* Thomas Bader <[EMAIL PROTECTED]> [2006-04-17 21:01]: > Claudio Jeker schrieb: > > On Mon, Apr 17, 2006 at 04:47:52PM +0200, Thomas Bader wrote: > >> [announce self is missing] > > Gosh! That's a fallout from Henning's nexthop self patch. > > I'll fix it ASAP. > Thanks, that works for me. > > I have established that I've mistaken the meaning of "self". I wanted to > use that to prevent routes learned over iBGP getting announced to eBGP > peers. They'll get announced though because they are (of course) from > the same AS. > > Probably I need a filter list to get the behaviour I expect. This works > well with IPv4, but not with IPv6: > > /etc/bgpd.conf:77: king bula sez: AF_INET only > > Whereas line 77 is: > > allow to $peer1 prefix fd:202c:ffff:80::/64 > > Is it planned to implement IPv6 prefixes in the filters?
hmm, I need to check again what is missing, I thought we had the basics all done... either I misremmeber or I missed sth. definately palnned. > > 3.8 shipped with basic IPv6 support but there were some bugs hidden in it. > > 3.9 fixed those issues and added special options to bgpctl to show only a > > specified address family. > OK. > > How well is IPv6 tested yet? I have seen some older postings which state > that it's not well tested. I plan to use v4 and v6, but I need the later > one for the time being only for testing and development purposes and not > for critical applications. Shall I worry about not fully tested v6 support? if nobody tests it, it won;t get tested... my test feed worked well whenever I tested. I do not have any v6 production systems. I would not worry too much. I doubt you'll see any dramatic failures. You might run into things like the above, where parts of the code are v4 only and complain. but that really doesn't affect v4 operation, right :) -- BS Web Services, http://www.bsws.de/ OpenBSD-based Webhosting, Mail Services, Managed Servers, ... Unix is very simple, but it takes a genius to understand the simplicity. (Dennis Ritchie)
