RFC 8764

Independent Submission                                       S. Cheshire
Request for Comments: 8764                                   M. Krochmal
Category: Informational                                       Apple Inc.
ISSN: 2070-1721                                                June 2020

                Apple's DNS Long-Lived Queries Protocol


   Apple's DNS Long-Lived Queries (LLQ) is a mechanism for extending the
   DNS protocol to support change notification, thus allowing clients to
   learn about changes to DNS data without polling the server.  From
   2005 onwards, LLQ was implemented in Apple products including Mac OS
   X, Bonjour for Windows, and AirPort wireless base stations.  In 2020,
   the LLQ protocol was superseded by the IETF Standards Track RFC 8765,
   "DNS Push Notifications", which builds on experience gained with the
   LLQ protocol to create a superior replacement.

   The existing LLQ protocol deployed and used from 2005 to 2020 is
   documented here to give background regarding the operational
   experience that informed the development of DNS Push Notifications,
   and to help facilitate a smooth transition from LLQ to DNS Push

Status of This Memo

   This document is not an Internet Standards Track specification; it is
   published for informational purposes.

   This is a contribution to the RFC Series, independently of any other
   RFC stream.  The RFC Editor has chosen to publish this document at
   its discretion and makes no statement about its value for
   implementation or deployment.  Documents approved for publication by
   the RFC Editor are not candidates for any level of Internet Standard;
   see Section 2 of RFC 7841.

   Information about the current status of this document, any errata,
   and how to provide feedback on it may be obtained at

Copyright Notice

   Copyright (c) 2020 IETF Trust and the persons identified as the
   document authors.  All rights reserved.

   This document is subject to BCP 78 and the IETF Trust's Legal
   Provisions Relating to IETF Documents
   (https://trustee.ietf.org/license-info) in effect on the date of
   publication of this document.  Please review these documents
   carefully, as they describe your rights and restrictions with respect
   to this document.

Table of Contents

   1.  Introduction
     1.1.  Transition to DNS Push Notifications
   2.  Conventions and Terminology Used in This Document
   3.  Mechanisms
     3.1.  Assigned Numbers
     3.2.  Opt-RR Format
   4.  LLQ Address and Port Identification
   5.  LLQ Setup
     5.1.  Setup Message Retransmission
     5.2.  LLQ Setup Four-Way Handshake
       5.2.1.  Setup Request
       5.2.2.  Setup Challenge
       5.2.3.  Challenge Response
       5.2.4.  ACK + Answers
     5.3.  Resource Record TTLs
   6.  Event Responses
     6.1.  Add Events
     6.2.  Remove Events
     6.3.  Gratuitous Response Acknowledgments
   7.  LLQ Lease-Life Expiration
     7.1.  Refresh Request
     7.2.  LLQ Refresh Acknowledgment
   8.  Security Considerations
     8.1.  Server DoS
     8.2.  Client Packet Storms
     8.3.  Spoofing
   9.  IANA Considerations
   10. References
     10.1.  Normative References
     10.2.  Informative References
   Appendix A.  Problems with the LLQ Protocol

   Authors' Addresses

1.  Introduction

   In dynamic environments, DNS-based Service Discovery [RFC6763]
   benefits significantly from clients being able to learn about changes
   to DNS information via a mechanism that is both more timely and more
   efficient than simple polling.  Such a mechanism enables "live
   browses" that (a) learn when a new instance of a service appears, (b)
   learn when an existing service instance disappears from the network,
   and (c) allows clients to monitor status changes to a service
   instance (e.g., printer ink levels).  Multicast DNS [RFC6762]
   supports this natively.  When a device on the network publishes or
   deletes Multicast DNS records, these changes are multicast to other
   hosts on the network.  Those hosts deliver the change notifications
   to interested clients (applications running on that host).  Hosts
   also send occasional queries to the network, in case gratuitous
   announcements are not received due to packet loss, and to detect
   records lost due to their publishers crashing or having become
   disconnected from the network.

   This document defines an Apple extension to unicast DNS that enables
   a client to issue long-lived queries that allow a unicast DNS server
   to notify clients about changes to DNS data.  This is a more scalable
   and practical solution than can be achieved by polling of the name
   server, because a low polling rate could leave the client with stale
   information, while a high polling rate would have an adverse impact
   on the network and server.

   The mechanism defined in this document is now being replaced by DNS
   Push Notifications [RFC8765] as explained in Section 1.1.

1.1.  Transition to DNS Push Notifications

   The LLQ protocol enjoyed over a decade of useful operation, enabling
   timely live updates for the service discovery user interface in
   Apple's Back to My Mac [RFC6281] service.

   However, some problems were discovered, as described in Appendix A.
   This operational experience with LLQ informed the design of its IETF
   Standards Track successor, DNS Push Notifications [RFC8765].  Since
   no further work is being done on the LLQ protocol, this LLQ
   specification will not be updated to remedy these problems.

   All existing LLQ implementations are RECOMMENDED to migrate to using
   DNS Push Notifications instead.

   Existing LLQ servers are RECOMMENDED to implement and support DNS
   Push Notifications so that clients can begin migrating to the newer

   Existing LLQ clients are RECOMMENDED to query for the
   "_dns-push-tls._tcp.<zone>" SRV record first, and then only if DNS
   Push Notifications fail, fall back to query for
   "_dns-llq._udp.<zone>" instead.  Use of the "_dns-llq._udp.<zone>"
   SRV record is described in Section 4.

   This will cause clients to prefer the newer protocol when possible.
   It is RECOMMENDED that clients always attempt DNS Push Notifications
   first for every new request, and only if that fails, then fall back
   to using LLQ.  Clients SHOULD NOT record that a given server only
   speaks LLQ and subsequently default to LLQ for that server, since
   server software gets updated and even a server that speaks only LLQ
   today may be updated to support DNS Push Notifications tomorrow.

   New client and server implementations are RECOMMENDED to support only
   DNS Push Notifications.

2.  Conventions and Terminology Used in This Document

   The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
   "OPTIONAL" in this document are to be interpreted as described in
   BCP 14 [RFC2119] [RFC8174] when, and only when, they appear in all
   capitals, as shown here.

3.  Mechanisms

   DNS Long-Lived Queries (LLQ) are implemented using the standard DNS
   message format [RFC1035] in conjunction with an EDNS(0) OPT pseudo-RR
   [RFC6891] with a new OPTION-CODE and OPTION-DATA format specified
   here.  Encoding the LLQ request in an OPT pseudo-RR allows for
   implementation of LLQ with minimal modification to a name server's
   front end.  If a DNS query containing an LLQ option is sent to a
   server that does not implement LLQ, a server that complies with the
   EDNS(0) specification [RFC6891] will silently ignore the unrecognized
   option and answer the request as a normal DNS query without
   establishing any long-lived state and without returning an LLQ option
   in its response.  If a DNS query containing an LLQ option is sent to
   a server that does not implement EDNS(0) at all, the server may
   silently ignore the EDNS(0) OPT pseudo-RR or it may return a nonzero
   RCODE.  However, in practice, this issue is mostly theoretical, since
   having a zone's _dns-llq._udp.<zone> SRV record target a host that
   does not implement LLQ is a configuration error.

   Note that this protocol is designed for data set sizes of a few dozen
   resource records at most and change rates no more than once every 10
   seconds on average.  Data sets that frequently exceed a single IP
   packet, or that experience a rapid change rate, may have undesirable
   performance implications.

3.1.  Assigned Numbers

   This section describes constants used in this document.

      EDNS(0) OPTION-CODE (recorded with IANA):

         LLQ               1

      LLQ-PORT 5352 (recorded with IANA)

      LLQ Opcodes (specific to this LLQ EDNS(0) Option):

         LLQ-SETUP         1
         LLQ-REFRESH       2
         LLQ-EVENT         3

      LLQ Error Codes (specific to this LLQ EDNS(0) Option):

         NO-ERROR          0
         SERV-FULL         1
         STATIC            2
         FORMAT-ERR        3
         NO-SUCH-LLQ       4
         BAD-VERS          5
         UNKNOWN-ERR       6

3.2.  Opt-RR Format

   As required by the EDNS(0) specification [RFC6891], all OPT
   pseudo-RRs used in LLQs are formatted as follows:

          | Field Name | Field Type   | Description             |
          | NAME       | domain name  | MUST be 0 (root domain) |
          | TYPE       | u_int16_t    | OPT (41)                |
          | CLASS      | u_int16_t    | 0*                      |
          | TTL        | u_int32_t    | 0                       |
          | RDLEN      | u_int16_t    | length of all RDATA     |
          | RDATA      | octet stream | (see below)             |

                       Table 1: OPT-RRs Used in LLQs

   * The CLASS field indicates, as per the EDNS(0) specification
   [RFC6891], the sender's UDP payload size.  However, clients and
   servers are not required to determine their reassembly buffer size,
   path MTU, etc., to support LLQ.  Thus, the sender of an LLQ Request
   or Response MAY set the CLASS field to 0.  The recipient MUST ignore
   the class field if it is set to 0.

   The RDATA of an EDNS(0) OPT pseudo-RR consists of zero or more
   options of the form { OPTION-CODE, OPTION-LENGTH, OPTION-DATA }
   packed together, with the RDLEN field set accordingly to indicate the
   total size.  An LLQ OPTION is illustrated below.  An EDNS(0) OPT
   pseudo-RR may contain zero or more LLQ OPTIONS in addition to zero or
   more other EDNS(0) options.

   | Field Name    | Field Type | Description                         |
   | OPTION-CODE   | u_int16_t  | LLQ (1)                             |
   | OPTION-LENGTH | u_int16_t  | Length of following fields (18)     |
   | LLQ-VERSION   | u_int16_t  | Version of LLQ protocol implemented |
   | LLQ-OPCODE    | u_int16_t  | Identifies LLQ operation            |
   | LLQ-ERROR     | u_int16_t  | Identifies LLQ errors               |
   | LLQ-ID        | u_int64_t  | Identifier for an LLQ               |
   | LLQ-LEASE     | u_int32_t  | Requested or granted life of LLQ,   |
   |               |            | in seconds                          |

                           Table 2: LLQ OPTION

   The size and meaning of the OPTION-CODE and OPTION-LENGTH fields are
   as described in the EDNS(0) specification [RFC6891].  The remainder
   of the fields comprise the OPTION-DATA of the EDNS(0) LLQ OPTION.
   Since for LLQ the OPTION-DATA is a fixed size, in EDNS(0) LLQ OPTIONS
   the OPTION-LENGTH field always has the value 18.

   In keeping with Internet convention, all multi-byte numeric
   quantities (u_int16_t, u_int32_t, and u_int64_t) are represented in
   big endian byte order (most significant byte first).

4.  LLQ Address and Port Identification

   The client requires a mechanism to determine to which server it
   should send LLQ operations.

   Additionally, some firewalls block direct communication with a name
   server on port 53 to avoid spoof responses.  However, this direct
   communication is necessary for LLQs.  Thus, servers MAY listen for
   LLQs on a different port (typically 5352).  Clients, therefore, also
   need a mechanism to determine to which port to send LLQ operations.

   The client determines the server responsible for a given LLQ much as
   a client determines to which server to send a DNS dynamic update.
   The client begins by sending a standard DNS query for the name of the
   LLQ, with type SOA.  If the record exists, then the server MUST
   answer with that SOA record in the Answer section.  If a record of
   type SOA with the LLQ name does not exist, then the server SHOULD
   include an SOA record for that name's zone in the Authority section.
   For example, a query for "_ftp._tcp.example.com" with type SOA, when
   there is no SOA record with that name, might return an SOA record
   named "example.com" in the Authority section.  If the named SOA
   record does not exist and the server fails to include the enclosing
   SOA record in the Authority section, the client strips the leading
   label from the name and tries again, repeating until an answer is

   This iterative zone apex discovery algorithm is described in more
   detail in the DNS Push Notifications specification [RFC8765].

   Upon learning the zone apex (SOA), the client then constructs and
   sends an SRV query for the name, "_dns-llq._udp.<zone>",
   e.g., "_dns-llq._udp.example.com".

   An authoritative server for a zone implementing LLQ MUST answer with
   an SRV record [RFC2782] for this name.  The SRV RDATA is as follows:

    | PRIORITY | typically 0                                          |
    | WEIGHT   | typically 0                                          |
    | PORT     | typically 53 or 5352                                 |
    | TARGET   | name of server providing LLQs for the requested zone |

                             Table 3: SRV RDATA

   The server SHOULD include the address record(s) for the target host
   in the Additional section of the response.

   If the server does not include the target host's address record(s) in
   the Additional section, the client SHOULD query explicitly for the
   address record(s) with the name of the SRV target.

   The client MUST send all LLQ requests, refreshes, and acknowledgments
   to the name server specified in the SRV target, at the address
   contained in the address record for that target.  Note that the
   queries described in this section (including those for SOA and SRV
   records) MAY be sent to an intermediate DNS recursive resolver --
   they need not be sent directly to the name server.

   If, on issuing the SRV query, the client receives a negative response
   indicating that the SRV record does not exist, the client SHOULD
   conclude that the zone does not support LLQ.  The client then SHOULD
send an LLQ request for the desired name, instead utilizing the
   behavior for LLQ-unaware servers described in Section 5, "LLQ Setup".

   Servers should send all messages to the source address and port of
   the LLQ setup message received from the client.

5.  LLQ Setup

   An LLQ is initiated by a client and is completed via a four-way
   handshake.  This handshake provides resilience to packet loss,
   demonstrates client reachability, and reduces denial-of-service
   attack opportunities (see Section 8, "Security Considerations").

5.1.  Setup Message Retransmission

   LLQ Setup Requests and Responses sent by the client SHOULD be
   retransmitted if no acknowledgments are received.  The client SHOULD
   retry up to two more times (for a total of 3 attempts) before
   considering the server down or unreachable.  The client MUST wait at
   least 2 seconds before the first retransmission and 4 seconds between
   the first and second retransmissions.  The client SHOULD listen for a
   response for at least 8 seconds after the 3rd attempt before
   considering the server down or unreachable.  Upon determining a
   server to be down, a client MAY periodically attempt to re-initiate
   an LLQ setup at a rate of not more than once per hour.

   Servers MUST NOT retransmit acknowledgments that do not generate
   responses from the client.  Retransmission in setup is client driven,
   freeing servers from maintaining timers for incomplete LLQ setups.
   If servers receive duplicate messages from clients (perhaps due to
   the loss of the server's responses mid-flight), the server MUST
   resend its reply (possibly modifying the LLQ-LEASE as described in
   Section 5.2.4, "ACK + Answers").

   Servers MUST NOT garbage collect LLQs that fail to complete the four-
   way handshake until the initially granted LLQ-LEASE has elapsed.

5.2.  LLQ Setup Four-Way Handshake

   The four phases of the handshake include:

      1) Setup Request        client to server, identifies LLQ(s)

      2) Setup Challenge      server to client, provides unique
                              identifiers for successful requested LLQs,
                              and error(s) for unsuccessful requested

      3) Challenge Response   client to server, echoes identifier(s),
                              demonstrating client's reachability and
                              willingness to participate

      4) ACK + Answers        server to client, confirms setup and
                              provides initial answers

5.2.1.  Setup Request

   A request for an LLQ is formatted like a standard DNS query but with
   an OPT pseudo-RR containing LLQ metadata in its Additional section.
   LLQ Setup Requests are identified by the LLQ-SETUP opcode and a
   zero-valued LLQ-ID.

   The request MAY contain multiple questions to set up multiple LLQs.
   A Setup Request consisting of multiple questions MUST contain
   multiple LLQ OPTIONS, one per question, with the LLQ OPTIONS in the
   same order as the questions they correspond to (i.e., the first LLQ
   OPTION corresponds to the first question, the second LLQ OPTION
   corresponds to the second question, etc.).  If requesting multiple
   LLQs, clients SHOULD request the same LLQ-LEASE for each LLQ.
   Requests over UDP MUST NOT contain multiple questions if doing so
   would cause the message to exceed a single IP packet.

   A client MUST NOT request multiple identical LLQs (i.e., containing
   the same qname/type/class) from the same source IP address and port.
   This requirement is to avoid unnecessary load on servers.  In the
   case of multiple independent client implementations that may run on
   the same device without knowledge of each other, it is allowable if
   they by chance send LLQ requests for the same qname/type/class.
   These independent implementations on the same client will be using
   different source ports.  Likewise, to the server, multiple
   independent clients behind the same NAT gateway will appear as if
   they were multiple independent clients using different ports on the
   same host, and this is also allowable.

   The query MUST NOT be for record type ANY (255), class ANY (255), or
   class NONE (0).

     | Field Name    | Field Type | Description                     |
     | OPTION-CODE   | u_int16_t  | LLQ (1)                         |
     | OPTION-LENGTH | u_int16_t  | Length of following fields (18) |
     | LLQ-VERSION   | u_int16_t  | Version of LLQ protocol         |
     |               |            | implemented by requester (1)    |
     | LLQ-OPCODE    | u_int16_t  | LLQ-SETUP (1)                   |
     | LLQ-ERROR     | u_int16_t  | NO-ERROR (0)                    |
     | LLQ-ID        | u_int64_t  | 0                               |
     | LLQ-LEASE     | u_int32_t  | Desired life of LLQ request     |

                 Table 4: Setup Request LLQ OPTION Format

   The Setup Request LLQ OPTION MUST be repeated once for each
   additional query in the Question section.

5.2.2.  Setup Challenge

   Upon receiving an LLQ Setup Request, a server implementing LLQs will
   send a Setup Challenge to the requester (client).  An LLQ Setup
   Challenge is a DNS response, with the DNS message ID matching that of
   the Setup Request, and with all questions contained in the Setup
   Request present in the Question section of the response.
   Additionally, the challenge contains a single OPT pseudo-RR with an
   LLQ OPTION for each LLQ request, indicating the success or failure of
   each request.  The LLQ OPTIONS MUST be in the same order as the
   questions they correspond to.  Note that in a Setup Request
   containing multiple questions, some LLQs may succeed while others may

     | Field Name    | Field Type | Description                     |
     | OPTION-CODE   | u_int16_t  | LLQ (1)                         |
     | OPTION-LENGTH | u_int16_t  | Length of following fields (18) |
     | LLQ-VERSION   | u_int16_t  | Version of LLQ protocol         |
     |               |            | implemented in server (1)       |
     | LLQ-OPCODE    | u_int16_t  | LLQ-SETUP (1)                   |
     | LLQ-ERROR     | u_int16_t  | [As Appropriate]                |
     | LLQ-ID        | u_int64_t  | [As Appropriate]                |
     | LLQ-LEASE     | u_int32_t  | [As Appropriate]                |

                Table 5: Setup Challenge LLQ OPTION Format

   The Setup Challenge LLQ OPTION MUST be repeated once for each query
   in the Questions section of the Setup Challenge.  Further details for
   LLQ-ERROR, LLQ-ID and LLQ-LEASE are given below.


      NO-ERROR:         The LLQ Setup Request was successful.

      FORMAT-ERR:       The LLQ was improperly formatted.  Note that if
                        the rest of the DNS message is properly
                        formatted, the DNS header error code MUST NOT
                        include a format error code, since to do so
                        would cause ambiguity between the case where a
                        client sends a valid LLQ Setup Request to a
                        server that does not understand LLQ and the case
                        where a client sends a malformed LLQ Setup
                        Request to a server that does understand LLQ.

      SERV-FULL:        The server cannot grant the LLQ request because
                        it is overloaded or the request exceeds the
                        server's rate limit (see Section 8, Security
                        Considerations).  Upon returning this error, the
                        server MUST include in the LLQ-LEASE field a
                        time interval, in seconds, after which the
                        client may retry the LLQ Setup.

      STATIC:           The data for this name and type is not expected
                        to change frequently, and the server, therefore,
                        does not support the requested LLQ.  The client
                        MUST honor the resource record TTLs returned and
                        MUST NOT poll sooner than indicated by those
                        TTLs, nor should it retry the LLQ Setup for this
                        name and type.

      BAD-VERS:         The protocol version specified in the client's
                        Setup Request is not supported by the server.

      UNKNOWN-ERR:      The LLQ was not granted for some other reason
                        not covered by the preceding error code values.

   LLQ-ID:           On success, a random number generated by the server
                     that is unique on the server for the requested
                     name/type/class.  The LLQ-ID SHOULD be an
                     unpredictable random number.  A possible method of
                     allocating LLQ-IDs with minimal bookkeeping would
                     be to store the time, in seconds since the Epoch,
                     in the high 32 bits of the field, and a
                     cryptographically generated 32-bit random integer
                     in the low 32 bits.

                     On error, the LLQ-ID is set to 0.

   LLQ-LEASE:        On success, the actual life of the LLQ, in seconds.
                     Value may be greater than, less than, or equal to
                     the value requested by the client, as per the
                     server administrator's policy.  The server MAY
                     discard the LLQ after this LLQ-LEASE expires unless
                     the LLQ has been renewed by the client (see
                     Section 7, "LLQ Lease-Life Expiration").  The
                     server MUST NOT generate events (see Section 6,
                     "Event Responses") for expired LLQs.

                     On SERV-FULL error, LLQ-LEASE MUST be set to a time
                     interval, in seconds, after which the client may
                     retry the LLQ Setup.

                     On other errors, the LLQ-LEASE MUST be set to 0.

5.2.3.  Challenge Response

   Upon issuing a Setup Request, a client listens for a Setup Challenge
   (Section 5.2.2) retransmitting the Setup Request as necessary
   (Section 5.1).  After receiving a successful Setup Challenge, the
   client SHOULD send a Challenge Response to the server.  This
   Challenge Response is a DNS request with questions as in the Setup
   Request and Setup Challenge, and a single OPT pseudo-RR in the
   Additional section, with the LLQ OPTIONS corresponding to the LLQ
   OPTIONS contained in the Setup Challenge (i.e., echoing, for each LLQ
   OPTION, the random LLQ-ID and the granted LLQ-LEASE).  If the
   Challenge Response contains multiple questions, the first question
   MUST correspond to the first LLQ OPTION, etc.

   If the Setup Request for a particular LLQ fails with a STATIC error,
   the client MUST NOT poll the server for that LLQ.  The client SHOULD
   honor the resource record TTLs contained in the response.

   If a Setup Request fails with a SERV-FULL error, the client MAY retry
   the LLQ Setup Request (Section 5.2.1) after the time indicated in the
   LLQ-LEASE field.

   If the Setup Request fails with an error other than STATIC or
   SERV-FULL, or the server is determined not to support LLQ (i.e., the
   client receives a DNS response with a nonzero RCODE, or a DNS
   response containing no LLQ option), the client MAY poll the server
   periodically with standard DNS queries, inferring Add and Remove
   Events (see Section 6, "Event Responses") by comparing answers to
   these queries.  The client SHOULD NOT poll more than once every 15
   minutes for a given query.  The client MUST NOT poll if it receives a
   STATIC error code in the acknowledgment.

5.2.4.  ACK + Answers

   Upon receiving a correct Challenge Response, a server MUST return an
   acknowledgment, completing the LLQ setup, and provide all current
   answers to the question(s).

   To acknowledge a successful Challenge Response, i.e., a Challenge
   Response in which the LLQ-ID and LLQ-LEASE echoed by the client match
   the values issued by the server, the server MUST send a DNS response
   containing all available answers to the question(s) contained in the
   original Setup Request, along with all additional resource records
   appropriate for those answers in the Additional section.  The
   Additional section also contains LLQ OPTIONS formatted as follows:

     | Field Name    | Field Type | Description                     |
     | OPTION-CODE   | u_int16_t  | LLQ (1)                         |
     | OPTION-LENGTH | u_int16_t  | Length of following fields (18) |
     | LLQ-VERSION   | u_int16_t  | Version of LLQ protocol         |
     |               |            | implemented in server (1)       |
     | LLQ-OPCODE    | u_int16_t  | LLQ-SETUP (1)                   |
     | LLQ-ERROR     | u_int16_t  | NO-ERROR (0)                    |
     | LLQ-ID        | u_int64_t  | Originally granted ID, echoed   |
     |               |            | in client's Response            |
     | LLQ-LEASE     | u_int32_t  | Remaining life of LLQ, in       |
     |               |            | seconds                         |

           Table 6: Successful ACK + Answers LLQ OPTION Format

   If there is a significant delay in receiving a Challenge Response, or
   multiple Challenge Responses are issued (possibly because they were
   lost en route to the client, causing the client to resend the
   Challenge Response), the server MAY decrement the LLQ-LEASE by the
   time elapsed since the Setup Challenge was initially issued.

   If the setup is completed over UDP and all initially available
   answers to the question(s), additional records, and the OPT pseudo-RR
   do not fit in a single IP packet, some or all additional records
   (excluding the OPT pseudo-RR) MUST be omitted.  If, after omission of
   all additional records, the answers still do not fit in a single
   message, answers MUST be removed until the message fits in a single
   IP packet.  These answers not delivered in the ACK + Answers MUST be
   delivered without undue delay to the client via Add Events
   (Section 6, "Event Responses").

5.3.  Resource Record TTLs

   The TTLs of resource records contained in answers to successful LLQs
   SHOULD be ignored by the client.  The client MAY cache LLQ answers
   until the client receives a gratuitous announcement (see Section 6,
   "Event Responses") indicating that the answer to the LLQ has changed.
   The client SHOULD NOT cache answers after the LLQs LLQ-LEASE expires
   without being refreshed (see Section 7, "LLQ Lease-Life Expiration").
   If an LLQ request fails, the client SHOULD NOT cache answers for a
   period longer than the client's polling interval.

   Note that resource records intended specifically to be transmitted
   via LLQs (e.g., DNS-based Service Discovery resource records) may
   have unusually short TTLs.  This is because it is assumed that the
   records may change frequently, and that a client's cache coherence
   will be maintained via the LLQ and gratuitous responses.  Short TTLs
   prevent stale information from residing in intermediate DNS recursive
   resolvers that are not LLQ aware.

   TTLs of resource records included in the Additional section of an LLQ
   response (which do not directly answer the LLQ) SHOULD be honored by
   the client.

6.  Event Responses

   When a change ("event") occurs to a name server's zone, the server
   MUST check if the new or deleted resource records answer any LLQs.
   If so, the changes MUST be communicated to the LLQ requesters in the
   form of a gratuitous DNS response sent to the client, with the
   relevant question(s) in the Question section, and the corresponding
   answers in the Answer section.  The response also includes an OPT
   pseudo-RR in the Additional section.  This OPT pseudo-RR contains, in
   its RDATA, an LLQ OPTION for each LLQ being answered in the message.
   Each LLQ OPTION must include the LLQ-ID.  This reduces the potential
   for spoof events being sent to a client.

     | Field Name    | Field Type | Description                     |
     | OPTION-CODE   | u_int16_t  | LLQ (1)                         |
     | OPTION-LENGTH | u_int16_t  | Length of following fields (18) |
     | LLQ-VERSION   | u_int16_t  | Version of LLQ protocol         |
     |               |            | implemented in server (1)       |
     | LLQ-OPCODE    | u_int16_t  | LLQ-EVENT (3)                   |
     | LLQ-ERROR     | u_int16_t  | NO-ERROR (0)                    |
     | LLQ-ID        | u_int64_t  | [As Appropriate]                |
     | LLQ-LEASE     | u_int32_t  | 0                               |

                Table 7: Event Response LLQ OPTION Format

   Gratuitous responses for a single LLQ MAY be batched such that
   multiple changes are communicated in a single message.  Responses
   MUST NOT be batched if this would cause a message that would
   otherwise fit in a single IP packet to be truncated.  While responses
   MAY be deferred to provide opportunities for batching, responses
   SHOULD NOT be delayed, for purposes of batching, for more than 30
   seconds, as this would cause an unacceptable latency for the client.

   After sending a gratuitous response, the server MUST listen for an
   acknowledgment from the client.  If the client does not respond, the
   server MUST resend the response.  The server MUST resend two times
   (for a total of 3 transmissions), after which the server MUST
   consider the client to be unreachable and delete its LLQ.  The server
   MUST listen for 2 seconds before resending the response, 4 more
   seconds before resending again, and must wait an additional 8 seconds
   after the 3rd transmission before terminating the LLQ.

   The DNS message header of the response SHOULD include an
   unpredictable random number in the DNS message ID field, which is to
   be echoed in the client's acknowledgment.

6.1.  Add Events

   Add Events occur when a new resource record appears, usually as the
   result of a dynamic update [RFC2136], that answers an LLQ.  This
   record must be sent in the Answer section of the event to the client.
   Records that normally accompany this record in responses MAY be
   included in the Additional section as per truncation restrictions
   described above.

6.2.  Remove Events

   Remove Events occur when a resource record previously sent to a
   client, either in an initial response or in an Add Event, becomes
   invalid (normally as a result of being removed via a dynamic update).
   The deleted resource record is sent in the Answer section of the
   event to the client.  The resource record TTL is set to -1,
   indicating that the record has been removed.

6.3.  Gratuitous Response Acknowledgments

   Upon receiving a gratuitous response ("event"), the client MUST send
   an acknowledgment to the server.  This acknowledgment is a DNS
   response echoing the OPT pseudo-RR contained in the event, with the
   message ID of the gratuitous response echoed in the message header.
   The acknowledgment MUST be sent to the source IP address and port
   from which the event originated.

7.  LLQ Lease-Life Expiration

7.1.  Refresh Request

   If the client desires to maintain the LLQ beyond the duration
   specified in the LLQ-LEASE field of the ACK + Answers
   (Section 5.2.4), the client MUST send a Refresh Request.  A Refresh
   Request is identical to an LLQ Challenge Response (Section 5.2.3) but
   with the LLQ-OPCODE set to LLQ-REFRESH.  Unlike a Challenge Response,
   a Refresh Request returns no answers.

   The client SHOULD refresh an LLQ when 80% of its LLQ-LEASE has

   As a means of reducing network traffic, when constructing refresh
   messages the client SHOULD include all LLQs established with a given
   server, even those not yet close to expiration.  However, at least
   one LLQ MUST have elapsed at least 80% of its original LLQ-LEASE.
   The client MUST NOT include additional LLQs if doing so would cause
   the message to no longer fit in a single IP packet.  In this case,
   the LLQs furthest from expiration should be omitted such that the
   message fits in a single IP packet.  (These LLQs SHOULD be refreshed
   in a separate message when 80% of one or more of their lease lives
   have elapsed.)  When refreshing multiple LLQs simultaneously, the
   message contains multiple questions and a single OPT pseudo-RR with
   multiple LLQ OPTIONS, one per question, with the LLQ OPTIONS in the
   same order as the questions they correspond to.

   The client SHOULD specify the original LLQ-LEASE granted in the LLQ
   response as the desired LLQ-LEASE in the Refresh Request.  If
   refreshing multiple LLQs simultaneously, the client SHOULD request
   the same LLQ-LEASE for all LLQs being refreshed (with the exception
   of termination requests; see below).

   To terminate an LLQ prior to its scheduled expiration (for instance,
   when the client terminates a DNS-based Service Discovery browse
   operation or when a client is about to go to sleep or shut down), the
   client specifies an LLQ-LEASE value of 0.

   The client MUST listen for an acknowledgment from the server.  The
   client MAY retry up to two more times (for a total of 3 attempts)
   before considering the server down or unreachable.  The client MUST
retry a first time before 90% of the LLQ-LEASE has expired and
   MUST NOT retry again before 95% of the LLQ-LEASE has expired.  If the
   server is determined to be down, the client MAY periodically attempt
   to re-establish the LLQ via an LLQ Setup Request message.  The client
   MUST NOT attempt the LLQ Setup Request more than once per hour.

7.2.  LLQ Refresh Acknowledgment

   Upon receiving an LLQ Refresh message, a server MUST send an
   acknowledgment of the Refresh.  This acknowledgment is formatted like
   the "ACK + Answers" message described in Section 5.2.4, but with the
   following variations:

   *  It contains no answers.

   *  The LLQ-OPCODE is set to LLQ-REFRESH.

   *  NO-SUCH-LLQ MUST be returned as an error code if the client
      attempts to refresh an expired or non-existent LLQ (as determined
      by the LLQ-ID in the request).

   *  The LLQ-ID in the acknowledgment is set to the LLQ-ID in the

8.  Security Considerations

   In datagram-based protocols (i.e., protocols running over UDP, or
   directly over IP, or similar), servers may be susceptible to denial-
   of-service (DoS) attacks, and clients may be subjected to packet
   storms.  Carefully designed mechanisms are needed to limit potential
   for these attacks.

   Note: This section contains no new protocol elements -- it serves
   only to explain the rationale behind protocol elements described
   above as they relate to security.

8.1.  Server DoS

   LLQs require that servers be stateful, maintaining entries for each
   LLQ over a potentially long period of time.  If unbounded in
   quantity, these entries may overload the server.  By returning
   SERV-FULL in Setup Challenges, the server may limit the maximum
   number of LLQs it maintains.  Additionally, the server may return
   SERV-FULL to limit the number of LLQs requested for a single name and
   type, or by a single client.  This throttling may be in the form of a
   hard limit, or, preferably, by token-bucket rate limiting.  Such rate
   limiting should occur rarely in normal use and is intended to prevent
   DoS attacks -- thus, it is not built into the protocol explicitly but
   is instead implemented at the discretion of an administrator via the
   SERV-FULL error and the LLQ-LEASE field to indicate a retry time to
   the client.

8.2.  Client Packet Storms

   In addition to protecting the server from DoS attacks, the LLQ
   protocol limits the ability of a malicious host to cause the server
   to flood a client with packets.  This is achieved via the four-way
   handshake upon setup, demonstrating reachability and willingness of
   the client to participate, and by requiring that gratuitous responses
   be ACK'd by the client.

   Additionally, rate limiting by LLQ client address, as described in
   Section 8.1, serves to limit the number of packets that can be
   delivered to an unsuspecting client.

8.3.  Spoofing

   A large random ID greatly reduces the risk of an off-path attacker
   sending spoof packets to the client (containing spoof events) or to
   the server (containing phony requests or refreshes).

9.  IANA Considerations

   The EDNS(0) OPTION CODE 1 has already been assigned for this DNS
   extension.  IANA has updated the record in the "DNS EDNS0 Option
   Codes (OPT)" registry from "On-hold" to "Optional" and has set the
   reference to this document.

   TCP and UDP ports 5352 have already been assigned for LLQ.  IANA has
   added a reference to this document.

10.  References

10.1.  Normative References

   [RFC1035]  Mockapetris, P., "Domain names - implementation and
              specification", STD 13, RFC 1035, DOI 10.17487/RFC1035,
              November 1987, <https://www.rfc-editor.org/info/rfc1035>.

   [RFC2119]  Bradner, S., "Key words for use in RFCs to Indicate
              Requirement Levels", BCP 14, RFC 2119,
              DOI 10.17487/RFC2119, March 1997,

   [RFC2782]  Gulbrandsen, A., Vixie, P., and L. Esibov, "A DNS RR for
              specifying the location of services (DNS SRV)", RFC 2782,
              DOI 10.17487/RFC2782, February 2000,

   [RFC6891]  Damas, J., Graff, M., and P. Vixie, "Extension Mechanisms
              for DNS (EDNS(0))", STD 75, RFC 6891,
              DOI 10.17487/RFC6891, April 2013,

   [RFC8174]  Leiba, B., "Ambiguity of Uppercase vs Lowercase in RFC
              2119 Key Words", BCP 14, RFC 8174, DOI 10.17487/RFC8174,
              May 2017, <https://www.rfc-editor.org/info/rfc8174>.

   [RFC8765]  Pusateri, T. and S. Cheshire, "DNS Push Notifications",
              RFC 8765, DOI 10.17487/RFC8765, June 2020,

10.2.  Informative References

   [RFC2136]  Vixie, P., Ed., Thomson, S., Rekhter, Y., and J. Bound,
              "Dynamic Updates in the Domain Name System (DNS UPDATE)",
              RFC 2136, DOI 10.17487/RFC2136, April 1997,

   [RFC4787]  Audet, F., Ed. and C. Jennings, "Network Address
              Translation (NAT) Behavioral Requirements for Unicast
              UDP", BCP 127, RFC 4787, DOI 10.17487/RFC4787, January
              2007, <https://www.rfc-editor.org/info/rfc4787>.

   [RFC4953]  Touch, J., "Defending TCP Against Spoofing Attacks",
              RFC 4953, DOI 10.17487/RFC4953, July 2007,

   [RFC5382]  Guha, S., Ed., Biswas, K., Ford, B., Sivakumar, S., and P.
              Srisuresh, "NAT Behavioral Requirements for TCP", BCP 142,
              RFC 5382, DOI 10.17487/RFC5382, October 2008,

   [RFC6281]  Cheshire, S., Zhu, Z., Wakikawa, R., and L. Zhang,
              "Understanding Apple's Back to My Mac (BTMM) Service",
              RFC 6281, DOI 10.17487/RFC6281, June 2011,

   [RFC6762]  Cheshire, S. and M. Krochmal, "Multicast DNS", RFC 6762,
              DOI 10.17487/RFC6762, February 2013,

   [RFC6763]  Cheshire, S. and M. Krochmal, "DNS-Based Service
              Discovery", RFC 6763, DOI 10.17487/RFC6763, February 2013,

   [RFC6886]  Cheshire, S. and M. Krochmal, "NAT Port Mapping Protocol
              (NAT-PMP)", RFC 6886, DOI 10.17487/RFC6886, April 2013,

   [RFC6887]  Wing, D., Ed., Cheshire, S., Boucadair, M., Penno, R., and
              P. Selkirk, "Port Control Protocol (PCP)", RFC 6887,
              DOI 10.17487/RFC6887, April 2013,

   [RFC7857]  Penno, R., Perreault, S., Boucadair, M., Ed., Sivakumar,
              S., and K. Naito, "Updates to Network Address Translation
              (NAT) Behavioral Requirements", BCP 127, RFC 7857,
              DOI 10.17487/RFC7857, April 2016,

   [RFC8490]  Bellis, R., Cheshire, S., Dickinson, J., Dickinson, S.,
              Lemon, T., and T. Pusateri, "DNS Stateful Operations",
              RFC 8490, DOI 10.17487/RFC8490, March 2019,

   [SYN]      Eddy, W., "Defenses Against TCP SYN Flooding Attacks",
              Volume 9, Number 4, The Internet Protocol Journal, Cisco
              Systems, December 2006,

Appendix A.  Problems with the LLQ Protocol

   In the course of using LLQ since 2005, some problems were discovered.
   Since no further work is being done on the LLQ protocol, this LLQ
   specification will not be updated to remedy these problems.

   LLQ's IETF Standards Track successor, "DNS Push Notifications"
   [RFC8765], does not suffer from these problems, so all existing LLQ
   implementations are RECOMMENDED to migrate to using DNS Push
   Notifications, and all new implementations are RECOMMENDED to
   implement DNS Push Notifications instead of LLQ.

   Known problems with LLQ are documented here as a cautionary tale
   about the challenges of building an application protocol directly
   using datagrams (like IP or UDP) without the benefit of a mature and
   thoroughly reviewed intervening transport layer (such as TCP or

   An LLQ "Setup Challenge" message from server to client is identical
   to an LLQ "ACK + Answers" message from server to client when there
   are no current answers for the query.  If there is packet loss,
   retransmission, and duplication in the network, then a duplicated
   "Setup Challenge" message arriving late at the client would look like
   an "ACK + Answers" message with no answers, causing the client to
   clear its cache of any records matching the query.

   Section 5.1 of this LLQ specification states, "Servers MUST NOT
   garbage collect LLQs that fail to complete the four-way handshake
   until the initially granted LLQ-LEASE has elapsed."  This is probably
   a mistake since it exposes LLQ servers to an easy resource-exhaustion
   denial-of-service attack.  LLQ's replacement, DNS Push Notifications
   [RFC8765], is built using DNS Stateful Operations [RFC8490], which
   uses TLS over TCP; a benefit of building on TCP is that there are
   already established industry best practices to guard against SYN
   flooding and similar attacks [SYN] [RFC4953].

   The attempts here to pack multiple questions into a single UDP/IP
   packet for efficiency are awkward and lead to error-prone programming
   to deal with cases where some requests in a packet succeed and other
   requests in the same packet fail.  Fully specifying the correct
   handling in all possible cases would be a lot of work to document, a
   lot of work to implement, and even more work to thoroughly test.  DNS
   Push Notifications [RFC8765] avoids this problem by using an
   underlying stream protocol (TLS/TCP) to deal with packing small
   multiple messages into larger IP packets for efficiency.

   In some cases, initial LLQ answers are delivered in the "ACK +
   Answers" message, and in other cases, such as when all the initial
   answers will not fit in a single IP packet, some of the initial
   answers are delivered in a subsequent "Add Event" message.  Having
   two different ways to accomplish the same thing increases the
   possibility for programming errors.  DNS Push Notifications [RFC8765]
   corrects this error by having only one single consistent way to
   deliver results.

   LLQ is built using UDP, and because UDP has no standardized way of
   indicating the start and end of a session, firewalls and NAT gateways
   tend to be fairly aggressive about recycling UDP mappings that they
   believe to be disused [RFC4787] [RFC5382] [RFC7857].  Using a high
   keepalive traffic rate to maintain firewall or NAT mapping state
   could remedy this but would largely defeat the purpose of using LLQ
   in the first place, which is to provide efficient change notification
   without wasteful polling.  Because of this, existing LLQ clients use
   the NAT Port Mapping Protocol (NAT-PMP) [RFC6886] and/or Port Control
   Protocol (PCP) [RFC6887] to establish longer port mapping lifetimes.
   This solves the problem but adds extra complexity and doesn't work
   with firewalls and NAT gateways that don't support NAT-PMP or PCP.
   By using TCP instead of UDP, the DNS Push Notifications protocol
   benefits from better longevity of sessions through firewalls and NAT
   gateways that don't support NAT-PMP or PCP.


   The concepts described in this document were originally explored,
   developed, and implemented with help from Chris Sharp and Roger

   Kiren Sekar made significant contributions to the first draft of this
   document and he wrote much of the code for the implementation of LLQ
   that shipped in Mac OS X 10.4 Tiger in April 2005.

   Thanks to Independent Stream Editor Adrian Farrel for his support and
   assistance in the publication of this RFC.

Authors' Addresses

   Stuart Cheshire
   Apple Inc.
   One Apple Park Way
   Cupertino, CA 95014
   United States of America

   Phone: +1 (408) 996-1010
   Email: cheshire@apple.com

   Marc Krochmal
   Apple Inc.
   One Apple Park Way
   Cupertino, CA 95014
   United States of America

   Phone: +1 (408) 996-1010
   Email: marc@apple.com