* 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.
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
* plugin/dnssec: filter bitmap also for NXDOMAIN responses
We change nxdomain to nodata, so at the point when we receive the
reply it can be nxdomain or nodata. In both cases we should filter the
nsec bitmap.
Change the code and add explicit tests for this.
* More tests
* plugin/health: add lameduck mode
Add a way to configure lameduck more, i.e. set health to false, stop
polling plugins. Then wait for a duration before shutting down. As the
health middleware is configured early on in the plugin list, it will
hold up all other shutdown, meaning we still answer queries.
* Add New
* More tests
* golint
* remove confusing text
* plugin/dnssec: check validityperiod of RRSIGs
Somehow we missed implementing this. If a sig a retrieved from the
cache, but not valid anymore, regenerate it instead of server invalid
signatures.
Fixes#1378
* drop from cache after 3/4 validity
* six days means 6 days
* Improve plugin/cache metrics
* Add coredns_cache_prefetch_total metric to track number of prefetches.
* Remove unnecessary Cache.get() call which would incorrectly increment
cache counters.
* Initialize all counters and gauges at zero.
* Allow prefetching of a single request per ttl
The original implementation didn't allow prefetching queries which are
only requested once during the duration of a TTL. The minimum amount of
queries which had to be seen was therefore capped at 2.
This change also implements a real prefetch test. The existing test was
a noop and always passed regardless of any prefetch implementation.
* Fix prefetching for items with a short TTL
The default prefetch threshold (percentage) is 10% of the lifetime of a
cache item. With the previous implementation, this disabled prefetching
for all items with a TTL < 10s (the resulting percentage would be 0, at
which point a cached item is already discarded).
This change uses a time based threshold calculation and ensures that
a prefetch is triggered at a TTL of 1 at the latest.
* Fix wrong duration reporting of cached responses
The logging and metrics plugins (among others) included the duration of
a cache prefetch in the request latency of client request. This change
fixes this wrong reporting and executes the prefetch request in a
goroutine in the background.
* Update vendor
Signed-off-by: Yong Tang <yong.tang.github@outlook.com>
* Add route53 plugin
This fix adds route53 plugin so that it is possible to
query route53 record through CoreDNS.
Signed-off-by: Yong Tang <yong.tang.github@outlook.com>
The cache plugin always returned a minimum TTL of 5 seconds, regardless
of the actual TTL of the records. A cache is not authoritative for the
record TTL and should not extend it.
* plugin/health: add 'overloaded metrics'
Query our on health endpoint and record (and export as a metric) the
time it takes. The Get has a 5s timeout, that, when reached, will set
the metric duration to 5s. The actually call "I'm I overloaded" is left
to an external entity.
* README
* golint and govet
* and the tests
* Align plugin/template usage and syntax with other plugins
* Use new fallthrough logic in plugin/template
* Use zone name normalization for plugin/template
* Test fallthrough parsing in plugin/template
* Rework scoping of match checks
Most matches are not plugin global but per template. The plugin does only a
very rough check while detailed checks are done per-template.
Per template checks include:
- Zones
- Class/Type
- Regex
- Fallthrough
* Remove trailing `.` from fully qualified domain names
* Register template metrics with zone/class/type instead of regex
* Remove trailing fqdn dot from multiple testcases
* Add a template plugin
The template plugin matches the incoming query by class, type and regex
and templates a response with go templates.
* Fix go style errors
* Fix template README example
* Fix corefile example in plugin/template
* Clarify plugin/template/README.md
Add more details and external links where needed.
* Fix code issues in plugin/template
* Add template metrics
* Add section and template to template plugin metrics
* Fix style / remove extra newline on go imports
* Fix typo in plugin/template/README.md
* Update README.md
I've change the format a bit in a PR that I merged yesterday.
* Add authority section to plugin/template
* Fix naming of incoming query name in plugin/template/README.md
* Fix doc syntax in plugin/template/README.md
* Add authority section to plugin/template/README.md config overview
* Add metric labels to plugin/template/README.md metrics section
* Use request.Request to pass state to the template matcher
* Add pkg/fall for Fallthrough
Move this into it's own package to facilitate tests. Important bug
was fixed: make the names fully qualified.
Add fall package to hosts, reverse, etcd, and fix kubernetes and any
tests. The k8s tests are still as-is, might need a future cleanup.
This is probably the first in a series to fix "crazy" pod queries.
If the namespace doesn't exist return NXDOMAIN.
It might be worth extending this 1:1 to findServices as well.
* Add manual pages
Generate manual pages from the README and extend README with Name and
Description sections.
The generation requires 'ronn' which may not be available. Just check in
all generated manual pages.
* plugin/dnssec: implement shotgun from CloudFlare
Put a whole bunch of types in the NSEC bitmap and remove the one that's
being asked for.
Add more records for queries to the apex, SOA, DNSKEY, MX.
* Update k8s client-go to v6.0.0
This fix updates k8s client-go to v6.0.0 as CoreDNS is supported
in 1.9 and v6.0.0 is the recommended version.
There are quite some massive changes that need to be made:
1. k8s.io/client-go/pkg/api/v1 has been changed to k8s.io/api/v1 (repo changed from `client-go` to `api`)
2. kubernetes.Clientset adds one extra layer, so that `kubernetes.Clientset.Services()` and like has been changed to `kubernetes.Clientset.CoreV1().Services()`
Also, we have to stick with specific commits of `k8s.io/apimachinery` and the newly introduced `k8s.io/api`
because go dep still could not figure out the right version to fetch.
Signed-off-by: Yong Tang <yong.tang.github@outlook.com>
* Update vendor with `dep ensure --update` and `dep prune`
Signed-off-by: Yong Tang <yong.tang.github@outlook.com>
After initial startup, see if prometheus is loaded and if so, register
our metrics with it.
Stop doing the init() func and just use the sync.Once so we don't double
registrer our metrics.
This leave most of the code intact, but we need to stop vendoring
prometheus, because, again, plugins what want to use it. Not vendoring
prometheus makes my forward metrics show up again. Code looks bit
convoluted, but works:
~~~
c.OnStartup(func() error {
once.Do(func() {
m := dnsserver.GetConfig(c).Handler("prometheus")
if m == nil {
return
}
if x, ok := m.(*metrics.Metrics); ok {
x.MustRegister(RequestCount)
x.MustRegister(RcodeCount)
x.MustRegister(RequestDuration)
x.MustRegister(HealthcheckFailureCount)
x.MustRegister(SocketGauge)
}
})
})
~~~
* dnstest: add multirecorder
This adds a new recorder that captures all messages written to it. This
can be useful when, for instance, testing AXFR which can write muliple
messages back to the client.
* docs
prometheus.Handler is deprecated according to the godoc for the package so
instead we're using promhttp.
Additionally, we are exposing the Registry that metrics is using so other
plugins that are not inside of coredns can read the registry. Otherwise, if
we kept using the Default one, there's no way to access that from outside
of the coredns repo since it is vendored.
If external plugins wanted to extend metrics there was no way since
zoneNames couldn't be initialized. Now plugins can call New to get an
instance of Metrics that they can extend.
