Hi Pete, Thanks for coming back to me. Yes we do have a few switches, but we have not moved the devices around (changed ports). The entry in the arp table looks something like this: jbpn1 (172.16.128.29) at <incomplete> on eth0 I mentioned this to someone else, and they also mentioned the possibility that the switch was the cause of the problem. I'll try and flush the switches/routers and other "intelligent" devices this weekend to see if that solves the problem. Regards Enrico ----- Original Message ----- From: "Pete Nesbitt" <pete@xxxxxxxxx> To: <redhat-list@xxxxxxxxxx> Sent: Friday, November 14, 2003 3:33 AM Subject: Re: ARP Table wierdness > On November 13, 2003 03:36 am, Enrico Payne wrote: > > Hi, > > > > I am running RH8 (Std ISO install) on 3 servers (kernel 2.4.18-14), and > > have noticed a strange thing over the last few days. > > > > Sometimes the client pc's (Windoze) or the servers for that matter may not > > be able to ping the one server. However if I do a traceroute, it knows > > about it, as I get !H messages on the traceroute. > > > > If I check the arp table (arp -a), I can see the server I am trying to > > ping, but it has an incomplete address. > > If I manually add the server to the arp table (arp -s), then all is well. > > What is strange is the fact that all the machines are on the same subnet > > and other machines can ping the server no problem. Any amount of rebooting > > does not seem to resolve the problem. This seems to be happening at random. > > > > I have also noticed an error on the server consoles "Neighbour table > > overflow". Does this perhaps have something to do with it? > > > > Does anyone know what could be causing the ARP weirdness, and what is > > causing the Neighbour table overflow error? > > > > Regards > > Enrico Payne > > not sure on why this is intermittent, but !H is Host Unreachable. So is it > just geting resolution from a local host file and failing to access the > network? > > -can you clearify the situation a bit (you say "it knows about it" hmmm) > -what exactly is in the incomplete entry of arp table? > > One thought, that may be unrelated, is that despite the way a switch should > work (send out all other ports if unknown), I have seen situation where you > had to "ping out" from a new machine to get it in the switches mac table, > this is often the case when you move a box from one port to another, as the > switches mac tables are wrong until it flushes the old info. Do you have > anything like that that may be causing problems? > > -- > Pete Nesbitt, rhce > > > -- > redhat-list mailing list > unsubscribe mailto:redhat-list-request@xxxxxxxxxx?subject=unsubscribe > https://www.redhat.com/mailman/listinfo/redhat-list -- ___________________________________________ This e-mail has been scanned for viruses. Pharma Natura will not be held responsible for the loss of data or any other loss caused by the use of the information contained in this e-mail. -- redhat-list mailing list unsubscribe mailto:redhat-list-request@xxxxxxxxxx?subject=unsubscribe https://www.redhat.com/mailman/listinfo/redhat-list
