I've just found out that with RIP-v1 everything works fine in a network as long as the same netmask is used on all interfaces of all routers from the network, no matter if that netmask is corresponding or not to Class A or B or C.
My understanding up to now was that only the default netmask corresponding to that IP class could be used, i.e /24 for Class C, /16 for Class B and /8 for Class A.
I used /28 netmask and the network I simulated on PaketTracer run smoothly.
But when I changed the nemask to /29 on only one interface of a router, the corresponding subnet from that interface was not anymore advertised by the router through the RIP updates.
What is the technical explanation behind this behaviour?
I'll just explain your original example, FE with /29 not being advertised through serial with /28
serial : 192.168.10.114/28
ethernet : 192.168.10.65/29
Major network for both these interface address are 192.168.10.0, because it's a class C address, so we use the standard /24 mask to derive the major network.
Now when advertising E/29 route over SE/28.
1)The router checks if the route belongs to the major network of the address configured on the interface. The answer is yes.
2)Second it checks if the SE interface mask (/28) matches the subnet mask of route from E (/29). It is not.
If the the answer in the second step is No, it will no advertise that route out SE.
Similarly the ethernet interface will not advertise the address from SE interface.
Now try the same example by changing the Ethernet address to another major network like 10.0.0.1/8. This will use the second rule I mentioned in my previous post and be advertised out SE interface.
- and you mean that if the 2 netmasks are different the router cannot compute the subnet?
I thought that it would compute the subnet 192.168.10.64 and advertise it. Why it can't?
It will not do this, because it will lead to ambiguity.