Query policies

This module can block, rewrite, or alter inbound queries based on user-defined policies. It does not affect queries generated by the resolver itself, e.g. when following CNAME chains etc.

Each policy rule has two parts: a filter and an action. A filter selects which queries will be affected by the policy, and action which modifies queries matching the associated filter.

Typically a rule is defined as follows: filter(action(action parameters), filter parameters). For example, a filter can be suffix which matches queries whose suffix part is in specified set, and one of possible actions is policy.DENY, which denies resolution. These are combined together into policy.suffix(policy.DENY, {todname('badguy.example.')}). The rule is effective when it is added into rule table using policy.add(), please see examples below.

This module is enabled by default because it implements mandatory RFC 6761 logic. When no rule applies to a query, built-in rules for special-use and locally-served domain names are applied. These rules can be overriden by action policy.PASS. For debugging purposes you can also add modules.unload('policy') to your config to unload the module.

Filters

A filter selects which queries will be affected by specified Actions. There are several policy filters available in the policy. table:

policy.all(action)

Always applies the action.

policy.pattern(action, pattern)

Applies the action if query name matches a Lua regular expression.

policy.suffix(action, suffix_table)

Applies the action if query name suffix matches one of suffixes in the table (useful for “is domain in zone” rules).

Note

For speed this filter requires domain names in DNS wire format, not textual representation, so each label in the name must be prefixed with its length. Always use convenience function policy.todnames() for automatic conversion from strings! For example:

policy.suffix(policy.DENY, policy.todnames({'example.com', 'example.net'}))
policy.suffix_common(action, suffix_table[, common_suffix])
Parameters
  • action – action if the pattern matches query name

  • suffix_table – table of valid suffixes

  • common_suffix – common suffix of entries in suffix_table

Like policy.suffix() match, but you can also provide a common suffix of all matches for faster processing (nil otherwise). This function is faster for small suffix tables (in the order of “hundreds”).

It is also possible to define custom filter function with any name.

policy.custom_filter(state, query)
Parameters
  • state – Request processing state kr_layer_state, typically not used by filter function.

  • query – Incoming DNS query as kr_query structure.

Returns

An action function or nil if filter did not match.

Typically filter function is generated by another function, which allows easy parametrization - this technique is called closure. An practical example of such filter generator is:

function match_query_type(action, target_qtype)
    return function (state, query)
        if query.stype == target_qtype then
            -- filter matched the query, return action function
            return action
        else
            -- filter did not match, continue with next filter
            return nil
        end
    end
end

This custom filter can be used as any other built-in filter. For example this applies our custom filter and executes action policy.DENY on all queries of type HINFO:

-- custom filter which matches HINFO queries, action is policy.DENY
policy.add(match_query_type(policy.DENY, kres.type.HINFO))

Actions

An action is a function which modifies DNS request, and is either of type chain or non-chain:

Non-chain actions

Following actions stop the policy matching on the query, i.e. other rules are not evaluated once rule with following actions matches:

policy.PASS

Let the query pass through; it’s useful to make exceptions before wider rules. For example:

More specific whitelist rule must preceede generic blacklist rule:

-- Whitelist 'good.example.com'
policy.add(policy.pattern(policy.PASS, todname('good.example.com.')))
-- Block all names below example.com
policy.add(policy.suffix(policy.DENY, {todname('example.com.')}))
policy.DENY

Deny existence of names matching filter, i.e. reply NXDOMAIN authoritatively.

policy.DENY_MSG(message)

Deny existence of a given domain and add explanatory message. NXDOMAIN reply contains an additional explanatory message as TXT record in the additional section.

policy.DROP

Terminate query resolution and return SERVFAIL to the requestor.

policy.REFUSE

Terminate query resolution and return REFUSED to the requestor.

policy.TC

Force requestor to use TCP. It sets truncated bit (TC) in response to true if the request came through UDP, which will force standard-compliant clients to retry the request over TCP.

policy.REROUTE({{subnet, target}, ...})

Reroute IP addresses in response matching given subnet to given target, e.g. {'192.0.2.0/24', '127.0.0.0'} will rewrite ‘192.0.2.55’ to ‘127.0.0.55’, see renumber module for more information. See policy.add() and do not forget to specify that this is postrule. Quick example:

-- this policy is enforced on answers
-- therefore we have to use 'postrule'
-- (the "true" at the end of policy.add)
policy.add(policy.REROUTE({'192.0.2.0/24', '127.0.0.0'}), true)
policy.ANSWER({ type = { rdata=data, [ttl=1] } }, [nodata=false])

Overwrite Resource Records in responses with specified values.

  • type - RR type to be replaced, e.g. [kres.type.A] or numeric value.

  • rdata - RR data in DNS wire format, i.e. binary form specific for given RR type. Set of multiple RRs can be specified as table { rdata1, rdata2, ... }. Use helper function kres.str2ip() to generate wire format for A and AAAA records.

  • ttl - TTL in seconds. Default: 1 second.

  • nodata - If type requested by client is not configured in this policy:

    • true: Return empty answer (NODATA).

    • false: Ignore this policy and continue processing other rules.

    Default: false.

-- policy to change IPv4 address and TTL for example.com
policy.add(
    policy.suffix(
        policy.ANSWER(
            { [kres.type.A] = { rdata=kres.str2ip('192.0.2.7'), ttl=300 } }
        ), { todname('example.com') }))
-- policy to generate two TXT records (specified in binary format) for example.net
policy.add(
    policy.suffix(
        policy.ANSWER(
            { [kres.type.TXT] = { rdata={'\005first', '\006second'}, ttl=5 } }
        ), { todname('example.net') }))

More complex non-chain actions are described in their own chapters, namely:

Chain actions

Following actions act on request and then processing continue until first non-chain action (specified in the previous section) is triggered:

policy.MIRROR(ip_address)

Send copy of incoming DNS queries to a given IP address using DNS-over-UDP and continue resolving them as usual. This is useful for sanity testing new versions of DNS resolvers.

policy.add(policy.all(policy.MIRROR('127.0.0.2')))
policy.FLAGS(set, clear)

Set and/or clear some flags for the query. There can be multiple flags to set/clear. You can just pass a single flag name (string) or a set of names. Flag names correspond to kr_qflags structure. Use only if you know what you are doing.

policy.QTRACE

Pretty-print DNS response packets from authoritative servers into debug logs for the query and its sub-queries. It’s useful for debugging weird DNS servers. Note that debug-level logs are off by default; see log_level().

-- log answers from all authoritative servers involved in resolving
-- requests for example.net. and its subdomains
policy.add(policy.suffix(policy.QTRACE, policy.todnames({'example.net'})))
policy.REQTRACE

Pretty-print DNS requests from clients into the verbose log. It’s useful for debugging weird DNS clients. Debug-level logging must be enabled for this policy to be effective; see log_level(). It makes most sense together with Views and ACLs (enabling per-client) and probably with verbose logging those request (e.g. DEBUG_ALWAYS).

policy.DEBUG_ALWAYS

Enable extra verbose logging for all requests, including cache hits. See caveats for policy.DEBUG_IF().

policy.DEBUG_CACHE_MISS

Enable extra verbose logging but print logs only for requests which required information which was not available locally (i.e. requests which forced resolver to communicate over network). Intended usage is for debugging problems with remote servers. This action typically produces less logs than policy.DEBUG_ALWAYS but all caveats from policy.DEBUG_IF() apply as well.

policy.add(policy.suffix(
    policy.DEBUG_CACHE_MISS,
    policy.todnames({'example.com.'})))
policy.DEBUG_IF(test_function)
Parameters

test_function – Function with single argument of type kr_request which returns true if debug logs for a given request should be generated and false otherwise.

Enable extra verbose logging but print logs only for requests which match condition specified by test_function. This allows to fine-tune which requests should be printed.

Warning

Logging on debug level has significant performance impact on resolver and might also overload you logging system because one request can easily generate tens of kilobytes of logs. Always use appropriate Filters to limit number of requests triggering this action to a minimum!

Note

test_function is evaluated only when request is finished. As a result debug logs for all requests must be collected until request is finished because it is not possible to know beforehand how test_function at the end evaluates given request. When a request is finalized logs are either printed or thrown away.

Example usage which gathers verbose logs for all requests in subtree dnssec-failed.org. and prints debug logs for all requests finished with states different than kres.DONE (most importantly kres.FAIL, see kr_layer_state).

policy.add(policy.suffix(
    policy.DEBUG_IF(function(req)
                        return (req.state ~= kres.DONE)
                    end),
    policy.todnames({'dnssec-failed.org.'})))

Custom actions

policy.custom_action(state, request)
Parameters
Returns

Returning a new kr_layer_state prevents evaluating other policy rules. Returning nil creates a chain action and allows to continue evaluating other rules.

This is real example of an action function:

-- Custom action which generates fake A record
local ffi = require('ffi')
local function fake_A_record(state, req)
    local answer = req:ensure_answer()
    if answer == nil then return nil end
    local qry = req:current()
    if qry.stype ~= kres.type.A then
        return state
    end
    ffi.C.kr_pkt_make_auth_header(answer)
    answer:rcode(kres.rcode.NOERROR)
    answer:begin(kres.section.ANSWER)
    answer:put(qry.sname, 900, answer:qclass(), kres.type.A, '\192\168\1\3')
    return kres.DONE
end

This custom action can be used as any other built-in action. For example this applies our fake A record action and executes it on all queries in subtree example.net:

policy.add(policy.suffix(fake_A_record, policy.todnames({'example.net'})))

The action function can implement arbitrary logic so it is possible to implement complex heuristics, e.g. to deflect Slow drip DNS attacks or gray-list resolution of misbehaving zones.

Warning

The policy module currently only looks at whole DNS requests. The rules won’t be re-applied e.g. when following CNAMEs.

Forwarding

Forwarding action alters behavior for cache-miss events. If an information is missing in the local cache the resolver will forward the query to another DNS resolver for resolution (instead of contacting authoritative servers directly). DNS answers from the remote resolver are then processed locally and sent back to the original client.

Actions policy.FORWARD(), policy.TLS_FORWARD() and policy.STUB() accept up to four IP addresses at once and the resolver will automatically select IP address which statistically responds the fastest.

policy.FORWARD(ip_address)
policy.FORWARD({ ip_address, [ip_address, ...] })

Forward cache-miss queries to specified IP addresses (without encryption), DNSSEC validate received answers and cache them. Target IP addresses are expected to be DNS resolvers.

-- Forward all queries to public resolvers https://www.nic.cz/odvr
policy.add(policy.all(
   policy.FORWARD(
       {'2001:148f:fffe::1', '2001:148f:ffff::1',
        '185.43.135.1', '193.14.47.1'})))

A variant which uses encrypted DNS-over-TLS transport is called policy.TLS_FORWARD(), please see section Forwarding over TLS protocol (DNS-over-TLS).

policy.STUB(ip_address)
policy.STUB({ ip_address, [ip_address, ...] })

Similar to policy.FORWARD() but without attempting DNSSEC validation. Each request may be either answered from cache or simply sent to one of the IPs with proxying back the answer.

This mode does not support encryption and should be used only for Replacing part of the DNS tree. Use policy.FORWARD() mode if possible.

-- Answers for reverse queries about the 192.168.1.0/24 subnet
-- are to be obtained from IP address 192.0.2.1 port 5353
-- This disables DNSSEC validation!
policy.add(policy.suffix(
    policy.STUB('192.0.2.1@5353'),
    {todname('1.168.192.in-addr.arpa')}))

Note

By default, forwarding targets must support EDNS and 0x20 randomization. See example in Replacing part of the DNS tree.

Forwarding over TLS protocol (DNS-over-TLS)

policy.TLS_FORWARD({ {ip_address, authentication}, [...] } )

Same as policy.FORWARD() but send query over DNS-over-TLS protocol (encrypted). Each target IP address needs explicit configuration how to validate TLS certificate so each IP address is configured by pair: {ip_address, authentication}. See sections below for more details.

Policy policy.TLS_FORWARD() allows you to forward queries using Transport Layer Security protocol, which hides the content of your queries from an attacker observing the network traffic. Further details about this protocol can be found in RFC 7858 and IETF draft dprive-dtls-and-tls-profiles.

Queries affected by policy.TLS_FORWARD() will always be resolved over TLS connection. Knot Resolver does not implement fallback to non-TLS connection, so if TLS connection cannot be established or authenticated according to the configuration, the resolution will fail.

To test this feature you need to either configure Knot Resolver as DNS-over-TLS server, or pick some public DNS-over-TLS server. Please see DNS Privacy Project homepage for list of public servers.

Note

Some public DNS-over-TLS providers may apply rate-limiting which makes their service incompatible with Knot Resolver’s TLS forwarding. Notably, Google Public DNS doesn’t work as of 2019-07-10.

When multiple servers are specified, the one with the lowest round-trip time is used.

CA+hostname authentication

Traditional PKI authentication requires server to present certificate with specified hostname, which is issued by one of trusted CAs. Example policy is:

policy.TLS_FORWARD({
    {'2001:DB8::d0c', hostname='res.example.com'}})
  • hostname must be a valid domain name matching server’s certificate. It will also be sent to the server as SNI.

  • ca_file optionally contains a path to a CA certificate (or certificate bundle) in PEM format. If you omit that, the system CA certificate store will be used instead (usually sufficient). A list of paths is also accepted, but all of them must be valid PEMs.

Key-pinned authentication

Instead of CAs, you can specify hashes of accepted certificates in pin_sha256. They are in the usual format – base64 from sha256. You may still specify hostname if you want SNI to be sent.

TLS Examples

modules = { 'policy' }
-- forward all queries over TLS to the specified server
policy.add(policy.all(policy.TLS_FORWARD({{'192.0.2.1', pin_sha256='YQ=='}})))
-- for brevity, other TLS examples omit policy.add(policy.all())
-- single server authenticated using its certificate pin_sha256
policy.TLS_FORWARD({{'192.0.2.1', pin_sha256='YQ=='}})  -- pin_sha256 is base64-encoded
-- single server authenticated using hostname and system-wide CA certificates
policy.TLS_FORWARD({{'192.0.2.1', hostname='res.example.com'}})
-- single server using non-standard port
policy.TLS_FORWARD({{'192.0.2.1@443', pin_sha256='YQ=='}})  -- use @ or # to specify port
-- single server with multiple valid pins (e.g. anycast)
policy.TLS_FORWARD({{'192.0.2.1', pin_sha256={'YQ==', 'Wg=='}})
-- multiple servers, each with own authenticator
policy.TLS_FORWARD({ -- please note that { here starts list of servers
    {'192.0.2.1', pin_sha256='Wg=='},
    -- server must present certificate issued by specified CA and hostname must match
    {'2001:DB8::d0c', hostname='res.example.com', ca_file='/etc/knot-resolver/tlsca.crt'}
})

Forwarding to multiple targets

With the use of policy.slice() function, it is possible to split the entire DNS namespace into distinct slices. When used in conjuction with policy.TLS_FORWARD(), it’s possible to forward different queries to different targets.

policy.slice(slice_func, action[, action[, ...])
Parameters
  • slice_func – slicing function that returns index based on query

  • action – action to be performed for the slice

This function splits the entire domain space into multiple slices (determined by the number of provided actions). A slice_func is called to determine which slice a query belongs to. The corresponding action is then executed.

policy.slice_randomize_psl(seed=os.time() / 3600 * 24 * 7)
Parameters

seed – seed for random assignment

The function initializes and returns a slicing function, which deterministically assigns query to a slice based on the query name.

It utilizes the Public Suffix List to ensure domains under the same registrable domain end up in a single slice. (see example below)

seed can be used to re-shuffle the slicing algorhitm when the slicing function is initialized. By default, the assigment is re-shuffled after one week (when resolver restart / reloads config). To force a stable distribution, pass a fixed value. To re-shuffle on every resolver restart, use os.time().

The following example demonstrates a distribution among 3 slices:

slice 1/3:
example.com
a.example.com
b.example.com
x.b.example.com
example3.com

slice 2/3:
example2.co.uk

slice 3/3:
example.co.uk
a.example.co.uk

These two functions can be used together to forward queries for names in different parts of DNS name space to different target servers:

policy.add(policy.slice(
    policy.slice_randomize_psl(),
    policy.TLS_FORWARD({{'192.0.2.1', hostname='res.example.com'}}),
    policy.TLS_FORWARD({
        -- multiple servers can be specified for a single slice
        -- the one with lowest round-trip time will be used
        {'193.17.47.1', hostname='odvr.nic.cz'},
        {'185.43.135.1', hostname='odvr.nic.cz'},
    })
))

Note

The privacy implications of using this feature aren’t clear. Since websites often make requests to multiple domains, these might be forwarded to different targets. This could result in decreased privacy (e.g. when the remote targets are both logging or otherwise processing your DNS traffic). The intended use-case is to use this feature with semi-trusted resolvers which claim to do no logging (such as those listed on dnsprivacy.org), to decrease the potential exposure of your DNS data to a malicious resolver operator.

Replacing part of the DNS tree

Following procedure applies only to domains which have different content publicly and internally. For example this applies to “your own” top-level domain example. which does not exist in the public (global) DNS namespace.

Dealing with these internal-only domains requires extra configuration because DNS was designed as “single namespace” and local modifications like adding your own TLD break this assumption.

Warning

Use of internal names which are not delegated from the public DNS is causing technical problems with caching and DNSSEC validation and generally makes DNS operation more costly. We recommend against using these non-delegated names.

To make such internal domain available in your resolver it is necessary to graft your domain onto the public DNS namespace, but grafting creates new issues:

These grafted domains will be rejected by DNSSEC validation because such domains are technically indistinguishable from an spoofing attack against the public DNS. Therefore, if you trust the remote resolver which hosts the internal-only domain, and you trust your link to it, you need to use the policy.STUB() policy instead of policy.FORWARD() to disable DNSSEC validation for those grafted domains.

Secondly, after disabling DNSSEC validation you have to solve another issue caused by grafting. For example, if you grafted your own top-level domain example. onto the public DNS namespace, at some point the root server might send proof-of-nonexistence proving e.g. that there are no other top-level domain in between names events. and exchange., effectivelly proving non-existence of example..

These proofs-of-nonexistence protect public DNS from spoofing but break grafted domains because proofs will be latter used by resolver (when the positive records for the grafted domain timeout from cache), effectivelly making grafted domain unavailable. The easiest work-around is to disable reading from cache for grafted domains.

Example configuration grafting domains onto public DNS namespace
extraTrees = policy.todnames(
    {'faketldtest.',
     'sld.example.',
     'internal.example.com.',
     '2.0.192.in-addr.arpa.'  -- this applies to reverse DNS tree as well
     })
-- Beware: the rule order is important, as policy.STUB is not a chain action.
-- Flags: for "dumb" targets disabling EDNS can help (below) as DNSSEC isn't
-- validated anyway; in some of those cases adding 'NO_0X20' can also help,
-- though it also lowers defenses against off-path attacks on communication
-- between the two servers.
policy.add(policy.suffix(policy.FLAGS({'NO_CACHE', 'NO_EDNS'}), extraTrees))
policy.add(policy.suffix(policy.STUB({'2001:db8::1'}), extraTrees))

Response policy zones

Warning

There is no published Internet Standard for RPZ and implementations vary. At the moment Knot Resolver supports limited subset of RPZ format and deviates from implementation in BIND. Nevertheless it is good enough for blocking large lists of spam or advertising domains.

The RPZ file format is basically a DNS zone file with very special semantics. For example:

; left hand side          ; TTL and class  ; right hand side
; encodes RPZ trigger     ; ignored        ; encodes action
; (i.e. filter)
blocked.domain.example    600 IN           CNAME .           ; block main domain
*.blocked.domain.example  600 IN           CNAME .           ; block subdomains

The only “trigger” supported in Knot Resolver is query name, i.e. left hand side must be a domain name which triggers the action specified on the right hand side.

Subset of possible RPZ actions is supported, namely:

RPZ Right Hand Side

Knot Resolver Action

BIND Compatibility

.

action is used

compatible if action is policy.DENY

*.

policy.ANSWER()

yes

rpz-passthru.

policy.PASS

yes

rpz-tcp-only.

policy.TC

yes

rpz-drop.

policy.DROP

no 1

fake A/AAAA

policy.ANSWER()

yes

fake CNAME

not supported

no

1

Our policy.DROP returns SERVFAIL answer (for historical reasons).

policy.rpz(action, path[, watch = true])
Parameters
  • action – the default action for match in the zone; typically you want policy.DENY

  • path – path to zone file

  • watch – boolean, if true, the file will be reloaded on file change

Enforce RPZ rules. This can be used in conjunction with published blocklist feeds. The RPZ operation is well described in this Jan-Piet Mens’s post, or the Pro DNS and BIND book.

For example, we can store the example snippet with domain blocked.domain.example (above) into file /etc/knot-resolver/blocklist.rpz and configure resolver to answer with NXDOMAIN plus the specified additional text to queries for this domain:

policy.add(
    policy.rpz(policy.DENY_MSG('domain blocked by your resolver operator'),
               '/etc/knot-resolver/blocklist.rpz',
               true))

Resolver will reload RPZ file at run-time if the RPZ file changes. Recommended RPZ update procedure is to store new blocklist in a new file (newblocklist.rpz) and then rename the new file to the original file name (blocklist.rpz). This avoids problems where resolver might attempt to re-read an incomplete file.

Additional properties

Most properties (actions, filters) are described above.

policy.add(rule, postrule)
Parameters
  • rule – added rule, i.e. policy.pattern(policy.DENY, '[0-9]+\2cz')

  • postrule – boolean, if true the rule will be evaluated on answer instead of query

Returns

rule description

Add a new policy rule that is executed either or queries or answers, depending on the postrule parameter. You can then use the returned rule description to get information and unique identifier for the rule, as well as match count.

-- mirror all queriesm, keep handle so we can retrieve information later
local rule = policy.add(policy.all(policy.MIRROR('127.0.0.2')))
-- we can print statistics about this rule any time later
print(string.format('id: %d, matched queries: %d', rule.id, rule.count)
policy.del(id)
Parameters

id – identifier of a given rule returned by policy.add()

Returns

boolean true if rule was deleted, false otherwise

Remove a rule from policy list.

policy.todnames({name, ...})
Param

names table of domain names in textual format

Returns table of domain names in wire format converted from strings.

-- Convert single name
assert(todname('example.com') == '\7example\3com\0')
-- Convert table of names
policy.todnames({'example.com', 'me.cz'})
{ '\7example\3com\0', '\2me\2cz\0' }