Deepz wrote:
> Hi, I have 2 win 2003 server machines with assigned Ip address
> 2.2.2.21( lets call it machine A) and 2.2.2.25 ( C)
> and n/w mask 255.255.255.0
>
> I have a Red hat machine ( machine B) that has 2 nics ( 2.2.2.30 and
> 2.2.2.31) on which I installed NISTnet with a warning
>
> rmmod: module rtc is not loaded
> Couldn't find rtc module -
> /dev/rtc will be mostly unusable after running istnet. Sorry about
> that.....
> To prevent this message recompile rtc (Enhanced Real Time Clock
> Support, under character devices) as a module.
>
> I am new to Linux and dont know what affect this warning may have on
> running Nistenet.
None at all. Nistnet needs to take over the realtime clock. When
nistnet is done, it leaves the rtc unusable. If the rtc code has been
compiled as a module you can just reload the module. Unfortunately, in
stock RedHat kernels, the rtc code is not compiled this way. If you
reboot when you are finished with nistnet, you may get some complaints
about rtc access, but that's all. I've been there & done that
occasionally, and had no trouble.
> So now all 3 machines are there are same subnet / switch.
>
> and the network configuration is
>
> [ A ] 2.2.2.21 --------------> 2.2.2.31 [ B ] 2.2.2.30
> <------------------- 2.2.2.25 [ C ]
>
> on B routing to C is via gateway 2.2.2.30
>
> on C routing to B is via gateway 2.2.2.31
Put A and C on separate class C networks, and add the appropriate routes
through B. Anything else is just going to cause you trouble.
> But I dint add any route from B to either A/C . When I do that it gives
> me error saying
>
> Host name lookup failure
You don't need to add routes on B. Were you adding routes by name
instead of by address?
>
> pinging B to both A / C works fine both ways ( ip_frowarding is set)
>
> Entries in NISTnet are
> src dest TOS delay duplicates
> .....
>
> A 2.2.2.30(B) - 10 ms 60%
> A 2.2.2.31(B) - 10 ms 60%
> C 2.2.2.30(B) - 10 ms 60%
> C 2.2.2.31(B) - 10 ms 60%
> A C - 10 ms 60%
>
>
> When I run 2 different instances of same application on both A and C
> that use TCp sockets to connect to each other ,the aplication runs the
> same way whether NISTnet is set ON or OFF. ( when I disable any of the
> nic on linux box , the applcation stops , meaning routing between A and
> C is hapening through the linux box)
>
> concerns/observations
>
> 1. when I ping from B to A/ C duplicates come as expected but delay is
> always less than 1 ms.
When you are testing, it's a lot easier to debug a single entry (from A
to C, for example) that adds a large deterministic delay and nothing
else. Then run ping continuously from A to C (I assume windows will let
you do that).
> 2. Is the problem is due to rtc warning I got .
Shouldn't be.
> I mean is nistnet
> installed properly.
Looks like it. Use cnistnet to see if the kernel module has the delay
right.
|