Network Working Group Alan DeKok INTERNET-DRAFT FreeRADIUS Category: Informational
<draft-ietf-radext-status-server-01.txt><draft-ietf-radext-status-server-02.txt> Expires: February 24, 2008 25 AugustMay 2, 2009 2 November 2008 Use of Status-Server Packets in the Remote Authentication Dial In User Service (RADIUS) Protocol By submitting this Internet-Draft, each author represents that any applicable patent or other IPR claims of which he or she is aware have been or will be disclosed, and any of which he or she becomes aware will be disclosed, in accordance with Section 6 of BCP 79. Internet-Drafts are working documents of the Internet Engineering Task Force (IETF), its areas, and its working groups. Note that other groups may also distribute working documents as Internet- Drafts. Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or to cite them other than as "work in progress." The list of current Internet-Drafts can be accessed at http://www.ietf.org/ietf/1id-abstracts.txt. The list of Internet-Draft Shadow Directories can be accessed at http://www.ietf.org/shadow.html. This Internet-Draft will expire on September 20, 2008.May 2, 2009. Copyright Notice Copyright (C) The IETF Trust (2008). Abstract RFC 2865 defines a Status-Server code for use in RADIUS, but labels it as "Experimental" without further discussion. This document describes a practical use for the Status-Server packet code, which is to let clients query the status of a RADIUS server. These queries, and responses (if any) enable the client to make more informed decisions. The result is a more stable, and more robust RADIUS architecture. Table of Contents 1. Introduction ............................................. 3 1.1. Terminology ......................................... 3 1.2. Requirements Language ............................... 4 2. Problem Statement ........................................ 5 2.1. Overloading Access-Request .......................... 5 2.1.1. Recommendation against Access-Request .......... 6 2.2. Overloading Accounting-Request ...................... 6 2.2.1. Recommendation against Accounting-Request ...... 7 2.3. Status-Server as a Solution ......................... 7 2.3.1. Status-Server to the RADIUS Authentication port 7 2.3.2. Status-Server to the RADIUS Accounting port ........................................................... 8 2.3.3. Status-Server to the RADIUS Change-of-Authorizat 8 3. Packet Format ............................................ 8 3.1. Consistent definition for Status-Server ............. 10 4. Implementation notes ..................................... 11 4.1. Client Requirements ................................. 12 4.2. Server Requirements ................................. 13 4.3. Change of Authorization and Status-Server ........... 15 4.4. More Robust Fail-over with Status-Server ............ 16 4.5. Proxy Server handling of Status-Server .............. 16 4.6. Realm Routing ....................................... 17 4.7. Management Information Base (MIB) Considerations .... 19 4.7.1. Interaction with RADIUS Server MIBs ............ 19 4.7.2. Interaction with RADIUS Client MIBs ............ 19 5. Additional considerations ................................ 20 5.1. Local site testing .................................. 20 5.2. RADIUS over reliable transports ..................... 21 5.3. Other uses for Status-Server ........................ 22 6. Table of Attributes ...................................... 22 7. Examples ................................................. 22 7.1. Minimal Query to Authentication Port ................ 23 7.2. Minimal Query to Accounting Port .................... 23 7.3. Verbose Query and Response .......................... 24 8. IANA Considerations ...................................... 25 9. Security Considerations .................................. 25 10. References .............................................. 25 10.1. Normative references ............................... 26 10.2. Informative references ............................. 26 Intellectual Property Statement .............................. 27 Disclaimer of Validity ....................................... 28 Full Copyright Statement ..................................... 28 1. Introduction The RADIUS Working Group was formed in 1995 to document the protocol of the same name, and created a number of standards surrounding the protocol. It also defined experimental commands within the protocol, without elaborating further on the potential uses of those commands. One of the commands so defined was Status-Server ([RFC2865] Section 3.). This document describes how some current implementations are using Status-Server packets as a method for querying the status of a RADIUS server. These queries do not otherwise affect the normal operation of a server, and do not result in any side effects other than perhaps incrementing an internal packet counter. These queries are not intended to implement the application-layer watchdog messages described in [RFC3539] Section 3.4. That document describes Authentication, Authorization, and Accounting (AAA) protocols that run over reliable transports which handle retransmissions internally. Since RADIUS runs over the User Datagram Protocol (UDP) rather than Transport Control Protocol (TCP), the full watchdog mechanism is not applicable here. The rest of this document is laid out as follows. Section 2 contains the problem statement, and explanations as to why some possible solutions can have unwanted side effects. Section 3 defines the Status-Server packet format. Section 4 contains client and server requirements, along with some implementation notes. Section 5 lists additional considerations not covered in the other sections. The remaining text contains a RADIUS table of attributes, and discussed security considerations not covered elsewhere in the document. 1.1. Terminology This document uses the following terms: Network Access Server (NAS) The device providing access to the network. Also known as the Authenticator (in IEEE 802.1x terminology) or RADIUS client. Home Server A RADIUS server that is authoritative for user authorization and authentication. Proxy Server A RADIUS server that acts as a Home Server to the NAS, but in turn proxies the request to another Proxy Server, or to a Home Server. silently discard This means the implementation discards the packet without further processing. The implementation MAY provide the capability of logging the error, including the contents of the silently discarded packet, and SHOULD record the event in a statistics counter. 1.2. Requirements Language In this document, several words are used to signify the requirements of the specification. The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in [RFC2119]. 2. Problem Statement It is often useful to know if a RADIUS server is alive and responding to requests. The most accurate way to obtain this information is to query the server via application protocol traffic, as other methods are either less accurate, or cannot be performed remotely. The reasons for wanting to know the status of a server are many. The administrator may simply be curious if the server is responding, and may not have access to NAS or traffic data that would give him that information. The queries may also be performed automatically by a NAS or proxy server, which is configured to send packets to a RADIUS server, and where that server may not be responding. That is, while [RFC2865] Section 2.6 indicates that sending Keep-Alives is harmful, it may be useful to send "Are you Alive" queries to a server once it has been marked "dead" due to prior unresponsiveness. The occasional query to a "dead" server offers little additional load on the network or server, and permits clients to more quickly discover when the server returns to a responsive state. Overall, status queries can be a useful part of the deployment of a RADIUS server. 2.1. Overloading Access-Request One possible solution to the problem of querying server status is for a NAS to send specially formed Access-Request packets to a RADIUS server's authentication port. The NAS can then look for a response, and use this information to determine if the server is active or unresponsive. However, the server may see the request as a normal login request for a user, and conclude that a real user has logged onto that NAS. The server may then perform actions that are undesirable for a simple status query. The server may alternatively respond with an Access- Challenge, indicating that it believes an extended authentication conversation is necessary. Another possibility is that the server responds with an Access- Reject, indicating that the user is not authorized to gain access to the network. As above, the server may also perform local site actions, such as warning an administrator of failed login attempts. The server may also delay the Access-Reject response, in the traditional manner of rate-limiting failed authentication attempts. This delay in response means that the querying administrator is unsure as to whether or not the server is down, is slow to respond, or is intentionally delaying it's response to the query. In addition, using Access-Request queries may mean that the server may have local users configured whose sole reason for existence is to enable these query requests. Unless the server's policy is designed carefully, it may be possible for an attacker to use those credentials to gain unauthorized network access. We note that some NAS implementations currently use Access-Request packets as described above, with a fixed (and non configurable) user name and password. Implementation issues with that equipment means that if a RADIUS server does not respond to those qeuries, it may be marked as unresponsive by the NAS. This marking may happen even if the server is actively responding to other Access-Requests from that same NAS. This behavior is confusing to administrators who then need to determine why an active server has been marked as "unresponsive". 2.1.1. Recommendation against Access-Request For the reasons outlined above, NAS implementors SHOULD NOT generate Access-Request packets solely to see if a server is alive. Similarly, site administrators SHOULD NOT configure test users whose sole reason for existence is to enable such queries via Access- Request packets. Note that it still may be useful to configure test users for the purpose of performing end-to-end or in-depth testing of a servers policy. While this practice is wide-spread, we caution administrators to use it with care. 2.2. Overloading Accounting-Request A similar solution for the problem of querying server status may be for a NAS to send specially formed Accounting-Request packets to a RADIUS servers authentication port. The NAS can then look for a response, and use this information to determine if the server is active or unresponsive. As seen above with Access-Request, the server may then conclude that a real user has logged onto a NAS, and perform local site actions that are undesirable for a simple status query. Another consideration is that some attributes are mandatory to include in an Accounting-Request. This requirement forces the administrator to query an accounting server with fake values for those attributes in a test packet. These fake values increase the work required to perform a simple query, and may pollute the server's accounting database with incorrect data. 2.2.1. Recommendation against Accounting-Request For the reasons outlined above, NAS implementors SHOULD NOT generate Accounting-Request packets solely to see if a server is alive. Similarly, site administrators SHOULD NOT configure accounting policies whose sole reason for existence is to enable such queries via Accounting-Request packets. Note that it still may be useful to configure test users for the purpose of performing end-to-end or in-depth testing of a servers policy. While this practice is wide-spread, we caution administrators to use it with care. 2.3. Status-Server as a Solution A better solution to the above problems is to use the Status-Server packet code. The name of the code leads us to conclude that it was intended for packets that query the status of a server. Since the packet is otherwise undefined, it does not cause interoperability issues to create implementation-specific definitions for it. The difficulty until now has been defining an inter-operable method of performing these queries. This document addresses that need. 2.3.1. Status-Server to the RADIUS Authentication port Status-Server SHOULD be used instead of Access-Request to query the responsiveness of a server. In this use-case, the protocol exchange between client and server is similar to the usual exchange of Access- Request and Access-Accept, as shown below. NAS RADIUS server --- ------------- Status-Server/ Message-Authenticator -> <- Access-Accept/ Reply-Message The Status-Server packet MUST contain a Message-Authenticator attribute for security. The Access-Acceptresponse (if any) to a Status-Server packet can optionally containsent to an informational Reply-Message attribute. Aauthentication port MUST be an Access-Accept packet. The list of attributes that are permitted in each type ofthe Access- Accept packet is given in the Table of attributesAttributes in Section 6, below. 2.3.2. Status-Server to the RADIUS Accounting port Status-Servermay be used instead of Accounting-Request to query the responsiveness of a server. In this use-case, the protocol exchange between client and server is similar to the usual exchange of Accounting-Request and Accounting-Response, as shown below. NAS RADIUS server --- ------------- Status-Server/ Message-Authenticator -> <- Accounting-Response The Status-Server packet MUST contain a Message-Authenticator attribute for security. The Accounting-Responseresponse (if any) to a Status-Server packet is empty. Asent to an accounting port MUST be an Accounting-Response packet. The list of attributes that are permitted in each type ofthe Accounting- Response packet is given in the Table of attributesAttributes in Section 6, below. 2.3.3. Status-Server to the RADIUS Change-of-Authorization port Status-Server may be pro-actively sent by a server to a NAS when the server first boots. This use mirrors the Accounting-Request use of the Acct-Status-Type attribute with value "Accounting On". This packet can serve as an indication to the NAS that the server is available for authentication and accounting requests. In this use-case, the protocol exchange between client and server is similar to the usual exchange of CoA-Request and CoA-ACK, as shown below. RADIUS Server NAS ------------- --- Status-Server/ Message-Authenticator -> <- CoA-ACK The Status-Server packet MUST contain a Message-Authenticator attribute for security. The CoA-ACKresponse (if any) to a Status-Server packet is empty. Asent to a Change-of-Authorization port MUST be a CoA-ACK packet. The list of attributes that are permitted in each type ofthe CoA-ACK packet is given in the Table of attributesAttributes in Section 6, below. 3. Packet Format Status-Server packets re-use the RADIUS packet format, with the fields and values for those fields as defined [RFC2865] Section 3. We do not include all of the text or diagrams of that section here, but instead explain the differences required to implement Status- Server. The Authenticator field of Status-Server packets MUST be generated using the same method as that used for the Request Authenticator field of Access-Request packets, as given below. The role of the Identifier field is the same for Status-Server as for other packets. However, as Status-Server is taking the role of Access-Request or Accounting-Request packets, there is the potential for Status-Server requests to be in conflict with Access-Request or Accounting-Request packets with the same Identifier. In Section 4.2, below, we describe a method for avoiding these problems. This method MUST be used to avoid conflicts between Status-Server and other packet types. Request Authenticator In Status-Server Packets, the Authenticator value is a 16 octet random number, called the Request Authenticator. The value SHOULD be unpredictable and unique over the lifetime of a secret (the password shared between the client and the RADIUS server), since repetition of a request value in conjunction with the same secret would permit an attacker to reply with a previously intercepted response. Since it is expected that the same secret MAY be used to authenticate with servers in disparate geographic regions, the Request Authenticator field SHOULD exhibit global and temporal uniqueness. The Request Authenticator value in a Status-Server packet SHOULD also be unpredictable, lest an attacker trick a server into responding to a predicted future request, and then use the response to masquerade as that server to a future Status-Server request from a client. Similarly, the Response Authenticator field of an Access-Accept packet sent in response to Status-Server queries MUST be generated using the same method as used for for calculating the Response Authenticator of the Access-Accept, with the Status-Server Request Authenticator taking the place of the Access-Request Request Authenticator. The Response Authenticator field of an Accounting-Response packet sent in response to Status-Server queries MUST be generated using the same method as used for for calculating the Response Authenticator of the Accounting-Response, with the Status-Server Request Authenticator taking the place of the Accounting-Request Request Authenticator. Note that when a server responds to a Status-Server request, it MUST NOTE send more than one response packet. Response Authenticator The value of the Authenticator field in Access-Accept, or Accounting-Response packets is called the Response Authenticator, and contains a one-way MD5 hash calculated over a stream of octets consisting of: the RADIUS packet, beginning with the Code field, including the Identifier, the Length, the Request Authenticator field from the Status-Server packet, and the response Attributes (if any), followed by the shared secret. That is, ResponseAuth = MD5(Code+ID+Length+RequestAuth+Attributes+Secret) where + denotes concatenation. In addition to the above requirements, all Status-Server packets MUST include a Message-Authenticator attribute. Failure to do so would mean that the packets could be trivially spoofed. Status-Server packets MAY include NAS-Identifier, one of NAS-IP- Address or NAS-IPv6-Address, and Reply-Message. These attributes are not necessary for the operation of Status-Server, but may be useful information to a server that receives those packets. Other attributes SHOULD NOT be included in a Status-Server packet. User authentication credentials such as User-Password, CHAP-Password, EAP-Message, etc. MUST NOT appear in a Status-Server packet sent to a RADIUS authentication port. User or NAS accounting attributes such as Acct-Session-Id, Acct-Status-Type, Acct-Input-Octets, etc. MUST NOT appear in a Status-Server packet sent to a RADIUS accounting port. The Access-Accept MAY contain a Reply-Message or Message- Authenticator attribute. It SHOULD NOT contain other attributes. The Accounting-Response packets sent in response to a Status-Server query SHOULD NOT contain any attributes. As the intent is to implement a simple query instead of user authentication or accounting, there is little reason to include other attributes in either the query or the corresponding response. Examples of Status-Server packet flows are given below in Section 7. 3.1. Consistent definition for Status-Server When sent to a RADIUS accounting port, contents of the Status-Server packets are calculated as described above. That is, even though the packets are being sent to an accounting port, they are not created using the same method as for Accounting-Requests. This difference has a number of benefits. Having one definition for Status-Server packets is simpler than having different definitions for different destination ports. In addition, if we were to define Status-Server as being similar to Accounting-Request or CoA-Request, but containing no attributes, then those packets could be trivially forged. We therefore define Status-Server consistently, and vary the response packets depending on the port to which the request is sent. When sent to an authentication port, the response to a Status-Server query is an Access-Accept packet. When sent to an accounting port, the response to a Status-Server query is an Accounting-Response packet. When sent to a change of authorization (CoA) port, the response to a Status-Server query is an CoA-ACK packet 4. Implementation notes There are a number of considerations to take into account when implementing support for Status-Server. This section describes implementation details and requirements for RADIUS clients and servers that support Status-Server. The following text applies to the authentication, accounting, and coa ports. We use the generic terms below to simplify the discussion: * Request packet An Access-Request packet sent to an authentication port, or an Accounting-Request packet sent to an accounting port, or a Change-Of-Authorization packet sent to a CoA port. * Response packet An Access-Accept, Access-Challenge, or Access-Reject packet sent from an authentication port, or an Accounting-Response packet sent from an accounting port, or a CoA-ACK packet sent from a CoA port. We also refer to "client" as the originator of the Status-Server packet, and "server" as the receiver of that packet, and the originator of the Response packet. Using generic terms to describe the Status-Server conversations is simpler than duplicating the text for authentication, accounting, and coa packets. 4.1. Client Requirements Clients SHOULD permit administrators to globally enable or disable the generation of Status-Server packets. The default SHOULD be that it is disabled. As it is undesirable to send queries to servers that do not support Status-Server, clients SHOULD also have a per-server configuration indicating whether or not to enable Status-Server for a particular destination. The default SHOULD be that it is disabled. The client SHOULD also have a configurable global timer (Tw) that is used when sending periodic Status-Server queries during server fail- over. The default value SHOULD be 30 seconds, and the value MUST NOT be permitted to be set below 6 seconds. If a response has not been received within the timeout period, the Status-Server packet is deemed to have received no corresponding Response packet, and MUST be discarded. Clients SHOULD use a jitter of +/- 2 seconds when sending periodic Status-Server packets, in order to avoid synchronization. When Status-Server packets are sent from a client, they MUST NOT be retransmitted. Instead, the Identity field MUST be changed every time a packet is transmitted. The old packet should be discarded, and a new Status-Server packet should be generated and sent, with new Identity and Authenticator fields. Clients MUST include the Message-Authenticator attribute in all Status-Server packets. Failure to do so would mean that the packets could be trivially spoofed, leading to potential denial of service (DoS) attacks. Other attributes SHOULD NOT appear in a Status-Server packet, except as outlined below in Section 6. As the intent of the packet is a simple status query, there is little reason for any additional attributes to appear in Status-Server packets. The client MAY increment packet counters as a result of sending a Status-Server request, or receiving a Response packet. The client MUST NOT perform any other action that is normally performed when it receives a Response packet, such as permitting a user to have login access to a port. Clients MAY send Status-Server requests to the RADIUS destination ports from the same source port used to send normal Request packets. Other clients MAY choose to send Status-Server requests from a unique source port, that is not used to send Request packets. The above suggestion for a unique source port for Status-Server packets aids in matching responses to requests. Since the response to a Status-Server packet is an Access-Accept, Accounting-Response, or CoA-ACK packet, those responses are indistinguishable from other packets sent in response to a Request packet. Therefore, the best way to distinguish them from other traffic is to have a unique port. A client MAY send a Status-Server packet from a source port also used to send Request packets. In that case, the Identifer field MUST be unique across all outstanding Request packets for that source port, independent of the value of the RADIUS Code field for those outstanding requests. Once the client has either received a response to the Status-Server packet, or has determined that the Status-Server packet has timed out, it may re-use that Identifier in another packet. Robust implementations SHOULD accept any Response packet as a valid response to a Status-Server packet, subject to the validation requirements defined above for the Response Authenticator. The code field of the packet matters less than the fact that a valid, signed, response has been received. That is, prior to accepting the response as valid, the client should check that the Response packet Code field is either Access-Accept (2), Accounting-Response (5), or CoA-ACK (44). If the code does not match any of these values, the packet MUST be silently discarded. The client MUST then validate the Response Authenticator via the algorithm given above in Section 3. If the Response Authenticator is not valid, the packet MUST be silently discarded. If the Response Authenticator is valid, then the packet MUST be deemed to be a valid response from the server. If the client instead discarded the response because the packet code did not match what it expected, then it could erroneously discard valid responses from a server, and mark that server as unresponsive. This behavior would affect the stability of a RADIUS network, as responsive servers would erroneously be marked as unresponsive. We therefore recommend that clients should be liberal in what they accept as responses to Status-Server queries. 4.2. Server Requirements Servers SHOULD permit administrators to globally enable or disable the acceptance of Status-Server packets. The default SHOULD be that it is enabled. Servers SHOULD also permit adminstrators to enable or disable acceptance of Status-Server packets on a per-client basis. The default SHOULD be that it is enabled. Status-Server packets originating from clients that are not permitted to send the server Request packets MUST be silently discarded. If a server does not support Status-Server packets, or is configured to not respond to them, then it MUST silently discard the packet. We note that [RFC2865] Section 3 defines a number of RADIUS Codes, but does not make statements about which Codes are valid for port 1812. In contrast, [RFC2866] Section 3 specifies that only RADIUS Accounting packets are to be sent to port 1813. This specification is compatible with [RFC2865], as it uses a known Code for packets to port 1812. This specification is not compatible with [RFC2866], as it adds a new code (Status-Server) that is valid for port 1812. However, as the category of [RFC2866] is Informational, this conflict is acceptable. Servers SHOULD silently discard Status-Server packets if they determine that a client is sending too many Status-Server requests in a particular time period. The method used by a server to make this determination is implementation-specific, and out of scope for this specification. If a server supports Status-Server packets, and is configured to respond to them, and receives a packet from a known client, it MUST validate the Message-Authenticator attribute as defined in [RFC3579] Section 3.2. Packets failing that validation MUST be silently discarded. Servers SHOULD NOT otherwise discard Status-Server packets if they have recently sent the client a Response packet. The query may have originated from an administrator who does not have access to the Response packet stream, or who is interested in obtaining additional information about the server. The server MAY prioritize the handling Status-Server queries over the handling of other requests, subject to the rate limiting described above. The server MAY decide to not respond to a Status-Server, depending on local site policy. For example, a server that is running but is unable to perform it's normal activities MAY silently discard Status- Server packets. This situation can happen, for example, when a server requires access to a database for normal operation, but the connection to that database is down. Or, it may happen when the accepted load on the server is lower than the offered load. Some server implementations require that Access-Request packets are accepted only on "authentication" ports, (e.g. 1812/udp), and that Accounting-Request packets are accepted only on "accounting" ports (e.g. 1813/udp). Those implementations SHOULD reply to Status-Server packets sent to an "authentication" port with an Access-Accept packet. Those implementations SHOULD reply to Status-Server packets sent to an "accounting" port with an Accounting-Response packet. Some server implementations accept both Access-Request and Accounting-Request packets on the same port, and do not distinguish between "authentication only" ports, and "accounting only" ports. Those implementations SHOULD reply to Status-Server packets with an Access-Accept packet. The server MAY increment packet counters as a result of receiving a Status-Server, or sending a Response packet. The server SHOULD NOT perform any other action that is normally performed when it receives a Request packet, other than sending a Response packet. 4.3. Change of Authorization and Status-Server The use of Status-Server with respect to Change of Authorization requires some additional discussion. When a Dynamic Authorization Client ([RFC5176] Section 1.3) reboots, it SHOULD send a Status-Server packet to a CoA port to IP addresses that are configured as both Dynamic Authorization Servers and RADIUS clients. Jitter SHOULD be used to avoid synchronization issues. If there is no response to a packet, the periodic timer above SHOULD be used to continue sending packets to that destination until a response has been received. When a response is received, the Dynamic Authorization Client MUST NOT send further Status-Server packets to the CoA port of any Dynamic Authorization until it next reboots. When a Dynamic Authorization Server receives a Status-Server packet to it's CoA port, it SHOULD respond with a CoA-ACK packet, as described above. It MAY use this information to modify it's authentication and/or accounting behavior, as described below. If the Status-Server packet came to a NAS CoA port from an IP address which is also configured as an authentication and/or accounting server. the NAS MAY decide to mark the RADIUS server as being responsive. If the RADIUS server had previously been marked as unresponsive, this change would enable the NAS to start packets to start sending packets to that RADIUS server again. The NAS MAY otherwise decide to receive multiple packets to it's CoA port before marking the RADIUS server as responsive. This behavior is implementation-defined, and SHOULD be configurable. Where possible, the Dynamic Authorization Client (usually a RADIUS server) SHOULD originate the Status-Server packet from the port to which the NAS would normally send Request packets. For example, a packet sent from from a RADIUS server with source port 1812 to a NAS with destination port 3799, would indicate to the NAS that the RADIUS authentication server at that address is alive. 4.4. More Robust Fail-over with Status-Server A common problem in RADIUS client implementations is the implementation of a robust fail-over mechanism between servers. A client may have multiple servers configured, with one server marked as primary and another marked as secondary. If the client determines that the primary is unresponsive, it can "fail over" to the secondary, and send requests to the secondary instead of to the primary. However, it is difficult in standard RADIUS for a client to know when it should start sending requests to the primary again. Sending test Access-Requests or Accounting-Requests to see if the server is alive has the issues outlined above in Section 2. Clients could alternately send real traffic to the primary, on the hope that it is responsive. If the server is still unresponsive, however, the result may be user login failures. The Status-Server solution is an ideal one to solve this problem. When a client fails over from one server to another because of a lack of responsiveness, it SHOULD send periodic Status-Server packets to the unresponsive server, using the timer (Tw) defined above. Once three time periods have passed where Status-Server messages have been sent and responded to, the server should be deemed responsive and RADIUS requests may sent to it again. This determination should be made separately for each server that the client has a relationship with. The same algorithm should be used for both authentication and accounting ports. The client MUST treat each destination (ip, port) combination as a unique server for the purposes of this determination. The practice of sending Status-Server packets to CoA ports (where applicable) can increase the information available in the network, and further help to stabilize the network, and to lower response times in the event of network changes. The above behavior is modelled after [RFC3539] Section 3.4.1. We note that if a reliable transport is used for RADIUS, then the algorithms specified in [RFC3539] MUST be used in preference to the ones given here. 4.5. Proxy Server handling of Status-Server Many RADIUS servers can act as proxy servers, and can forward requests to home servers. Such servers MUST NOT proxy Status-Server packets. The purpose of Status-Server as specified here is to permit the client to query the responsiveness of a server that it has a direct relationship with. Proxying Status-Server queries would negate any usefulness that may be gained by implementing support for them. Proxy servers MAY be configured to respond to Status-Server queries from clients, and MAY act as clients sending Status-Server queries to other servers. However, those activities MUST be independent of one another. 4.6. Realm Routing RADIUS servers are commonly used in an environment where Network Access Identifiers (NAIs) are used as routing identifiers [RFC4282]. In this practice, the User-Name attribute is decorated with realm routing information, commonly in the format of "user@realm". Since a particular RADIUS server may act as a proxy for more than one realm, the mechanism outlined above may be inadequate. The schematic below demonstrates this scenario. /-> Proxy Server P -----> Home Server for Realm A / \ / NAS X \ / \ \-> Proxy Server S -----> Home Server for Realm B That is, the NAS has relationships with two Proxy Servers, P and S. Each Proxy Server has relationships with Home Servers for both Realm A and Realm B. In this scenario, the Proxy Servers can determine if one or both of the Home Servers are dead or unreachable. The NAS can determine if one or both of the Proxy Servers are dead or unreachable. There is an additional case to consider, however. If Proxy Server P cannot reach the Home Server for Realm A, but the Proxy Server S can reach that Home Server, then the NAS cannot discover this information using the Status-Server queries as outlined above. It would therefore be useful for the NAS to know that Realm A is reachable from Proxy Server S, as it can then route all requests for Realm A to that Proxy Server. Without this knowledge, the client may route requests to Proxy Server P, where they may be discarded or rejected. To complicate matters, the behavior of Proxy Servers P and S in this situation is not well defined. Some implementations simply fail to respond to the request, and other implementations respond with an Access-Reject. If the implementation fails to respond, then the NAS cannot distinguish between the Proxy Server being down, or the next server along along the proxy chain is unreachable. In the worst case, failures in routing for Realm A may affect users Realm B. For example, if Proxy Server P can reach Realm B but not Realm A, and Proxy Server S can reach Realm A but not Realm B, then active paths exist to handle all RADIUS requests. However, depending on the NAS and Proxy Server implementation choices, the NAS may not be able to determine which server requests may be sent to in order to maintain network stability. This problem cannot, unfortunately be solved by using Status-Server requests. A robust solution would involve either a RADIUS routing table for the NAI realms, or a RADIUS "destination unreachable" response to authentication requests. Either solution would not fit into the traditional RADIUS model, and both are therefore outside of the scope of this specification. The problem is discussed here in order to define how best to use Status-Server in this situation, rather than to define a new solution. When a server has responded recently to a request from a client, that client MUST mark the server as "responsive". In the above case, a Proxy Server may be responding to requests destined for Realm A, but not responding to requests destined for Realm B. The client therefore considers the server to be responsive, as it is receiving responses from the server. The client will then continue to send requests to the Proxy Server for destination Realm B, even though the Proxy Server cannot route the requests to that destination. This failure is a known limitation of RADIUS, and can be partially addressed through the use of failover in the Proxy Servers. A more realistic situation than the one outlined above is where each Proxy Server also has multiple choices of Home Servers for a realm, as outlined below. /-> Proxy Server P -----> Home Server P / \ / NAS X \ / \ \-> Proxy Server S -----> Home Server S In this situation, if all participants impement Status-Server as defined herein, any one link may be broken, and all requests from the NAS will still reach a home server. If two links are broken at different places, (i.e. not both links from the NAS), then all requests from the NAS will still reach a home server. In many situations where three or more links are broken, then requests from the NAS may still reach a home server. It is RECOMMENDED, therefore, that implementations desiring the most benefit from Status-Server also implement server failover. The combination of these two practices will maximize network reliability and stability. 4.7. Management Information Base (MIB) Considerations 4.7.1. Interaction with RADIUS Server MIBs Since Status-Server packets are sent to the defined RADIUS ports, they can affect the [RFC4669] and [RFC4671] RADIUS server MIBs. [RFC4669] defines a counter named radiusAuthServTotalUnknownTypes, that counts "The number of RADIUS packets of unknown type that were received". [RFC4671] defines a similar counter named radiusAcctServTotalUnknownTypes. Implementations not supporting Status-Server, or implementations that are configured to not respond to Status-Server packets MUST use these counters to track received Status-Server packets. If, however, Status-Server is supported and the server is configured to respond as described above, then the counters defined in [RFC4669] and [RFC4671] MUST NOT be used to track Status-Server requests or responses to those requests. That is, when a server fully implements Status-Server, the counters defined in [RFC4669] and [RFC4671] MUST be unaffected by the transmission or reception of packets relating to Status-Server. If a server supports Status-Server and the [RFC4669] or [RFC4671] MIBs, then it SHOULD also support vendor-specific MIBs containing similar information as the standard MIBs, but which are instead dedicated solely to tracking Status-Server requests and responses. Any definition of the server MIBs for Status-Server is outside of the scope of this document. 4.7.2. Interaction with RADIUS Client MIBs Clients implementing Status-Server MUST NOT increment [RFC4668] or [RFC4670] counters upon reception of Response packets to Status- Server queries. That is, when a server fully implements Status- Server, the counters defined in [RFC4668] and [RFC4670] MUST be unaffected by the transmission or reception of packets relating to Status-Server. If an implementation supports Status-Server and the [RFC4668] or [RFC4670] MIBs, then it SHOULD also support vendor-specific MIBs containing similar information as those MIBs, but which are instead dedicated solely to tracking Status-Server requests and responses. Any definition of the client MIBs for Status-Server is outside of the scope of this document. 5. Additional considerations There are additional topics related to the use of Status-Server that may be covered. As those topics do not fit well into the preceding sections, they are covered herein. 5.1. Local site testing There is at least one situation where using Access-Request or Accounting-Request packets may be useful, despite the recommendations above in Section 2.1.1 and Section 2.2.1. That situation is local site testing, where the RADIUS client, server, and user store are under the control of a single administrator or administrative entity. In that situation, administrators MAY configure a well-known "test" user to enable local site testing. The advantage to creating such a local user is that it is now possible for the administrator to send a RADIUS request that performs end-to-end testing of the RADIUS server. As above with Status- Server, this testing includes RADIUS server responsiveness. It may also include querying databases of user authentication credentials, or storing accounting data to a billing database. The information obtained from performing those queries is that the entire RADIUS server infrastructure, including all of it's dependencies, is functioning as expected. These queries are most useful in deployments where an administrator has internal RADIUS server that proxy to other internal RADIUS servers, such as for load balancing or fail over. If used, the names used for these test users SHOULD be difficult to guess by an attacker. An Access-Request packet for a test user otherwise should be treated as follows, depending on its origin: o Packets from localhost (127.0.0.1 or ::1): RADIUS servers SHOULD treat the request according to local site policy. o Packets from NASes that normally originate Access-Request packets (i.e. not proxy servers): RADIUS servers SHOULD respond with an Access-Reject packet, as the use of Status-Server is preferred. o Packets from other machines controlled by the administrator: RADIUS servers SHOULD treat the request according to local site policy. o Packets originating from machines not controlled by the administrator: RADIUS servers MUST respond with an Access-Reject packet. If a RADIUS server is configured to support test users for Accounting-Request packets, it MAY respond with an Accounting- Response packet, independent of the origin of the request. However, any subsequent analysis of the accounting data such as billing or usage MUST NOT include the data for the test user. If these recommendations are implemented, then it may be possible in some situations to safely query a RADIUS server for responsiveness using Access-Request or Accounting-Request packets. However, this behavior is still NOT RECOMMENDED. 5.2. RADIUS over reliable transports Although RADIUS has been assigned two TCP ports (1812/tcp and 1813/tcp) in addition to the commonly used UDP ports, there has been as yet no specification for using TCP as a reliable transport for RADIUS. If such a specification were to be created, then the transport issues discussed in [RFC3539] would apply. Further, when RADIUS is run over reliable transports, the watchdog algorithm described in [RFC3539] Section 3.4 MUST be used rather than the algorithm described above. For the reasons outlined above in Section 2, Status-Server packets SHOULD be used as the watchdog request, in preference to Access-Request or Accounting-Request packets. Clients sending Status-Server over reliable transport MUST ensure that the Identifier field is unique for all requests on a particular connection, independent of the packet code. That is, if a Status- Server with a particular value in the Identifier field is sent to a server, the client MUST NOT simultaneously send an Access-Request or Accounting-Request packet with that same Identifier value, on that connection. Once the client has either received a response to the Status-Server packet, or has determined that the Status-Server packet has timed out, it may re-use that Identifier in another packet. 5.3. Other uses for Status-Server While other uses of Status-Server are possible, uses beyond those specified here are beyond the scope of this document. It may be tempting to increase the utility of Status-Server by having the responses carry additional information, implementors are warned that such uses have not been analyzed for potential security issues or network problems. Specifically, it may seem useful to leverage a combination of Status- Server and CoA ports in order to send realm routing information "upstream" from the home servers to the proxy servers, and finally to the NAS. This use of Status-Server is NOT RECOMMENDED, as there has been insufficient analysis and deployment experience to know if it is useful, or even if it makes the network less reliable. 6. Table of Attributes The following table provide a guide to which attributes may be found in Status-Server packets, and in what quantity. No attributes other than the ones listed below should be found in a Status-Server packet. Status- Access- Accounting- CoA- Server Accept Response ACK # Attribute 0-1 0 0 0 4 NAS-IP-Address 0 0+ 0 0 18 Reply-Message 0+ 0+ 0+ 0 26 Vendor-Specific 0+ 0+ 0 0 31 Calling-Station-Id 0-1 0 0 0 32 NAS-Identifier 1 0-1 0-1 0-1 80 Message-Authenticator 0-1 0 0 0 95 NAS-IPv6-Address The following table defines the meaning of the above table entries. 0 This attribute MUST NOT be present in packet. 0+ Zero or more instances of this attribute MAY be present in packet. 0-1 Zero or one instance of this attribute MAY be present in packet. 1 Exactly one instance of this attribute MUST be present in packet. 7. Examples A few examples are presented to illustrate the flow of packets to both the authentication and accounting ports. These examples are not intended to be exhaustive, many others are possible. Hexadecimal dumps of the example packets are given in network byte order, using the shared secret "xyzzy5461". 7.1. Minimal Query to Authentication Port The NAS sends a Status-Server UDP packet with minimal content to a RADIUS server on port 1812. The Request Authenticator is a 16 octet random number generated by the NAS. Message-Authenticator is included in order to authenticate that the request came from a known client. 0c da 00 26 8a 54 f4 68 6f b3 94 c5 28 66 e3 02 18 5d 06 23 50 12 5a 66 5e 2e 1e 84 11 f3 e2 43 82 20 97 c8 4f a3 1 Code = Status-Server (12) 1 ID = 218 2 Length = 38 16 Request Authenticator Attributes: 18 Message-Authenticator (80) = 5a665e2e1e8411f3e243822097c84fa3 The Response Authenticator is a 16-octet MD5 checksum of the code (2), id (218), Length (20), the Request Authenticator from above, and the shared secret. 02 da 00 14 ef 0d 55 2a 4b f2 d6 93 ec 2b 6f e8 b5 41 1d 66 1 Code = Access-Accept (2) 1 ID = 218 2 Length = 20 16 Request Authenticator Attributes: None. 7.2. Minimal Query to Accounting Port The NAS sends a Status-Server UDP packet with minimal content to a RADIUS server on port 1813. The Request Authenticator is a 16 octet random number generated by the NAS. Message-Authenticator is included in order to authenticate that the request came from a known client. 0c b3 00 26 92 5f 6b 66 dd 5f ed 57 1f cb 1d b7 ad 38 82 60 80 12 e8 d6 ea bd a9 10 87 5c d9 1f da de 26 36 78 58 1 Code = Status-Server (12) 1 ID = 179 2 Length = 38 16 Request Authenticator Attributes: 18 Message-Authenticator (80) = e8d6eabda910875cd91fdade26367858 The Response Authenticator is a 16-octet MD5 checksum of the code (5), id (179), Length (20), the Request Authenticator from above, and the shared secret. 02 b3 00 1a 0f 6f 92 14 5f 10 7e 2f 50 4e 86 0a 48 60 66 9c 1 Code = Accounting-Response (5) 1 ID = 179 2 Length = 20 16 Request Authenticator Attributes: None. 7.3. Verbose Query and Response The NAS at 192.0.2.16 sends a Status-Server UDP packet to the RADIUS server on port 1812. The Request Authenticator is a 16 octet random number generated by the NAS. 0c 47 00 2c bf 58 de 56 ae 40 8a d3 b7 0c 85 13 f9 b0 3f be 04 06 c0 00 02 10 50 12 85 2d 6f ec 61 e7 ed 74 b8 e3 2d ac 2f 2a 5f b2 1 Code = Status-Server (12) 1 ID = 71 2 Length = 44 16 Request Authenticator Attributes: 6 NAS-IP-Address (4) = 192.0.2.16 18 Message-Authenticator (80) = 852d6fec61e7ed74b8e32dac2f2a5fb2 The Response Authenticator is a 16-octet MD5 checksum of the code (2), id (71), Length (52), the Request Authenticator from above, the attributes in this reply, and the shared secret. The Reply-Message is "RADIUS Server up 2 days, 18:40" 02 47 00 34 46 f4 3e 62 fd 03 54 42 4c bb eb fd 6d 21 4e 06 12 20 52 41 44 49 55 53 20 53 65 72 76 65 72 20 75 70 20 32 20 64 61 79 73 2c 20 31 38 3a 34 30 1 Code = Access-Accept (2) 1 ID = 71 2 Length = 52 16 Request Authenticator Attributes: 32 Reply-Message (18) 8. IANA Considerations This specification does not create any new registries, nor does it require assignment of any protocol parameters. 9. Security Considerations This document defines the Status-Server packet as being similar in treatment to the Access-Request packet, and is therefore subject to the same security considerations as described in [RFC2865], Section 8. Status-Server packets also use the Message-Authenticator attribute, and are therefore subject to the same security considerations as [RFC3579], Section 4. We reiterate that Status-Server packets MUST contain a Message- Authenticator attribute. Early implementations supporting Status- Server did not enforce this requirement, and may have been vulnerable to DoS attacks as a result. Where this document differs from [RFC2865] is that it defines a new request/response method in RADIUS; the Status-Server request. As this use is based on previously described and implemented standards, we know of no additional security considerations that arise from the use of Status-Server as defined herein. 10. References 10.1. Normative references [RFC2865] Rigney, C., Willens, S., Rubens, A. and W. Simpson, "Remote Authentication Dial In User Service (RADIUS)", RFC 2865, June 2000. [RFC2866] Rigney, C., "RADIUS Accounting", RFC 2866, June 2000. [RFC3579] Aboba, B., Calhoun, P., "RADIUS (Remote Authentication Dial In User Service) Support For Extensible Authentication Protocol (EAP)", RFC 3579, September 2003. [RFC4282] Aboba, B., and Beadles, M. at al, "The Network Access Identifier", RFC 4282, December 2005. 10.2. Informative references [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate Requirement Levels", RFC 2119, March, 1997. [RFC3539] Aboba, B., Wood, J., "Authentication, Authorization, and Accounting (AAA) Transport Profile", RFC 3539, June 2003. [RFC4668] Nelson, D., "RADIUS Authentication Client MIB for IPv6", RFC 4668, August 2006. [RFC4669] Nelson, D., "RADIUS Authentication Server MIB for IPv6", RFC 4669, August 2006. [RFC4670] Nelson, D., "RADIUS Accounting Client MIB for IPv6", RFC 4670, August 2006. [RFC4671] Nelson, D., "RADIUS Accounting Server MIB for IPv6", RFC 4671, August 2006. [RFC5176] Chiba, M., Eklund, M., et al, "Dynamic Authorization Extensions to Remote Authentication Dial In User Service (RADIUS)", RFC 5176, January 2008. Acknowledgments Parts of the text in Section 3 defining the Request and Response Authenticators were taken with minor edits from [RFC2865] Section 3. The author would like to thank Mike McCauley of Open Systems Consultants for making a Radiator server available for inter- operability testing. Authors' Addresses Alan DeKok The FreeRADIUS Server Project http://freeradius.org Email: email@example.com Intellectual Property Statement The IETF takes no position regarding the validity or scope of any Intellectual Property Rights or other rights that might be claimed to pertain to the implementation or use of the technology described in this document or the extent to which any license under such rights might or might not be available; nor does it represent that it has made any independent effort to identify any such rights. Information on the procedures with respect to rights in RFC documents can be found in BCP 78 and BCP 79. Copies of IPR disclosures made to the IETF Secretariat and any assurances of licenses to be made available, or the result of an attempt made to obtain a general license or permission for the use of such proprietary rights by implementers or users of this specification can be obtained from the IETF on-line IPR repository at http://www.ietf.org/ipr. The IETF invites any interested party to bring to its attention any copyrights, patents or patent applications, or other proprietary rights that may cover technology that may be required to implement this standard. Please address the information to the IETF at ietf- firstname.lastname@example.org. Disclaimer of Validity This document and the information contained herein are provided on an "AS IS" basis and THE CONTRIBUTOR, THE ORGANIZATION HE/SHE REPRESENTS OR IS SPONSORED BY (IF ANY), THE INTERNET SOCIETY, THE IETF TRUST AND THE INTERNET ENGINEERING TASK FORCE DISCLAIM ALL WARRANTIES, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO ANY WARRANTY THAT THE USE OF THE INFORMATION HEREIN WILL NOT INFRINGE ANY RIGHTS OR ANY IMPLIED WARRANTIES OF MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. Full Copyright Statement Copyright (C) The IETF Trust (2008). This document is subject to the rights, licenses and restrictions contained in BCP 78, and except as set forth therein, the authors retain all their rights. Acknowledgment Funding for the RFC Editor function is currently provided by the Internet Society.