* add manifest list

* nit

* change check

In Makefile.release `shell uname -m` was used to find the current architecture.
On amd64 systems this might be reported as `x86_64` depending on the Linux.
However, in order to have manifest docker images we need to stick with
golang's architecture tag of `amd64`.

This fix changes instead so that the correct architecture conforming to
golang (`amd64`) could be build.
Signed-off-by: Yong Tang <yong.tang.github@outlook.com>

Add a dynamic SOA record (at least the serial and minttl). This create
another interface that should be implemented by the backends.

For now default to returning epoch in as a uint32 (no change from
before).

Lower the minTTL returned to 30s (from 60s)

Fix typo in kubernetes/README.md and remove DEV-README.md as it is stale
and information on the website is more up to date.

Remove large sections of text in plugin.md; just talk about how to
structure your plugin and docs.

* add goroutine to check hosts file for updates

* rename parseFile to parseReader, remove extra error check

* readme: more tests

Add dnssec and file plugin to the test readme. This requires creating a
bunch of files with the right content. Doing so already unconvered an
unconditional type assertion in DNSSEC. This PR will include the fix for
that as well.

Also extended the snippets in the file plugin README, so that they are
whole Corefile - showing more value and checking all corefile snippets.

Create outliner right now is the kubernetes plugin, because even setting
the right env vars will result in:

open /var/run/secrets/kubernetes.io/serviceaccount/token: no such file or directory":

Which we can't create for a test.

* lint

Don't use the 'check and godep' target when building for a release, this
is now repeated 6 times for no reason as this is already checked on
travis.

Some other cleanups in the documentation as well.

* release: add more build targets

This adds amd64, ppc and s390. Rework some other builds/directories.

This builds:
% find build
build
build/darwin
build/darwin/x86_64
build/darwin/x86_64/coredns
build/linux
build/linux/ppc64
build/linux/ppc64/coredns
build/linux/x86_64
build/linux/x86_64/coredns
build/linux/arm64
build/linux/arm64/coredns
build/linux/s390
build/linux/s390/coredns
build/linux/arm
build/linux/arm/coredns

% make -f Makefile.release tar
rm -rf release && mkdir release
tar -zcf release/coredns_0.9.9_linux_x86_64.tgz -C build/linux/x86_64 coredns
tar -zcf release/coredns_0.9.9_darwin_x86_64.tgz -C build/darwin/x86_64 coredns
tar -zcf release/coredns_0.9.9_linux_armv6l.tgz -C build/linux/arm coredns
tar -zcf release/coredns_0.9.9_linux_armv8l.tgz -C build/linux/arm64 coredns
tar -zcf release/coredns_0.9.9_linux_ppc64le.tgz -C build/linux/ppc64 coredns
tar -zcf release/coredns_0.9.9_linux_s390x.tgz -C build/linux/s390 coredns

Checking:
% for i in $(find build -type f); do file $i; done
build/darwin/x86_64/coredns: Mach-O 64-bit x86_64 executable, flags:<NOUNDEFS>
build/linux/ppc64/coredns: ELF 64-bit LSB executable, 64-bit PowerPC or cisco 7500, version 1 (SYSV), statically linked, stripped
build/linux/x86_64/coredns: ELF 64-bit LSB executable, x86-64, version 1 (SYSV), statically linked, stripped
build/linux/arm64/coredns: ELF 64-bit LSB executable, ARM aarch64, version 1 (SYSV), statically linked, stripped
build/linux/s390/coredns: ELF 64-bit MSB executable, IBM S/390, version 1 (SYSV), statically linked, stripped
build/linux/arm/coredns: ELF 32-bit LSB executable, ARM, EABI5 version 1 (SYSV), statically linked, stripped

* code review

People are using it in prod and some other updates.

Emphasize `go get` as the main way to get the project.

* integration ci

* rename test

* unfunctionalize DoIntegrationTests

* alphabetize expected answers

* Enable out-of-cluster test

* Enable out-of-cluster test

* move integration tests back to ci repo

Some cleanup in proxy and dnstap:
* just use time pkg directly and side step the indirection for Epoch
* Use Set in SetQueryEpoch to be more Go like. (Looked like a reader)
* Don't maintain two sets of time, we already track start, so use that.
* Use time.Time and convert when needed
* dedent the toDnstap function and put in a separate file

If we don't need it, don't initialize it.

Fixes #1156

* core: allow all CIDR ranges in zone specifications

Allow (e.g.) a v4 reverse on a /17. If a zone is specified in such a
way a FilterFunc is set in the config. This filter is checked against
incoming queries.

For all other queries this adds a 'x != nil' check which will not impact
performace too much. Benchmark function is added as well to check for
this as wel.

Add multiple tests in tests/server_reverse_test.go.

Benchmark shows in the non-reverse case this hardly impact the speed:

~~~
classless:
pkg: github.com/coredns/coredns/core/dnsserver
BenchmarkCoreServeDNS-4   	 1000000	      1431 ns/op	      16 B/op	       1 allocs/op

pkg: github.com/coredns/coredns/core/dnsserver
BenchmarkCoreServeDNS-4   	 1000000	      1429 ns/op	      16 B/op	       1 allocs/op

master:
pkg: github.com/coredns/coredns/core/dnsserver
BenchmarkCoreServeDNS-4   	 1000000	      1412 ns/op	      16 B/op	       1 allocs/op

pkg: github.com/coredns/coredns/core/dnsserver
BenchmarkCoreServeDNS-4   	 1000000	      1429 ns/op	      16 B/op	       1 allocs/op
~~~

* README.md updates

This flakyness is driving me crazy. Add the sleep(3s) back.

* stalebot: use wontfix-stalebot

I've added two new labels: wontfix-stalebot and pinned.

Wontfix-stalebot will be used by stalebot and pinned can be used to keep
issues open.

* later as well

Make the locking slightly smarter in Evict and add benchmark function.

Seems a bit faster (there was some variance while performing these
benchmarks)

Master:
BenchmarkCache-2   	 1000000	      2317 ns/op	       0 B/op	       0 allocs/op
BenchmarkCache-2   	 1000000	      2032 ns/op	       0 B/op	       0 allocs/op

This branch:
BenchmarkCache-2   	 1000000	      1806 ns/op	       0 B/op	       0 allocs/op
BenchmarkCache-2   	 1000000	      1809 ns/op	       0 B/op	       0 allocs/op

These file still had middleware in their name, rename to plugin.

* core: add ServeDNS benchmark

Add benchmark function so we can perf test future additions to the
servers' ServeDNS function.

* naming

* plugin/kubernetes: wait until api is ready

Wait for HasSynced before allowing startup to avoid startup race.

Also do a small refactor in findServices() to pull a check out of the
loop - only needs to be done once.

* sigh

Various other changes.

* plugin/dnssec; insert and sign DS records

Sign a delegation as well and insert DS records.

Fixes #698

* better

We support ENV variables in config files, add an example in the one for
metric, note that `localhost:` is a valid host (in Go), so the
TestReadme will parse this correctly.

Fixes #1150

* indexing

* corrections

* Don't number the plugins

The number is not needed, because the ordering is already specified.
It's also annoying when you move plugins, because you need to renumber
them. Remove this.

'go gen' shows no changes in the generated files, meaning this just
works.

* better naming

* plugin/autopath: Add namespace selector and metrics

Add a namespace, so autopathing only is performed in this namespace.
This will make caching work for the cluster again.

Also export metrics that we've done a successful autopath

* dont shadow

* Fix

* Back the namespacing changes

* healthchecks: check on every 3rd failure

Check on every third failure and some cleanups to make this possible. A
failed healthcheck will never increase Fails, a successfull healthceck
will reset Fails to 0. This is a chance this counter now drops below 0,
making the upstream super? healthy.

This removes the okUntil smartness and condences everything back to 1
metrics: Fails; so it's simpler in that regard.

Timout errors are *not* attributed to the local upstream, and don't get
counted into the Fails anymore. Meaning the 'dig any isc.org' won't kill
your upstream.

Added extra test the see if the Fails counter gets reset after 3 failed
connection.

There is still a disconnect beween HTTP healthceck working the proxy (or
lookup) not being able to connect to the upstream.

* Fix tests

Small optimization.

Flaky test because it used a fix port; just drop it. Might be worthwhile
to return it as a unit tests, but on that level we can't use a coredns
"process" to exercise the reload.

See https://github.com/kubernetes/kubernetes/pull/53604, explicitaly add
test to make sure we do the right thing.

* doc update

Go through all README and fix mistakes, extend example and let more
corefile snippets be test for validity.

* Cant use spefic addr in test

Add Counter metrics and fix duration to use upstream name (and only use
it when we have one).

Fix the documentation to reflect this.

Fixes #1134

* add NSEC-records if entry exists but the requested type is missing

* added test for dnssec fix