[quagga-dev 4420] Re: Problem using undefined access-list in distrbute lists

Alain Ritoux alain.ritoux at 6wind.com
Thu Sep 28 17:21:54 BST 2006


Hi,

Paul Jakma wrote:

>> But if take a look at route-map management, this is the opposite 
>> behaviour,
>> i.e. when an acces-list is used in a a test, but not defined, the 
>> result is
>> NOMATCH.
>>
>> I think the second behaviour is better.
>
>
> Is this a bug or an intentional compatibility behaviour though?

I would think it's a bug.

>
>
> Yeah, very likely. We have a class problem here, as well as in how 
> delete/update of lists is managed, see, e.g.:
>
>   http://bugzilla.quagga.net/show_bug.cgi?id=292
>
> Which suggests the interfaces to these needs improving.

Just some idea:

In OSPF, route-maps simply don't keep pointers to ACL, they just
keep the name, and evey time the need to use the ACL, they do an
 access_list_lookup()
and managed a NULL result as a RMAP_NOMATCH

It seems very safe, and inspires me a suggestion:

If any ACL-user stuff behaves in a same way, ACL change and/or
creation and/or removal only needs to trigger an update fct
(which is the case). No more pb with free'd ACL. The hard thing
if for the update fct to call ad-hoc update fct for each ACL-user.
In OPSF, it seems to be the case.

I think the same thing can be used for prefix lists.

This would solve the class problem. Or do I miss something ?

Regards,
Alain
-- 
Alain RITOUX
Tel +33-1-39-30-92-32
Fax +33-1-39-30-92-11
visit our web http://www.6wind.com



More information about the Quagga-dev mailing list