Re: Heimdal and OpenBSD 2.6

[Assar Westerlund <assar@sics.se>]

( I assume some of the errors here are because of textual problems in
mail? )

Naomaru Itoi <itoi@eecs.umich.edu> writes:
> 23:01:38.026245 alice.citi.umich.edu.39995 > citi.umich.edu.domain: 33487+ AA
>  babble.citi.umich.edu. (39)
> 23:01:38.026766 citi.umich.edu.domain > alice.citi.umich.edu.39995: 33487* 0/
>  (93)

that would be asking for an AAAA (ipv6 address) and getting back that
there are no such thing

> 23:01:38.027264 alice.citi.umich.edu.16477 > citi.umich.edu.domain: 50512+ AA
>  babble.citi.umich.edu.citi.umich.edu. (54)
> 23:01:38.027738 citi.umich.edu.domain > alice.citi.umich.edu.16477: 50512 NXD
> in* 0/1/0 (108)

once more, when the default search domain added

> 23:01:38.028493 alice.citi.umich.edu.1430 > citi.umich.edu.domain: 49831+ A? 
> ble.citi.umich.edu. (39)
> 23:01:38.029075 citi.umich.edu.domain > alice.citi.umich.edu.1430: 49831* 1/3
> A babble.citi.umich.edu (179)

A look-up for babble, A (v4).  Now we should just send a packet there.

> 23:01:38.029811 alice.citi.umich.edu.43935 > citi.umich.edu.domain: 46927+ AA
>  babble.citi.umich.edu. (39)

But we don't, instead we do more lookups.  Weird.  Now, lots of things
have changed in getaddrinfo() between 2.6 and 2.7, so getaddrinfo()
behaving in some funky way is not a bad guess.  To try to verify if
that's the case, can you hack lib/roken/Makefile to include
getaddrinfo.o in libroken_la_LIBADD, rebuild in `lib/roken' and then
make sure you re-link `kinit' and try again?

/assar