[ntp:questions] (Software) timeserver for windows being broadcast-able incl. keys

Tom Smith smith at cag.zko.hp.com
Wed Mar 14 21:54:38 UTC 2007


Erik wrote:
> The clients that need to be time-corrected reside in the following
> address-ranges:
> 
> 145.47.51.[016-167]
> with netmask 255.255.255.128

The netmask is not consistent with the address range
The netmask should be 255.255.255.0 and the broadcast
address for that network would then be 145.47.51.255

> 
> 145.47.52.[032-175]
> with netmask 255.255.255.128

The netmask is not consistent with the address range
The netmask should be 255.255.255.0 and the broadcast
broadcast address 145.47.52.255

> 
> 145.47.53.[076-091]
> with netmask 255.255.255.194

The netmask is wrong. It should be 255.255.255.192
The broadcast address would then be 145.47.53.127

> 
> - Does the above require me to use more than one broadcast-address?

Yes.

> - Does the broadcast address always need to be the highest address
> covered by the netmask-address?

Yes.

> - These broadcast-addresses: are they claimed/occupied by the
> broadcast-command just as if they were the address of a PC in the
> network?
 > i.e. they can not be used anymore by other clients in the network?

These are just ordinary IP broadcast addresses that would be
used by any IP broadcast application. It is the way you specify
that an IP packet should be sent to every address in the network.
There is nothing reserved or otherwise special about them.

> - does this result in certain requirements for the PC's network
> address hosting the broadcast service (server). That is: does this PC
> have to have this same address or not?

Yes, it has to have interfaces in each of these networks. Otherwise,
you have to configure your network (switches, routers) to pass
broadcast packets addressed to a wider network from a smaller one.
In that case, in addition to reconfiguring your network, you
would use a minimal-scope broadcast address of 145.47.63.255
and you would have to have an interface on the server in the
range 145.47.[48-63].*.

> 
> I am sorry; pretty much a newby on this terrain  I'm affraid
> 
>> Why don't you just look at how the previous server was configured
>> (the one that was in use by the existing clients) and copy that?
> 
> There was no time server before this one; the clients were just
> prtepared for such a service, the service itself was never there.
> Therefore, no possibility to peek...

OK. Then if there is no legacy system to support, why don't
you just reconfigure the clients into something rational?
For example, get rid of the keys and the broadcast, and just
point each client to the server with a simple "server [address]"
declaration.

> 
> Thanks once again
> Erik
> 




More information about the questions mailing list