[quagga-dev 11512] Re: [PATCH] pimd merge

Everton Marques everton.marques at gmail.com
Fri Sep 19 21:00:53 BST 2014


Thank you for the awesome remarks! See my comments below.

On Thu, Sep 18, 2014 at 3:07 PM, Christian Franke <nobody at nowhere.ws> wrote:

> - Should static_add_ipv4/static_delete_ipv4 be kept around? They don't
>   seem to be used.


> - The DEFUN for "ip mroute" uses "[<1-255>]" to specify an optional
>   distance. This won't work all that well currently. The reason being
>   that anything in square brackets will be treated as optional string.
>   So "ip mroute abcd" will be considered valid
>   input. The easiest way around this is to e.g. use a DEFUN without the
>   distance and add an ALIAS with the distance as a mandatory argument.
>   You have to make sure that your config write output doesn't put a "0"
>   if the user didn't specify a distance, as "0" wouldn't be a valid
>   input for <1-255> and that would cause an error when the config gets
>   loaded.


> - ZEBRA_IPV4_NEXTHOP_LOOKUP/rib_match_ipv4 were solely used by bgpd to
>   get information concerning recursive nexthops. Therefore,
>   rib_match_ipv4 excludes any BGP routes from the results. This behavior
>   has been inherited by rib_match_ipv4_safi in your patch, so
>   ZEBRA_IPV4_NEXTHOP_LOOKUP_MRIB will exclude all bgp routes. I wonder
>   whether this is the correct behavior for RPF lookups.

Fixed. I have added a flag about considering BGP routes:

rib_match_ipv4_safi (struct in_addr addr, safi_t safi, int skip_bgp)

May you please take a look?

> One final observation which I made just now is that "no ip mroute"
> doesn't seem to work correctly for some reason. The route disappears
> from "show running-config", but it still shows up in "show ip rpf". See
> the attached session for an example.

Fixed. static_uninstall_ipv4() was missing SAFI support.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quagga.net/pipermail/quagga-dev/attachments/20140919/815ad75d/attachment.html>

More information about the Quagga-dev mailing list