I have trying to figure out a problem with a computer loosing
connectivity on a network. I have run ping tests against the computer
and I am seeing a 16% packet loss. I changed the switch the computer
was connecting with. The problem continued. I tried again and was
seeing roughly the same error rate. I then took one of the switches
out and put in a tap. Still seeing > 15% packet loss over an 8 hour
window. I looked for something in the dump logs that might give me a
clue, but what I saw puzzled me even more. The computer was
communicating with the file server , but was not receiving ping
requests anymore from the computer which I am using to monitor.
The original problem was that the users noticed they couldn't
connect to the file server. But now I'm really baffled as to what
could be causing the problem.

Any ideas?
Any other tests?
What might I look for in the tcpdump log?

On 2008-07-17, baggins <clarkjohnston@gmail.com> wrote:
>
>
> I have trying to figure out a problem with a computer loosing
> connectivity on a network. I have run ping tests against the computer
> and I am seeing a 16% packet loss. I changed the switch the computer
> was connecting with. The problem continued. I tried again and was
> seeing roughly the same error rate. I then took one of the switches
> out and put in a tap. Still seeing > 15% packet loss over an 8 hour
> window. I looked for something in the dump logs that might give me a
> clue, but what I saw puzzled me even more. The computer was
> communicating with the file server , but was not receiving ping
> requests anymore from the computer which I am using to monitor.
> The original problem was that the users noticed they couldn't
> connect to the file server. But now I'm really baffled as to what
> could be causing the problem.
>
> Any ideas?
> Any other tests?
> What might I look for in the tcpdump log?

You might try replacing the network card (if the NIC is on the
motherboard, use a card instead).

On Jul 18, 4:04 am, Bill Marcum <marcumb...@bellsouth.net> wrote:
> On 2008-07-17, baggins <clarkjohns...@gmail.com> wrote:
>
>
>
>
>
> > I have trying to figure out a problem with a computer loosing
> > connectivity on a network. I have run ping tests against the computer
> > and I am seeing a 16% packet loss. I changed the switch the computer
> > was connecting with. The problem continued. I tried again and was
> > seeing roughly the same error rate. I then took one of the switches
> > out and put in a tap. Still seeing > 15% packet loss over an 8 hour
> > window. I looked for something in the dump logs that might give me a
> > clue, but what I saw puzzled me even more. The computer was
> > communicating with the file server , but was not receiving ping
> > requests anymore from the computer which I am using to monitor.
> > The original problem was that the users noticed they couldn't
> > connect to the file server. But now I'm really baffled as to what
> > could be causing the problem.
>
> > Any ideas?
> > Any other tests?
> > What might I look for in the tcpdump log?
>
> You might try replacing the network card (if the NIC is on the
> motherboard, use a card instead).

yes, we replaced it twice.
The last time it happened. I was able to get on the fileserver and
ping from there and saw no response. I then pinged from my computer
and saw a response.
So now I'm thinking that the switch cache is getting messed up. But I
don't know how it is getting messed up. This is the second switch
that I have used. So my assumption is that there is something on the
network which is causing the problem (probably this computer).

if only 1 computer is impacted, these are the steps that never fail

1. replace cable
2. replace cable again
3. force full duplex
4. change switch port to a known good port
5. change NIC
6. replace NIC driver

when you change ports and NIC, clear arp cache or wait


> yes, we replaced it twice.
> The last time it happened. *I was able to get on the fileserver and
> ping from there and saw no response. *I then pinged from my computer
> and saw a response.
> So now I'm thinking that the switch cache is getting messed up. *But I
> don't know how it is getting messed up. *This is the second switch
> that I have used. *So my assumption is that there is something on the
> network which is causing the problem (probably this computer).- Hide quoted text -
>
> - Show quoted text -

On Jul 20, 4:20 pm, stevehunte...@hotmail.com wrote:
> if only 1 computer is impacted, these are the steps that never fail
>
We had a linux and windows box going to the same switch port. Pulled
the linux computer off and problem still remains. B and A switches
changed. Alls that remains is the windows box plugged into
a switch. I'm kind of thinking now that somehow the windows computer
is causing the problem, but I don't know how.
New software was installed that controls a robot via an rs-232 port.
Nobody can tell me though, whether this was installed before or after
the problem. Sometimes this will go fine for a couple of days and
then it will have a day where it just keeps dropping a network
connection.
Lately though the problem has gotten worse.

> 1. replace cable
> 2. replace cable again
Done this.
> 3. force full duplex
Don't know how on the Windows NT box.
> 4. change switch port to a known good port
Changed twice.
> 5. change NIC
> 6. replace NIC driver
Five and six have been done twice.
>
> when you change ports and NIC, clear arp cache or wait
>
> > yes, we replaced it twice.
> > The last time it happened. I was able to get on the fileserver and
> > ping from there and saw no response. I then pinged from my computer
> > and saw a response.
> > So now I'm thinking that the switch cache is getting messed up. But I
> > don't know how it is getting messed up. This is the second switch
> > that I have used. So my assumption is that there is something on the
> > network which is causing the problem (probably this computer).- Hide quoted text -
>
> > - Show quoted text -

There were 2 computers affected.

baggins wrote:

> On Jul 20, 4:20 pm, stevehunte...@hotmail.com wrote:
>> if only 1 computer is impacted, these are the steps that never fail
>>
> We had a linux and windows box going to the same switch port. Pulled
> the linux computer off and problem still remains. B and A switches
> changed. Alls that remains is the windows box plugged into
> a switch. I'm kind of thinking now that somehow the windows computer
> is causing the problem, but I don't know how.
> New software was installed that controls a robot via an rs-232 port.
> Nobody can tell me though, whether this was installed before or after
> the problem. Sometimes this will go fine for a couple of days and
> then it will have a day where it just keeps dropping a network
> connection.
> Lately though the problem has gotten worse.
>
>> 1. replace cable
>> 2. replace cable again
> Done this.
>> 3. force full duplex
> Don't know how on the Windows NT box.
>> 4. change switch port to a known good port
> Changed twice.
>> 5. change NIC
>> 6. replace NIC driver
> Five and six have been done twice.
>>
>> when you change ports and NIC, clear arp cache or wait
>>
>> > yes, we replaced it twice.
>> > The last time it happened. I was able to get on the fileserver and
>> > ping from there and saw no response. I then pinged from my computer
>> > and saw a response.
>> > So now I'm thinking that the switch cache is getting messed up. But I
>> > don't know how it is getting messed up. This is the second switch
>> > that I have used. So my assumption is that there is something on the
>> > network which is causing the problem (probably this computer).- Hide
>> > quoted text -
>>
>> > - Show quoted text -
>
> There were 2 computers affected.

I noticed that you mentioned a robot... is it possible that you are
experiencing some sort of EMI?