that was my favorite solution also, but unfortunately i'm not in charge so i don't get to make that decision.
Are you running dhcp on the Novell server? And what about a network diagram and protocol breakdown? Interesting problem. Have you sniffed the segment where this is occuring?
<< Are you running dhcp on the Novell server? >>
Yes.
<< And what about a network diagram and protocol breakdown? Interesting problem. Have you sniffed the segment where this is occuring? >>
i have no network diagram to offer. protocol breakdown meaning what? and i honestly don't know what a segment is or how i'd go about sniffing it. none of us here have really been trained to handle this stuff, which is undoubtedly a large part of the problem. most of this work is done by consultants. our novell consultant says it's the layer 3 switches. our cisco consultant says it's the DHCP server. and me, i dunno WTF is going on, i'm just tired of all the helpdesk calls for IP conflicts.
Have you found a pattern, like IP's out of a particular scope getting conflicts?
Draw out a diagram of how your network is layed out, label the routers/switches/hubs/servers (all interfaces on the servers) as such. You are obviously running IP, how about IPX? You'd be surprised what you can find out by tracing your cables and getting a handle on what goes into what. Your consultant might be right about the L3 switches, but if so why doesn't he fix it or tell you exactly what the problem is? Suspicious. There are some really sharp people on this forum who would be happy to help you if you could just provide some more detailed information. Unfortunately, this is not going to be a quick-fix situation.
Draw out a diagram of how your network is layed out, label the routers/switches/hubs/servers (all interfaces on the servers) as such. You are obviously running IP, how about IPX? You'd be surprised what you can find out by tracing your cables and getting a handle on what goes into what. Your consultant might be right about the L3 switches, but if so why doesn't he fix it or tell you exactly what the problem is? Suspicious. There are some really sharp people on this forum who would be happy to help you if you could just provide some more detailed information. Unfortunately, this is not going to be a quick-fix situation.
Hi,
It sounds like your client lease is expiring at the server before the client renews it. If you are sure this is not happening then write a batch file or hack the registry so that the client releases the address on shutdown. This will give you some time to put a sniffer on the wire to see what's going on. Ethereal is a great free sniffer.
It sounds like your client lease is expiring at the server before the client renews it. If you are sure this is not happening then write a batch file or hack the registry so that the client releases the address on shutdown. This will give you some time to put a sniffer on the wire to see what's going on. Ethereal is a great free sniffer.
<< Have you found a pattern, like IP's out of a particular scope getting conflicts?
Draw out a diagram of how your network is layed out, label the routers/switches/hubs/servers (all interfaces on the servers) as such. You are obviously running IP, how about IPX? You'd be surprised what you can find out by tracing your cables and getting a handle on what goes into what. Your consultant might be right about the L3 switches, but if so why doesn't he fix it or tell you exactly what the problem is? Suspicious. There are some really sharp people on this forum who would be happy to help you if you could just provide some more detailed information. Unfortunately, this is not going to be a quick-fix situation. >>
Most of them are in one building, but we do get them in other places occasionally. IP ranges are distributed by building or by dept in the large buildings. 90% or more of our conflicts are in that one building. Yes we are running IPX also. The novell consultant isn't responsible for configuring the L3 or any of the cisco equip. He just takes care of Netware basically. The cisco consultant says the L3 is fine. I'd be happy to supply more information, i just don't know a lot about this stuff so i don't know what to give you until you ask for it.
<< Hi,
It sounds like your client lease is expiring at the server before the client renews it. If you are sure this is not happening then write a batch file or hack the registry so that the client releases the address on shutdown. This will give you some time to put a sniffer on the wire to see what's going on. Ethereal is a great free sniffer. >>
i tried that registry-hack fix a while back. from that point on that computer had conflicts every minute of the day.
well think about what the MS computer is doing. It sends out an ARP for its own IP address and if somebody answers you'll get the "address conflict" error message. If you have switches you can track down the MAC address of the machine that has that IP address. If you see that the MAC address of the conflict is indeed your computer or the interface of another router then something is wrong with the network configuration. My main concern is that you have more serious underlying network problems.
I'd be telling your "consultants" to fix it. Add up the cost of a support call with these kinds of messages and start billing them, they'll fix it.
<edit> the MS computer is doing exactly what it is supposed to do by arping for itself. You should never get this kind of error message unless you truly do have a conflict with another machine.
BTW - what is the lease time set to on the server?
I'd be telling your "consultants" to fix it. Add up the cost of a support call with these kinds of messages and start billing them, they'll fix it.
<edit> the MS computer is doing exactly what it is supposed to do by arping for itself. You should never get this kind of error message unless you truly do have a conflict with another machine.
BTW - what is the lease time set to on the server?
weird...it should only release on shutdown...
Ok,
Have you tried to increase the lease expiration? How many hops are the clients from the server? Does this happen on the local segment? What version of DHCP (modules dhcpsrvr), mine is 3.01d. Are you on Win98 second release? Are you running the latest Client32 or are you running the Microsoft Netware client?
Ok,
Have you tried to increase the lease expiration? How many hops are the clients from the server? Does this happen on the local segment? What version of DHCP (modules dhcpsrvr), mine is 3.01d. Are you on Win98 second release? Are you running the latest Client32 or are you running the Microsoft Netware client?
<< Ok,
Have you tried to increase the lease expiration? How many hops are the clients from the server? Does this happen on the local segment? What version of DHCP (modules dhcpsrvr), mine is 3.01d. Are you on Win98 second release? Are you running the latest Client32 or are you running the Microsoft Netware client? >>
We've discussed, but not tried, changing the lease expiration. To my knowledge, even our most remote users are no more than 2 hops from the server and the building that gets hit most has a DHCP server in it. It is 98SE and Novell Client 3.3 or 3.2 on the client machines.
Here is a description of the error from Novell
My guess would just be that whoever set up that DHCP scope did not exclude some static addresses that you are using or you just have some rougue users that think they can just pick and address and use it.
Have you bee tracking what the addresses are that are getting conflicts - are they always the same addresses? Also - you are seeing the MAC addresses of the devices that are in conflict - any information that you can get from there? Take a look at the OUI (vendor code) portion of the MAC address - you can search for it here. Are they all HP MAC addresses - maybe they didn't exclude the static IP's assigned to the printers, if all your machines use 3COM NICS and the vendor code belongs to 3COM then possibly it's a user that hard coded his IP address.
My guess would just be that whoever set up that DHCP scope did not exclude some static addresses that you are using or you just have some rougue users that think they can just pick and address and use it.
Have you bee tracking what the addresses are that are getting conflicts - are they always the same addresses? Also - you are seeing the MAC addresses of the devices that are in conflict - any information that you can get from there? Take a look at the OUI (vendor code) portion of the MAC address - you can search for it here. Are they all HP MAC addresses - maybe they didn't exclude the static IP's assigned to the printers, if all your machines use 3COM NICS and the vendor code belongs to 3COM then possibly it's a user that hard coded his IP address.
Oh one more thing -- how is your timesync on the server? Clients keeping good time? DSREPAIR's look normal? You'd be surprised how a little lost sync can really hose up a good Novell network. Can you post the console log from a server getting this DHCP error? You are logging console, right?
<< Here is a description of the error from Novell
My guess would just be that whoever set up that DHCP scope did not exclude some static addresses that you are using or you just have some rougue users that think they can just pick and address and use it.
Have you bee tracking what the addresses are that are getting conflicts - are they always the same addresses? Also - you are seeing the MAC addresses of the devices that are in conflict - any information that you can get from there? Take a look at the OUI (vendor code) portion of the MAC address - you can search for it here. Are they all HP MAC addresses - maybe they didn't exclude the static IP's assigned to the printers, if all your machines use 3COM NICS and the vendor code belongs to 3COM then possibly it's a user that hard coded his IP address. >>
we tracked offending addresses for several months and never had a recurring number. the printers do not have static IPs. we've been looking for users setup with a static IP, but haven't found any. most users have been checked, but not all of them yet.
<< Oh one more thing -- how is your timesync on the server? Clients keeping good time? DSREPAIR's look normal? You'd be surprised how a little lost sync can really hose up a good Novell network. Can you post the console log from a server getting this DHCP error? You are logging console, right? >>
Timesync is always flawless. everything is fiber between servers and switches. DSRepair occasionally turns up an error or two. It's done weekly. Help me locate the file and I'll post it.
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 25K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes Discussion Threads
- Started by Tigerick
- Replies: 21K
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
-
-
Facebook
Twitter