[Dibbler] DHCPv6 server negative tests failure

Tomasz Mrugalski thomson at klub.com.pl
Mon Jan 30 14:15:56 CET 2012


On 30.01.2012 13:19, Ron, Omer1 wrote:
> Hi,
> All tests below had DHCPv6 server failure.
> Please find on each test the received result vs. expected result
Can you describe your test environment? Are these automated tests or
manual? Is this a validation environment that is available, like TAHI or
TAHI extensions? I'm interested in your tests if you are willing to
share them.

> TEST 1: 
>     Verify server sends UseMulticast status code when client sends
> unicast Renew
> 
> the client send DHCPv6 Renew using the server's unicast address.
> 
> The server instead of sending status code of 5 'UseMulticast' send
> status code ‘Success’
Long known bug. See bug #83 in bugzilla:
https://klub.com.pl/bugzilla3/show_bug.cgi?id=83

> TEST 2:
>     Verify server sends UseMulticast status code when client sends a
> unicast Request
> 
> the client send DHCPv6 Request using the server's unicast address.
> 
> The server instead of sending status code of 5 'UseMulticast' send
> status code ‘Success’
Depending on the details, that is actually not a valid test. Client can
send REQUEST using unicast if it knows server unicast AND has address
with sufficient scope to communicate with that unicast. For example:
Client goes through SARR and gets global address (using IA_NA with
iaid=1). Later client decides that another address is needed. It then
sends another REQUEST using unicast (using IA_NA with iaid=2). That's a
valid scenario. Client *can* use unicast transmission to send request as
it has address with sufficient scope to receive answer.

This test should be modified to verify that REQUEST sent to unicast
address is sent from global address (or at least not from link-local
address).

That's how much theory says. I'm 100% sure that it is not implemented
correctly in Dibbler, so your lead on finding bug here is good :)

> TEST 3:
>     Verify server sends Reply with lifetimes of 0 when client attempts
> to renew unknown address (relevant also for rebind message)
> 
> the client send DHCPv6 Renew with a valid IAID containing unknown address
> 
> IA IPv6 address is wrong , expecting the same IP send in the renew
> 
> IA Preferred lifetime is 4294967295, expecting 0
> 
> IA Valid lifetime is 4294967295, expecting 0
Ok, that looks like a valid bug. Please file a bug in bugzilla.

> TEST 4:
>     Verify server sends NotOnLink status code when client sends Request
> with invalid link address (relevant also for confirm message)
> 
> the client send DHCPv6 Request with an address not considered "on link"
> 
> Found unexpected status code Success (0), expecting NotOnLink (4)
Yikes. Another bug. Please report it in bugzilla.

Good tests! I assume you run them on latest Dibbler version (0.8.1) or
GIT code?

Thanks a lot for testing Dibbler.
Tomek Mrugalski


More information about the Dibbler mailing list