In article <BAY22-DAV17goFT3kMU000388d7@hotmail.com>,
Justin Robbs <justin_robbs@hotmail.com> wrote:
>
>----- Original Message -----
>From: "Tony Lawrence" <pcunix@gmail.com>
>Newsgroups: comp.unix.sco.misc
>To: <distro@jpr.com>
>Sent: Friday, September 10, 2004 11:20 AM
>Subject: Re: telnet problem
>
>
>>
>> Justin Robbs wrote:
>> > ----- Original Message -----
>> > From: "Justin Robbs" <justin_robbs@hotmail.com>
>> > Newsgroups: comp.unix.sco.misc
>> > To: <distro@jpr.com>
>> > Sent: Wednesday, September 08, 2004 8:31 AM
>> > Subject: Re: telnet problem
>> >
>> >
>> > >
>> > > ----- Original Message -----
>> > > From: "Justin Robbs" <justin_robbs@hotmail.com>
>> > > Newsgroups: comp.unix.sco.misc
>> > > To: <distro@jpr.com>
>> > > Sent: Tuesday, September 07, 2004 3:38 PM
>> > > Subject: telnet problem
>> > >
>> > >
>> > > > I am having a weird problem with two different sites. I cannot
>> telnet
>> > > > directly to either one over our satellite network. I can ping
>> the
>> > > devices,
>> > > > so I know the routes are ok, just no telnet. I can also telnet
>> to other
>> > > > OpenServer machines on those same networks, then from there
>> telnet into
>> > > the
>> > > > affected machine over the LAN just fine. Also, once I get on to
>> one of
>> > > the
>> > > > affected machines via this circuitous route, I can telnet
>> directly to
>> > the
>> > > > original machine I am coming from. Any ideas what could be
>> causing
>> > this?
>> > > >
>> > > > The machines are running Open Server 5.0.6, they have the correct
>> > default
>> > > > gateway setup as well. This began happening over the weekend.
>> > > >
>> > > > Thanks,
>> > > > Justin
>> > > >
>> > >
>> > > Here is some additional information. All of our sites are
>> connected via
>> > > satellite WAN. From the main office we have a T-1 connection to
>> the main
>> > > hub for the satellite network. Now, from one of our other sites on
>> the
>> > WAN,
>> > > I can telnet directly to these specific machines that are causing
>> > problems.
>> > > However, I cannot get to these two machines from the main office
>> using
>> > > telnet or ftp. Ping still works just fine. I can reach any of the
>> other
>> > > machines at any of our other sites just fine. Any help would be
>> greatly
>> > > appreciated. BTW, I am not doing any DNS.
>> > >
>> > > Thanks,
>> > > Justin
>> > >
>> >
>> > FTP is behaving the same way as telnet. I forgot to mention that
>> earlier.
>>
>> WAGS:
>>
>> Firewall reconfigured?
>> MTU problem somewhere?
>>
>> Cheap lan card?
>>

>Since I last posted, I talked to the satellite company. They have
>fixed one site by refreshing the ARP tables, at least that is
>what they said.

>However, shouldn't pinging the site do that. The last time they
>told me our ARP tables were messed up, they only way we could
>telnet/ftp was to ping the sight first. Then from what I can
>gather, the tables were refreshed at least temporarily.

It all depends on how things are set up. On smaller routers the
arp tables will get refreshed more often, but on bigger routers you
really don't want to do that.

At another ISP a client was upgrading their hardware and put in a
second rack with four 8RU Compaq servers. He called and said
they weren't working so he put the others back in place and wanted
to know what was wrong. I told him he should have called me first
and had him swap the the cables to bring the new rack into play.

I went into the 7513, and cleared the arp-cache. And before I
could say "it should be working" he said "it's working now".

The default arp-cache time on those is 4 hours. But since those
can handle huge routing tables and have thousands of connected
machines, refreshing less than that was not needed.

As to fixing things by pinging it first, if the IP/MAC mapping
was not in the arp cache, then the ping would put that into the
arp-cache.

>The other site I have been unable to test as one of our
>technicians replaced the computer to try to fix the problem and
>now the new one has crashed. Unfortunately, he can't get out
>there until this weekend.

Sounds like Murphy struck again.

>At any rate, I don't know a lot about ARP. Does that sound like a
>plausible solution? I am not comfortable with that solution, but
>I haven't had any time to do any research about ARP.

ARP - Address Resolution Protocol.

Machines that are directly connected only use the IP address to
get the MAC address - and then all communications is done
MAC to MAC. [The MAC address is the 6 octets speparated by
colons when you type arp -a or arp -an

? (192.168.0.1) at 00:09:5b:01:ac:de on fxp0 [ethernet]
? (192.168.0.5) at 00:60:08:02:d0:14 on fxp0 [ethernet]
? (192.168.0.133) at 00:0c:6e:58:b0:33 on fxp0 [ethernet]
? (192.168.0.255) at ff:ff:ff:ff:ff:ff on fxp0 permanent [ethernet]

If something had changed and the arp-cache had a different MAC
address associated with the IP you were pinging, the pings would
go nowhere, as it would be using a MAC address that was probably
non-existant. At that point you can 1) wait until the cache times
out or 2) have whoever controls the router clear the arp cache.

>At any rate, I don't know a lot about ARP. Does that sound like a
>plausible solution? I am not comfortable with that solution, but
>I haven't had any time to do any research about ARP.

All you probably need to know about ARP is what I wrote above.
man arp should give you more info on how to see it locally.
On local machines arp -d <ip> will delete one, or
arp -d -a will delete all, which is similar to clear arp-cache
in a router.

Bill
--
Bill Vermillion - bv @ wjv . com