Karel Slaný authored 8 years ago and Ondřej Surý committed 8 years ago

Cookies are not stored in separate cache.

Karel Slaný authored 8 years ago and Ondřej Surý committed 8 years ago

The cookies layer injects a new query into the plan when a DADCOOKIE
response is detected. After failing the second attempt a TCP fallback is
signalised.

Karel Slaný authored 8 years ago and Ondřej Surý committed 8 years ago

The source address was always put into a subsequent query.

Karel Slaný authored 8 years ago and Ondřej Surý committed 8 years ago

This simplifies the response source identification on the cookie module.

this caused a bug in pseudo-lru table that negated
stickiness of values to table slots

If stdout is buffered, kr_log_info() might take ages to show up in the
output stream.  Since this stream could be interleaved with stderr
(e.g. kr_log_error()), it would be good to be able to see the messages
in the order in which they are generated.

Marek Vavrusa authored 8 years ago and Ondřej Surý committed 8 years ago

there are cases where switches or middle-boxes
block DNS/UDP answers >512 octets completely,
this gives user an option to mitigate that.
however, there are authoritatives serving
large answers that don't support TCP, so it's
a compromise as always

iterator already checks this and also chain length,
however these checks were omitted in the rrcache
CNAME unroll loop

there are cases where switches or middle-boxes
block DNS/UDP answers >512 octets completely,
this gives user an option to mitigate that.
however, there are authoritatives serving
large answers that don't support TCP, so it's
a compromise as always

in normal mode, only final CNAME target is refetched, but
not intermediate CNAMEs. intermediate CNAMEs are *never* cached,
but they are used to get final name for requery. in strict mode now,
every CNAME target is explicitly fetched even if it's a chained CNAME.

forks are connected with IPC pipes to process
group leader and can execute expressions on its
behalf. so running commands over all workers
is easy now:

> hostname() -- single
localhost

> map 'hostname()' -- all
localhost
localhost
localhost

* in the begin() layer, the incoming query is
  exposed as req->qsource.packet, it is invalidated
  after begin() and should not be modified
* the destination address (local interface) is
  also tracked for filtering purposes

now including <1ms, <50ms, <250ms, <500ms, <1.5s

during the consume step, the information about
upstream authoritative (address and current rtt)
is exposed in the request structure, just like
information about current query

fixed incorrect tagging in rrmap where secure rank would overflow

found by @gdemidov

* SOA MINTTL always preferred for negative answers
* only SOA used for negative answers

refs #75