dns/man/coredns-forward.7

.\" Generated by Mmark Markdown Processer - mmark.miek.nl
.TH "COREDNS-FORWARD" 7 "March 2026" "CoreDNS" "CoreDNS Plugins"

.SH "NAME"
.PP
\fIforward\fP - facilitates proxying DNS messages to upstream resolvers.

.SH "DESCRIPTION"
.PP
The \fIforward\fP plugin re-uses already opened sockets to the upstreams. It supports UDP, TCP and
DNS-over-TLS and uses in band health checking.

.PP
When it detects an error a health check is performed. This checks runs in a loop, performing each
check at a \fI0.5s\fP interval for as long as the upstream reports unhealthy. Once healthy we stop
health checking (until the next error). The health checks use a recursive DNS query (\fB\fC. IN NS\fR)
to get upstream health. Any response that is not a network error (REFUSED, NOTIMPL, SERVFAIL, etc)
is taken as a healthy upstream. The health check uses the same protocol as specified in \fBTO\fP. If
\fB\fCmax_fails\fR is set to 0, no checking is performed and upstreams will always be considered healthy.

.PP
When \fIall\fP upstreams are down it assumes health checking as a mechanism has failed and will try to
connect to a random upstream (which may or may not work).

.SH "SYNTAX"
.PP
In its most basic form, a simple forwarder uses this syntax:

.PP
.RS

.nf
forward FROM TO...

.fi
.RE

.IP \(bu 4
\fBFROM\fP is the base domain to match for the request to be forwarded. Domains using CIDR notation
that expand to multiple reverse zones are not fully supported; only the first expanded zone is used.
.IP \(bu 4
\fBTO...\fP are the destination endpoints to forward to. The \fBTO\fP syntax allows you to specify
a protocol, \fB\fCtls://9.9.9.9\fR or \fB\fCdns://\fR (or no protocol) for plain DNS. The number of upstreams is
limited to 15.


.PP
Multiple upstreams are randomized (see \fB\fCpolicy\fR) on first use. When a healthy proxy returns an error
during the exchange the next upstream in the list is tried.

.PP
Extra knobs are available with an expanded syntax:

.PP
.RS

.nf
forward FROM TO... {
    except IGNORED\_NAMES...
    force\_tcp
    prefer\_udp
    expire DURATION
    max\_idle\_conns INTEGER
    max\_fails INTEGER
    max\_connect\_attempts INTEGER
    tls CERT KEY CA
    tls\_servername NAME
    policy random|round\_robin|sequential
    health\_check DURATION [no\_rec] [domain FQDN]
    max\_concurrent MAX
    next RCODE\_1 [RCODE\_2] [RCODE\_3...]
    failfast\_all\_unhealthy\_upstreams
    failover RCODE\_1 [RCODE\_2] [RCODE\_3...]
}

.fi
.RE

.IP \(bu 4
\fBFROM\fP and \fBTO...\fP as above.
.IP \(bu 4
\fBIGNORED_NAMES\fP in \fB\fCexcept\fR is a space-separated list of domains to exclude from forwarding.
Requests that match none of these names will be passed through.
.IP \(bu 4
\fB\fCforce_tcp\fR, use TCP even when the request comes in over UDP.
.IP \(bu 4
\fB\fCprefer_udp\fR, try first using UDP even when the request comes in over TCP. If response is truncated
(TC flag set in response) then do another attempt over TCP. In case if both \fB\fCforce_tcp\fR and
\fB\fCprefer_udp\fR options specified the \fB\fCforce_tcp\fR takes precedence.
.IP \(bu 4
\fB\fCmax_fails\fR is the number of subsequent failed health checks that are needed before considering
an upstream to be down. If 0, the upstream will never be marked as down (nor health checked).
Default is 2.
.IP \(bu 4
\fB\fCmax_connect_attempts\fR caps the total number of upstream connect attempts
performed for a single incoming DNS request. Default value of 0 means no per-request
cap.
.IP \(bu 4
\fB\fCexpire\fR \fBDURATION\fP, expire (cached) connections after this time, the default is 10s.
.IP \(bu 4
\fB\fCmax_idle_conns\fR \fBINTEGER\fP, maximum number of idle connections to cache per upstream for reuse.
Default is 0, which means unlimited.
.IP \(bu 4
\fB\fCtls\fR \fBCERT\fP \fBKEY\fP \fBCA\fP define the TLS properties for TLS connection. From 0 to 3 arguments can be
provided with the meaning as described below

.RS
.IP \(en 4
\fB\fCtls\fR - no client authentication is used, and the system CAs are used to verify the server certificate
.IP \(en 4
\fB\fCtls\fR \fBCA\fP - no client authentication is used, and the file CA is used to verify the server certificate
.IP \(en 4
\fB\fCtls\fR \fBCERT\fP \fBKEY\fP - client authentication is used with the specified cert/key pair.
The server certificate is verified with the system CAs
.IP \(en 4
\fB\fCtls\fR \fBCERT\fP \fBKEY\fP  \fBCA\fP - client authentication is used with the specified cert/key pair.
The server certificate is verified using the specified CA file

.RE
.IP \(bu 4
\fB\fCtls_servername\fR \fBNAME\fP allows you to set a server name in the TLS configuration; for instance 9.9.9.9
needs this to be set to \fB\fCdns.quad9.net\fR. Using TLS forwarding but not setting \fB\fCtls_servername\fR results in anyone
being able to man-in-the-middle your connection to the DNS server you are forwarding to. Because of this,
it is strongly recommended to set this value when using TLS forwarding.


.PP
Per destination endpoint TLS server name indication is possible in the form of \fB\fCtls://9.9.9.9%dns.quad9.net\fR.
  \fB\fCtls_servername\fR must not be specified when using per destination endpoint TLS server name indication
  as it would introduce clash between the server name indication spectifications. If destination endpoint
  is to be reached via a port other than 853 then the port must be appended to the end of the destination
  endpoint specifier. In case of port 10853, the above string would be: \fB\fCtls://9.9.9.9%dns.quad9.net:10853\fR.

.IP \(bu 4
\fB\fCpolicy\fR specifies the policy to use for selecting upstream servers. The default is \fB\fCrandom\fR.

.RS
.IP \(en 4
\fB\fCrandom\fR is a policy that implements random upstream selection.
.IP \(en 4
\fB\fCround_robin\fR is a policy that selects hosts based on round robin ordering.
.IP \(en 4
\fB\fCsequential\fR is a policy that selects hosts based on sequential ordering.

.RE
.IP \(bu 4
\fB\fChealth_check\fR configure the behaviour of health checking of the upstream servers

.RS
.IP \(en 4
\fB\fC<duration>\fR - use a different duration for health checking, the default duration is 0.5s.
.IP \(en 4
\fB\fCno_rec\fR - optional argument that sets the RecursionDesired-flag of the dns-query used in health checking to \fB\fCfalse\fR.
The flag is default \fB\fCtrue\fR.
.IP \(en 4
\fB\fCdomain FQDN\fR - set the domain name used for health checks to \fBFQDN\fP.
If not configured, the domain name used for health checks is \fB\fC.\fR.

.RE
.IP \(bu 4
\fB\fCmax_concurrent\fR \fBMAX\fP will limit the number of concurrent queries to \fBMAX\fP.  Any new query that would
raise the number of concurrent queries above the \fBMAX\fP will result in a REFUSED response. This
response does not count as a health failure. When choosing a value for \fBMAX\fP, pick a number
at least greater than the expected \fIupstream query rate\fP * \fIlatency\fP of the upstream servers.
As an upper bound for \fBMAX\fP, consider that each concurrent query will use about 2kb of memory.
.IP \(bu 4
\fB\fCnext\fR If the \fB\fCRCODE\fR (i.e. \fB\fCNXDOMAIN\fR) is returned by the remote then execute the next plugin. If no next plugin is defined, or the next plugin is not a \fB\fCforward\fR plugin, this setting is ignored
.IP \(bu 4
\fB\fCfailfast_all_unhealthy_upstreams\fR - determines the handling of requests when all upstream servers are unhealthy and unresponsive to health checks. Enabling this option will immediately return SERVFAIL responses for all requests. By default, requests are sent to a random upstream.
.IP \(bu 4
\fB\fCfailover\fR - By default when a DNS lookup fails to return a DNS response (e.g. timeout), \fIforward\fP will attempt a lookup on the next upstream server. The \fB\fCfailover\fR option will make \fIforward\fP do the same for any response with a response code matching an \fB\fCRCODE\fR ( e.g. \fB\fCSERVFAIL\fR、\fB\fCREFUSED\fR). \fB\fCNOERROR\fR cannot be used. If all upstreams have been tried, the response from the last attempt is returned.


.PP
Also note the TLS config is "global" for the whole forwarding proxy if you need a different
\fB\fCtls_servername\fR for different upstreams you're out of luck.

.PP
On each endpoint, the timeouts for communication are set as follows:

.IP \(bu 4
The dial timeout by default is 30s, and can decrease automatically down to 1s based on early results.
.IP \(bu 4
The read timeout is static at 2s.


.SH "METADATA"
.PP
The forward plugin will publish the following metadata, if the \fImetadata\fP
plugin is also enabled:

.IP \(bu 4
\fB\fCforward/upstream\fR: the upstream used to forward the request


.SH "METRICS"
.PP
If monitoring is enabled (via the \fIprometheus\fP plugin) then the following metric are exported:

.IP \(bu 4
\fB\fCcoredns_forward_healthcheck_broken_total{}\fR - count of when all upstreams are unhealthy,
and we are randomly (this always uses the \fB\fCrandom\fR policy) spraying to an upstream.
.IP \(bu 4
\fB\fCcoredns_forward_max_concurrent_rejects_total{}\fR - count of queries rejected because the
number of concurrent queries were at maximum.
.IP \(bu 4
\fB\fCcoredns_proxy_request_duration_seconds{proxy_name="forward", to, rcode}\fR - histogram per upstream, RCODE
.IP \(bu 4
\fB\fCcoredns_proxy_healthcheck_failures_total{proxy_name="forward", to, rcode}\fR- count of failed health checks per upstream.
.IP \(bu 4
\fB\fCcoredns_proxy_conn_cache_hits_total{proxy_name="forward", to, proto}\fR- count of connection cache hits per upstream and protocol.
.IP \(bu 4
\fB\fCcoredns_proxy_conn_cache_misses_total{proxy_name="forward", to, proto}\fR - count of connection cache misses per upstream and protocol.


.PP
Where \fB\fCto\fR is one of the upstream servers (\fBTO\fP from the config), \fB\fCrcode\fR is the returned RCODE
from the upstream, \fB\fCproto\fR is the transport protocol like \fB\fCudp\fR, \fB\fCtcp\fR, \fB\fCtcp-tls\fR.

.PP
The following metrics have recently been deprecated:
* \fB\fCcoredns_forward_healthcheck_failures_total{to, rcode}\fR
  * Can be replaced with \fB\fCcoredns_proxy_healthcheck_failures_total{proxy_name="forward", to, rcode}\fR
* \fB\fCcoredns_forward_requests_total{to}\fR
  * Can be replaced with \fB\fCsum(coredns_proxy_request_duration_seconds_count{proxy_name="forward", to})\fR
* \fB\fCcoredns_forward_responses_total{to, rcode}\fR
  * Can be replaced with \fB\fCcoredns_proxy_request_duration_seconds_count{proxy_name="forward", to, rcode}\fR
* \fB\fCcoredns_forward_request_duration_seconds{to, rcode}\fR
  * Can be replaced with \fB\fCcoredns_proxy_request_duration_seconds{proxy_name="forward", to, rcode}\fR

.SH "EXAMPLES"
.PP
Proxy all requests within \fB\fCexample.org.\fR to a nameserver running on a different port:

.PP
.RS

.nf
example.org {
    forward . 127.0.0.1:9005
}

.fi
.RE

.PP
Send all requests within \fB\fClab.example.local.\fR to \fB\fC10.20.0.1\fR, all requests within \fB\fCexample.local.\fR (and not in
\fB\fClab.example.local.\fR) to \fB\fC10.0.0.1\fR, all others requests to the servers defined in \fB\fC/etc/resolv.conf\fR, and
caches results. Note that a CoreDNS server configured with multiple \fIforward\fP plugins in a server block will evaluate those
forward plugins in the order they are listed when serving a request.  Therefore, subdomains should be
placed before parent domains otherwise subdomain requests will be forwarded to the parent domain's upstream.
Accordingly, in this example \fB\fClab.example.local\fR is before \fB\fCexample.local\fR, and \fB\fCexample.local\fR is before \fB\fC.\fR.

.PP
.RS

.nf
\&. {
    cache
    forward lab.example.local 10.20.0.1
    forward example.local 10.0.0.1
    forward . /etc/resolv.conf
}

.fi
.RE

.PP
The example above is almost equivalent to the following example, except that example below defines three separate plugin
chains (and thus 3 separate instances of \fIcache\fP).

.PP
.RS

.nf
lab.example.local {
    cache
    forward . 10.20.0.1
}
example.local {
    cache
    forward . 10.0.0.1
}
\&. {
    cache
    forward . /etc/resolv.conf
}

.fi
.RE

.PP
Load balance all requests between three resolvers, one of which has a IPv6 address.

.PP
.RS

.nf
\&. {
    forward . 10.0.0.10:53 10.0.0.11:1053 [2003::1]:53
}

.fi
.RE

.PP
Forward everything except requests to \fB\fCexample.org\fR

.PP
.RS

.nf
\&. {
    forward . 10.0.0.10:1234 {
        except example.org
    }
}

.fi
.RE

.PP
Proxy everything except \fB\fCexample.org\fR using the host's \fB\fCresolv.conf\fR's nameservers:

.PP
.RS

.nf
\&. {
    forward . /etc/resolv.conf {
        except example.org
    }
}

.fi
.RE

.PP
Proxy all requests to 9.9.9.9 using the DNS-over-TLS (DoT) protocol, and cache every answer for up to 30
seconds. Note the \fB\fCtls_servername\fR is mandatory if you want a working setup, as 9.9.9.9 can't be
used in the TLS negotiation. Also set the health check duration to 5s to not completely swamp the
service with health checks.

.PP
.RS

.nf
\&. {
    forward . tls://9.9.9.9 {
       tls\_servername dns.quad9.net
       health\_check 5s
    }
    cache 30
}

.fi
.RE

.PP
Or configure other domain name for health check requests

.PP
.RS

.nf
\&. {
    forward . tls://9.9.9.9 {
       tls\_servername dns.quad9.net
       health\_check 5s domain example.org
    }
    cache 30
}

.fi
.RE

.PP
Or with multiple upstreams from the same provider

.PP
.RS

.nf
\&. {
    forward . tls://1.1.1.1 tls://1.0.0.1 {
       tls\_servername cloudflare\-dns.com
       health\_check 5s
    }
    cache 30
}

.fi
.RE

.PP
Or when you have multiple DoT upstreams with different \fB\fCtls_servername\fRs, you can do the following:

.PP
.RS

.nf
\&. {
    forward . 127.0.0.1:5301 127.0.0.1:5302
}

\&.:5301 {
    forward . tls://8.8.8.8 tls://8.8.4.4 {
        tls\_servername dns.google
    }
}

\&.:5302 {
    forward . tls://1.1.1.1 tls://1.0.0.1 {
        tls\_servername cloudflare\-dns.com
    }
}

.fi
.RE

.PP
The following would try 1.2.3.4 first. If the response is \fB\fCNXDOMAIN\fR, try 5.6.7.8. If the response from 5.6.7.8 is \fB\fCNXDOMAIN\fR, try 9.0.1.2.

.PP
.RS

.nf
\&. {
  forward . 1.2.3.4 {
    next NXDOMAIN
  }
  forward . 5.6.7.8 {
    next NXDOMAIN
  }
  forward . 9.0.1.2 {
  }
}

.fi
.RE

.PP
In the following example, if the response from \fB\fC1.2.3.4\fR is \fB\fCSERVFAIL\fR or \fB\fCREFUSED\fR, it will try \fB\fC5.6.7.8\fR. If the response from \fB\fC5.6.7.8\fR is \fB\fCSERVFAIL\fR or \fB\fCREFUSED\fR, it will try \fB\fC9.0.1.2\fR.

.PP
.RS

.nf
\&. {
  forward . 1.2.3.4 5.6.7.8 9.0.1.2 {
     policy sequential
     failover SERVFAIL REFUSED
  }
}

.fi
.RE

.SH "SEE ALSO"
.PP
RFC 7858
\[la]https://tools.ietf.org/html/rfc7858\[ra] for DNS over TLS.