I have written a small expect script to telnet a FreeBSD machine and
send some commands to that machine remotely, through Windows XP
machine
These commands are being sent periodically in a loop. After sending
few command , the scripts throws the following exception:


write(spawn_id=]: broken pipe
    while executing
"send -s "ipfw pipe 1 config delay $mod_delay \r""
    invoked from within
"expect -nobrace 1# {

                 for { set cycle 0 } { $cycle < $no_of_cycles } {incr
cycle } {
                     puts "###..."
    invoked from within
"expect {
          "1#" {

                 for { set cycle 0 } { $cycle < $no_of_cycles } {incr
cycle } {


   Can anybody please let me know the cause for this.
   Is this issue of the freeBSD machine or my PC. Can what can be the
workaround for this

Extract from my script:
..
..
expect {
          "1#" {

                 for { set cycle 0 } { $cycle < $no_of_cycles } {incr
cycle } {
                     puts "##### CYCLE NUMBER -- $cycle #####"
                      while { $delay <= $max_delay } {
                        set mod_delay [join "$delay ms" ""]
                        puts stdout "\nDelay set to $mod_delay\n"

                        send -s "ipfw pipe 1 config delay $mod_delay
\r"
                        set delay [expr $delay + $incr_delay]
                        after $wait
                    }

                   set delay $min_delay
                }
            }
         timeout {  puts "Timeout !!!"
                    exp_continue }

        }
..
..
..
Additional Info: I am using Dummynet application of FreeBSD to
simulate Network delays.
                      And these commands are used to insert variable
delays in packets received from        one port and are then sent out
on another interface.

On Apr 14, 12:11 pm, "Ranjan" <ranjan.kap...@gmail.com> wrote:
> I have written a small expect script to telnet a FreeBSD machine and
> send some commands to that machine remotely, through Windows XP
> machine
> These commands are being sent periodically in a loop. After sending
> few command , the scripts throws the following exception:
> write(spawn_id=]: broken pipe

This tells you one thing: the telnet process has died. From there, you
should investigate:

(1) Which side broke the connection first (use WireShark on your
windows machine, or a tcpdump/tethereal on the unix one).

(2) If the first F packet comes from the remote host, what happened
there (the in.telnetd and children). I don't know FreeBSD but if there
were something like strace(Linux) or truss(Solaris), I'd try it.

-Alex

I started ethereal on windows XP machine. Before the problem occurred,
there were some packets like "This frame is a (suspected)
retransmission", and there after it is sending ARP packets for the
FreeBSD PC IP. I am now able to ping the machine manually, but telnet
to the machine is very slow. (I waiting for 30 seconds after entering
password). Both the machines are connected on same switch , and I am
able to successfully telnet other machines (ie there is no networking
issue otherwise).

On Apr 16, 7:40 am, "Ranjan" <ranjan.kap...@gmail.com> wrote:
> I started ethereal on windows XP machine. Before the problem occurred,
> there were some packets like "This frame is a (suspected)
> retransmission", and there after it is sending ARP packets for the
> FreeBSD PC IP. I am now able to ping the machine manually, but telnet
> to the machine is very slow. (I waiting for 30 seconds after entering
> password). Both the machines are connected on same switch , and I am
> able to successfully telnet other machines (ie there is no networking
> issue otherwise).

Maybe a half/full-duplex mismatch or autoneg failure between the
switch and one of the hosts ? Look at the collision counts on both
sides...

-Alex

On Apr 16, 5:03 pm, "Alexandre Ferrieux"
<alexandre.ferri...@gmail.com> wrote:
> On Apr 16, 7:40 am, "Ranjan" <ranjan.kap...@gmail.com> wrote:
>
> > I started ethereal on windows XP machine. Before the problem occurred,
> > there were some packets like "This frame is a (suspected)
> > retransmission", and there after it is sending ARP packets for the
> > FreeBSD PC IP. I am now able to ping the machine manually, but telnet
> > to the machine is very slow. (I waiting for 30 seconds after entering
> > password). Both the machines are connected on same switch , and I am
> > able to successfully telnet other machines (ie there is no networking
> > issue otherwise).
>
> Maybe a half/full-duplex mismatch or autoneg failure between the
> switch and one of the hosts ? Look at the collision counts on both
> sides...
>
> -Alex

Alex,

I too couldnt interpret much from the packet captures.
I think I need to isolate the two machines and check if network
congestion has something to do.
Or else I will try installing tcl on freebsd (dont know if it will
work..) and do directly on that machine....

Ranjan wrote:
> Both the machines are connected on same switch , and I am
> able to successfully telnet other machines (ie there is no networking
> issue otherwise).

Sometimes individual ports on switches go bad or get flakey. Try moving 
wires about and see if the problem moves to another machine.


-- 
   Darren New / San Diego, CA, USA (PST)
     His kernel fu is strong.
     He studied at the Shao Linux Temple.

On Apr 14, 5:11 am, "Ranjan" <ranjan.kap...@gmail.com> wrote:
> I have written a small expect script to telnet a FreeBSD machine and
> send some commands to that machine remotely, through Windows XP
> machine
> These commands are being sent periodically in a loop. After sending
> few command , the scripts throws the following exception:
>
> write(spawn_id=]: broken pipe
>     while executing
> "send -s "ipfw pipe 1 config delay $mod_delay \r""
>     invoked from within
> "expect -nobrace 1# {
>
>                  for { set cycle 0 } { $cycle < $no_of_cycles } {incr
> cycle } {
>                      puts "###..."
>     invoked from within
> "expect {
>           "1#" {
>
>                  for { set cycle 0 } { $cycle < $no_of_cycles } {incr
> cycle } {
>
>    Can anybody please let me know the cause for this.
>    Is this issue of the freeBSD machine or my PC. Can what can be the
> workaround for this
>
> Extract from my script:
> .
> .
> expect {
>           "1#" {
>
>                  for { set cycle 0 } { $cycle < $no_of_cycles } {incr
> cycle } {
>                      puts "##### CYCLE NUMBER -- $cycle #####"
>                       while { $delay <= $max_delay } {
>                         set mod_delay [join "$delay ms" ""]
>                         puts stdout "\nDelay set to $mod_delay\n"
>
>                         send -s "ipfw pipe 1 config delay $mod_delay
> \r"
>                         set delay [expr $delay + $incr_delay]
>                         after $wait
>                     }
>
>                    set delay $min_delay
>                 }
>             }
>          timeout {  puts "Timeout !!!"
>                     exp_continue }
>
>         }
> .
> .
> .
> Additional Info: I am using Dummynet application of FreeBSD to
> simulate Network delays.
>                       And these commands are used to insert variable
> delays in packets received from        one port and are then sent out
> on another interface.

You also use "puts stdout" to report you loop to the user on the XP
machine you
should use send_user "##### ....".

Carl

I moved the two machines to an isolated switch but still the problem
is occuring.
It has been noticed that (after the problem occurs) the ping reply
comes in <1ms but the telnet to the machine is very slow, you need to
wait for many seconds before the next line appears. I changed the XP
machine but the telnet was still slow, and after restarting the
freeBSD machine everything becomes OK. So seems probably something
goes wrong on FreeBSD machine.

Now I want to avoid telnet the FreeBSD machine. Is there is any
package of Tcl compatible with FreeBSD?

Thanks,
Ranjan

On Apr 19, 7:45 pm, Ranjan <ranjan.kap...@gmail.com> wrote:
> machine but the telnet was still slow, and after restarting the
> freeBSD machine everything becomes OK. So seems probably something
> goes wrong on FreeBSD machine.

Rebooting and being fine is a bit frustrating...
Again, have you looked at the collision count (ifconfig) and ethernet
autoneg status (mii-diag et al) ?

-Alex

On 19 Apr, 18:45, Ranjan <ranjan.kap...@gmail.com> wrote:
> I moved the two machines to an isolated switch but still the problem
> is occuring.
> It has been noticed that (after the problem occurs) the ping reply
> comes in <1ms but the telnet to the machine is very slow, you need to
> wait for many seconds before the next line appears. I changed the XP
> machine but the telnet was still slow, and after restarting the
> freeBSD machine everything becomes OK. So seems probably something
> goes wrong on FreeBSD machine.
>
> Now I want to avoid telnet the FreeBSD machine. Is there is any
> package of Tcl compatible with FreeBSD?
>
> Thanks,
> Ranjan

Hows about

set sfd [ socket <host> <port> ]
spawn -open $sfd

Derek

On Apr 20, 1:27 am, Alexandre Ferrieux <alexandre.ferri...@gmail.com>
wrote:
> On Apr 19, 7:45 pm, Ranjan <ranjan.kap...@gmail.com> wrote:
>
> > machine but the telnet was still slow, and after restarting the
> > freeBSD machine everything becomes OK. So seems probably something
> > goes wrong on FreeBSD machine.
>
> Rebooting and being fine is a bit frustrating...
> Again, have you looked at the collision count (ifconfig) and ethernet
> autoneg status (mii-diag et al) ?
>
> -Alex


Alex,

Does ifconfig command show collision count? Also mii-diag,
ifplugstatus command is not working on FreeBSD machine.

Regards,
Ranjan

On Apr 20, 2:04 pm, derek.phi...@csr.com wrote:
>
> > Now I want to avoid telnet the FreeBSD machine. Is there is any
> > package of Tcl compatible with FreeBSD?
>
> Hows about
> set sfd [ socket <host> <port> ]
> spawn -open $sfd

He has *network* problems. Replacing telnet by a handmade socket
changes nothing (telnet is also on top of TCP).

-Alex

In article <1176731829.533894.124450@n76g2000hsh.googlegroups.com>,
Ranjan <ranjan.kapoor@gmail.com> wrote:
			.
			.
			.
>Or else I will try installing tcl on freebsd (dont know if it will
>work..) and do directly on that machine....
>

I'm uncertain of what you're saying here.  I can assure you,
though, that Tcl has a looooooooong history of working quite
nicely under FreeBSD--did you have a question about that?

In article <3g79g4-6qh.ln1@lairds.us>, Cameron Laird <claird@lairds.us> wrote:
>In article <1176731829.533894.124450@n76g2000hsh.googlegroups.com>,
>Ranjan <ranjan.kapoor@gmail.com> wrote:
>			.
>			.
>			.
>>Or else I will try installing tcl on freebsd (dont know if it will
>>work..) and do directly on that machine....
>>
>
>I'm uncertain of what you're saying here.  I can assure you,
>though, that Tcl has a looooooooong history of working quite
>nicely under FreeBSD--did you have a question about that?

There *was* a question.

I like to think <URL: http://wiki.tcl.tk/2422 > answers it.