* allow only one k8s section

* add test

* plugin/forward: add it

This moves coredns/forward into CoreDNS. Fixes as a few bugs, adds a
policy option and more tests to the plugin.

Update the documentation, test IPv6 address and add persistent tests.

* Always use random policy when spraying

* include scrub fix here as well

* use correct var name

* Code review

* go vet

* Move logging to metrcs

* Small readme updates

* Fix readme

This adds a generic way of start a check function to check a backend.
This package can be used to kick off healthchecks. The package makes
sure only 1 is run at any one time.

It should allow for:
See upstream error -> kick off healthcheck

and not to worry about overwhelming the upstream with a barrage of
queries.

Add man-pages for the new plugins.

* tune usage of var global, add limit to options

* update readme for minimal values

* useless change to quick-off codecov

* fix msgs for min values and tune the flag for end of reload usage, with a 'maybe' option

* adding UT for min values, adding MD5 of corefile on the log

* add new policy always_first to mimic windows dns resolvers
fill documentation, add UT and cleanup fmt

* change name of policy from always_first to first. Update docs

As plugin/proxy always returns compressed messages, it's important to
set this before calling Scrub(), as some messages will be unnecessarily
truncated otherwise.

HTTP healthchecking will be implicitely activated for proxy upstream
hosts, even if not configured. The README states that not using the
health_check directive will disable HTTP healthchecks though.

It seems to me that the availability of the HealthCheck.Path attribute
is used as indicator whether HTTP healthchecks should be used or not.
The normalizeCheckURL() function didn't check that attribute though,
always returning a CheckURL. This would increase the healthcheck failure
on every third failure in plugin/proxy, without any possibility for the
upstream host to be marked as healthy again. This would eventually
remove all upstream hosts from the serving pool.

This adds the inital top-level OWNERS file. Each plugin should also get
one, but that will be more work.

Ref: #1454

Golint fix to make goreportcard happy
Signed-off-by: Yong Tang <yong.tang.github@outlook.com>

This fix uses external context package for server-tls.go
This fix is a follow up to 1468.
Signed-off-by: Yong Tang <yong.tang.github@outlook.com>

This was used for docker automation or someting?
I don't think there are uses for this, lets remove it.

With Go 1.9 you *can* include the std lib's context package and nothing
breaks. However we never officially made the move (and grpc also doesn't
ues the std lib's one).

Standardize all plugins on using the extern context package.

Fixes #1466

Let caddyfile parser give us the token, the comma sep thing also didn't
work.

Fixes #1446

* plugin.md updates

Talk about return code, logging and other bits.

Also checked the code for fmt.Printf logging, there is none.

Fixes #1449 #1450

* review

* more typos

* Add reload directive

* gofmt

* Fix default jitter and error message

* remove unneeded call to NextArg, add a couple negative setup tests

* Review feedback

* testpr3

* hack

* chop

* tolerate coredns/coredns get fail

* remove TEST from readme

Last release I forgot steps, put them in Makefile.release. Generating
the manual files will be skipped if 'ronn' can't be found.

* Enable forward

* Regen all docs

CoreVersion needs to be documented for golint, but we can't use too much
text because we grep this file for the coredns version.

* move flag blacklisting to main

Doing it in init() is the wrong place or something else changed. Doing
it in main() makes it easy to see *when* this happens.

* keep trapsignals

* plugin/forward: add out of tree forward plugin

This is a simpler proxy than *proxy*, include by default so it is easier
to switch (i.e. no recompile). It lacks features compared to proxy (did I
say it was simpler), but does cache udp and tcp connection, so it is
faster than proxy.

* Muck with the makefile

* Versioning forward now

Enable this Caddy plugin by default. Docs will go up coredns.io for this
as well.

See https://caddyserver.com/docs/on

These test cases became invalid with the fix of message truncation and
don't validate anything.

* Fix truncation of messages longer than permitted by the client

CoreDNS currently doesn't respect the maximum response size advertised
by the client and returns the full answer on a message with the TC bit
set. This breaks client implementations which rely on DNS servers
respecting the advertised size limit, for example the Ruby stdlib
client. It also has negative network performance implications, as large
messages will be split up into multiple UDP packets, even though the
client will discard the truncated response anyway.

While RFC 2181 permits the response of partial RRSets, finding the
correct number of records fitting into the advertised response size is
non-trivial. As clients should ignore truncated messages, this change
simply removes the full RRSet on truncated messages.

* Remove incorrect etcd test assertion

If a client requests a TXT record larger than its advertised buffer
size, a DNS server should _not_ respond with the answer, but truncate
the message and set the TC bit, so that the client can retry using TCP.

In order to track the rollout status of CoreDNS versions, add the common
build_info metric.

See #1400 and the discussion in that bug.

Fixes #1400

Fix the error handling. Log when we have an error during any of the
transfer state. And if there isn't an error transfer the zones.

Also fix the tests in test/ so we, at least, check the initial transfer.

Update the docs to show more about how errors are handled.

Ref #1400