idnits 2.17.1 draft-andrews-dns-no-response-issue-05.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt: ---------------------------------------------------------------------------- == No 'Intended status' indicated for this document; assuming Proposed Standard Checking nits according to https://www.ietf.org/id-info/checklist : ---------------------------------------------------------------------------- ** The document seems to lack a Security Considerations section. ** The document seems to lack an IANA Considerations section. (See Section 2.2 of https://www.ietf.org/id-info/checklist for how to handle the case when there are no actions for IANA.) Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year -- The document date (Oct 2014) is 3475 days in the past. Is this intentional? Checking references for intended status: Proposed Standard ---------------------------------------------------------------------------- (See RFCs 3967 and 4897 for information about using normative references to lower-maturity documents in RFCs) ** Obsolete normative reference: RFC 5966 (Obsoleted by RFC 7766) Summary: 3 errors (**), 0 flaws (~~), 2 warnings (==), 1 comment (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 Network Working Group M. Andrews 3 Internet-Draft ISC 4 Expires: April 4, 2015 Oct 2014 6 A Common Operational Problem in DNS Servers - Failure To Respond. 7 draft-andrews-dns-no-response-issue-05.txt 9 Abstract 11 The DNS is a query / response protocol. Failure to respond to 12 queries causes both immediate operational problems and long term 13 problems with protocol development. 15 This document identifies a number of common classes of queries that 16 some servers fail to respond too. This document also suggests 17 procedures for TLD and other similar zone operators to apply to 18 reduce / eliminate the problem. 20 Status of this Memo 22 This Internet-Draft is submitted in full conformance with the 23 provisions of BCP 78 and BCP 79. 25 Internet-Drafts are working documents of the Internet Engineering 26 Task Force (IETF). Note that other groups may also distribute 27 working documents as Internet-Drafts. The list of current Internet- 28 Drafts is at http://datatracker.ietf.org/drafts/current/. 30 Internet-Drafts are draft documents valid for a maximum of six months 31 and may be updated, replaced, or obsoleted by other documents at any 32 time. It is inappropriate to use Internet-Drafts as reference 33 material or to cite them other than as "work in progress." 35 This Internet-Draft will expire on April 4, 2015. 37 Copyright Notice 39 Copyright (c) 2014 IETF Trust and the persons identified as the 40 document authors. All rights reserved. 42 This document is subject to BCP 78 and the IETF Trust's Legal 43 Provisions Relating to IETF Documents 44 (http://trustee.ietf.org/license-info) in effect on the date of 45 publication of this document. Please review these documents 46 carefully, as they describe your rights and restrictions with respect 47 to this document. Code Components extracted from this document must 48 include Simplified BSD License text as described in Section 4.e of 49 the Trust Legal Provisions and are provided without warranty as 50 described in the Simplified BSD License. 52 Table of Contents 54 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 3 55 2. Common queries class that result in non responses. . . . . . . 4 56 2.1. EDNS Queries - Version Independent . . . . . . . . . . . . 4 57 2.2. EDNS Queries - Version Specific . . . . . . . . . . . . . . 4 58 2.3. EDNS Options . . . . . . . . . . . . . . . . . . . . . . . 4 59 2.4. EDNS Flags . . . . . . . . . . . . . . . . . . . . . . . . 5 60 2.5. Unknown / Unsupported Type Queries . . . . . . . . . . . . 5 61 2.6. TCP Queries . . . . . . . . . . . . . . . . . . . . . . . . 5 62 3. Remediating . . . . . . . . . . . . . . . . . . . . . . . . . . 5 63 4. Firewalls and Load Balancers . . . . . . . . . . . . . . . . . 7 64 5. Response Code Selection . . . . . . . . . . . . . . . . . . . . 7 65 6. Normative References . . . . . . . . . . . . . . . . . . . . . 8 66 Author's Address . . . . . . . . . . . . . . . . . . . . . . . . . 8 68 1. Introduction 70 The DNS [RFC1034], [RFC1035] is a query / response protocol. Failure 71 to respond to queries causes both immediate operational problems and 72 long term problems with protocol development. 74 Failure to respond to a query is indistinguishable from a packet loss 75 without doing a analysis of query response patterns and results in 76 unnecessary additional queries being made by DNS clients and 77 unnecessary delays being introduced to the resolution process. 79 Due to the inability to distingish between packet loss and 80 nameservers dropping EDNS [RFC6891] queries, packet loss is sometimes 81 misclassified as lack of EDNS support which can lead to DNSSEC 82 validation failures. 84 Allowing servers which fail to respond to queries to remain results 85 in developers being afraid to deploy implementations of recent 86 standards. Such servers need to be identified and corrected / 87 replaced. 89 The DNS has response codes that cover almost any conceivable query 90 response. A nameserver should be able to respond to any conceivable 91 query using them. 93 Unless a nameserver is under attack, it should respond to all queries 94 directed to it as a result of following delegations. Additionally 95 code should not assume that there isn't a delegation to the server 96 even if it is not configured to serve the zone. Broken delegation 97 are a common occurrence in the DNS and receiving queries for zones 98 that you are not configured for is not a necessarily a indication 99 that you are under attack. Parent zone operators are supposed to 100 regularly check that the delegating NS records are consistent with 101 those of the delegated zone and to correct them when they are not 102 [RFC1034]. If this was being done regularly the instances of broken 103 delegations would be much lower. 105 When a nameserver is under attack it may wish to drop packets. A 106 common attack is to use a nameserver as a amplifier by sending 107 spoofed packets. This is done because response packets are bigger 108 than the queries and big amplification factors are available 109 especially if EDNS is supported. Limiting the rate of responses is 110 reasonable when this is occuring and the client should retry. This 111 however only works if legitimate clients are not being forced to 112 guess whether EDNS queries are accept or not. While there is still a 113 pool of servers that don't repsond to EDNS requests, clients have no 114 way to know if the lack of response is due to packet loss, EDNS 115 packets not being supported or rate limiting due to the server being 116 under attack. Mis-classifications of server characteristics are 117 unavoidable when rate limiting is done. 119 2. Common queries class that result in non responses. 121 There are three common query classes that result in non responses 122 today. These are EDNS queries, queries for unknown (unallocated) or 123 unsupported types and filtering of TCP queries. 125 2.1. EDNS Queries - Version Independent 127 Identifying servers that fail to respond to EDNS queries can be done 128 by first identifying that the server responds to regular DNS queries 129 then making a series of otherwise identical responses using EDNS, 130 then making the original query again. A series of EDNS queries is 131 needed as at least one DNS implementation responds to the first EDNS 132 query with FORMERR but fails to respond to subsequent queries from 133 the same address for a period until a regular DNS query is made. The 134 EDNS query should specify a UDP buffer size of 512 bytes to avoid 135 false classification of not supporting EDNS due to response packet 136 size. 138 If the server responds to the first and last queries but fails to 139 respond to most or all of the EDNS queries it is probably faulty. 140 The test should be repeated a number of times to eliminate the 141 likelihood of a false positive due to packet loss. 143 Firewalls may also block larger EDNS responses but there is no easy 144 way to check authoritative servers to see if the firewall is 145 misconfigured. 147 2.2. EDNS Queries - Version Specific 149 Some servers respond correctly to EDNS version 0 queries but fail to 150 respond to EDNS queries with version numbers that are higher than 151 zero. Servers should respond with BADVERS to EDNS queries with 152 version numbers that they do not support. 154 Some servers respond correctly to EDNS version 0 queries but fail to 155 set QR=1 when responding to EDNS versions they do not support. Such 156 answer are discarded or treated as requests. 158 2.3. EDNS Options 160 Some servers fail to respond to EDNS queries with EDNS options set. 161 Unknown EDNS options are supposed to be ignored by the server 162 [RFC6891]. 164 2.4. EDNS Flags 166 Some servers fail to respond to EDNS queries with EDNS Flags set. 167 Server should ignore EDNS flags there do not understand and not add 168 them to the response [RFC6891]. 170 2.5. Unknown / Unsupported Type Queries 172 Identifying servers that fail to respond to unknown or unsupported 173 types can be done by making an initial DNS query for an A record, 174 making a number of queries for an unallocated type, them making a 175 query for an A record again. IANA maintains a registry of allocated 176 types. 178 If the server responds to the first and last queries but fails to 179 respond to the queries for the unallocated type it is probably 180 faulty. The test should be repeated a number of times to eliminate 181 the likely hood of a false positive due to packet loss. 183 2.6. TCP Queries 185 All DNS servers are supposed to respond to queries over TCP 186 [RFC5966]. Firewalls that drop TCP connection attempts rather that 187 resetting the connect attempt or send a ICMP/ICMPv6 administratively 188 prohibited message introduce excessive delays to the resolution 189 process. 191 Whether a server accepts TCP connections can be tested by first 192 checking that it responds to UDP queries to confirm that it is up and 193 operating then attempting the same query over TCP. An additional 194 query should be made over UDP if the TCP connection attempt fails to 195 confirm that the server under test is still operating. 197 3. Remediating 199 While the first step in remediating this problem is to get the 200 offending nameserver code corrected, there is a very long tail 201 problem with DNS servers in that it can often take over a decade 202 between the code being corrected and a nameserver being upgraded with 203 corrected code. With that in mind it is requested that TLD, and 204 other similar zone operators, take steps to identify and inform their 205 customers, directly or indirectly through registrars, that they are 206 running such servers and that the customers need to correct the 207 problem. 209 TLD operators should construct a list of servers child zones are 210 delegated to along with a delegated zone name. This name shall be 211 the query name used to test the server as it is supposed to exist. 213 For each server the TLD operator shall make an SOA query the 214 delegated zone name. This should result in the SOA record being 215 returned in the answer section. If the SOA record is not return but 216 some other response is returned this is a indication of a bad 217 delegation and the TLD operator should take whatever steps it 218 normally takes to rectify a bad delegation. If more that one zone is 219 delegated to the server it should choose another zone until it finds 220 a zone which responds correctly or it exhausts the list of zones 221 delegated to the server. 223 If the server fails to get a response to a SOA query the TLD operator 224 should make a A query as some nameservers fail to respond to SOA 225 queries but respond to A queries. If it gets no response to the A 226 query another delegated zone should be queried for as some 227 nameservers fail to respond to zones they are not configured for. If 228 subsequent queries find a responding zone all delegation to this 229 server need to be checked and rectified using the TLD's normal 230 procedures. 232 Having identified a working tuple the TLD 233 operator should now check that the server responds to EDNS, Unknown 234 Query Type and TCP tests as described above. If the TLD operator 235 finds that server fails any of the tests, the TLD operator shall take 236 steps to inform the operator of the server that they are running a 237 faulty nameserver and that they need to take steps to correct the 238 matter. The TLD operator shall also record the 239 for followup testing. 241 If repeated attempts to inform and get the customer to correct / 242 replace the faulty server are unsuccessful the TLD operator shall 243 remove all delegations to said server from the zone. 245 It will also be necessary for TLD operators to repeat the scans 246 periodically. It is recommended that this be performed monthly 247 backing off to bi-annually once the numbers of faulty servers found 248 drops off to less than 1 in 100000 servers tested. Follow up tests 249 for faulty servers still need to be performed monthly. 251 Some operators claim that they can't perform checks at registration 252 time. If a check is not performed at registration time it needs to 253 be performed within a week of registration in order to detect faulty 254 servers swiftly. 256 Checking of delegations by TLD operators should be nothing new as 257 they have been required from the very beginings of DNS to do this 258 [RFC1034]. Checking for compliance of nameserver operations should 259 just be a extension of such testing. 261 It is recommended that TLD operators setup a test web page which 262 performs the tests the TLD operator performs as part of their regular 263 audits to allow nameserver operators to test that they have correctly 264 fixed their servers. Such tests should be rate limited to avoid 265 these pages being a denial of service vector. 267 4. Firewalls and Load Balancers 269 Firewalls and load balancers can affect the externally visible 270 behaviour of a nameserver. Tests for conformance need to be done 271 from outside of any firewall so that the system as a whole is tested. 273 Firewalls and load balancers should not drop DNS packets that they 274 don't understand. They should either pass through the packets or 275 generate a appropriate error response. 277 Requests for unknown query types are not attacks and should not be 278 treated as such. 280 Requests with unassigned flags set (DNS or EDNS) are not attacks and 281 should not be treated as such. The behavaviour for unassigned is to 282 ignore them in the request and to not set them in the response. All 283 dropping DNS / EDNS packets with unassigned flags does is make it 284 harder to deploy extension that make use of them due to the need to 285 reconfigure / update firewalls. 287 Requests with unknown EDNS options are not a attack and should not be 288 treated as such. The correct behaviour for unknown EDNS options is 289 to ignore them. 291 Requests with unknown EDNS versions are not a attack and should not 292 be treated as such. The correct behaviour for unknown EDNS versions 293 is to return BADVERS along with the highest EDNS version the server 294 supports. All dropping EDNS packets does is break EDNS version 295 negotiation. 297 5. Response Code Selection 299 Choosing the correct response code when fixing a nameserver is 300 important. Just because a type is not implemented does not mean that 301 NOTIMP is the correct response code to return. Response codes need 302 to be choosen considering how clients will handle them. 304 For unimplemented opcodes NOTIMP is the expected response code. 306 In general, for unimplemented type codes Name Error (NXDOMAIN) and 307 NOERROR (no data) are the expected response codes. A server is not 308 supposed to serve a zone which contains unsupported types ([RFC1034]) 309 so the only thing left is return if the QNAME exists or not. NOTIMP 310 and REFUSED are not useful responses as they force the clients to try 311 all the authoritative servers for a zone looking for a server which 312 will answer the query. 314 Meta queries type may be the exception but these need to be thought 315 about on a case by case basis. 317 If you support EDNS and get a query with a unsupported EDNS version 318 the correct response is BADVERS [RFC6891]. 320 If you do not support EDNS at all FORMERR and NOTIMP are the expected 321 error codes. That said a mimimal EDNS server implementation just 322 requires parsing the OPT records and responding with a empty OPT 323 record. There is no need to interpret any EDNS options present in 324 the request as unsupported options are expected to be ignored 325 [RFC6891]. 327 6. Normative References 329 [RFC1034] Mockapetris, P., "Domain names - concepts and facilities", 330 STD 13, RFC 1034, November 1987. 332 [RFC1035] Mockapetris, P., "Domain names - implementation and 333 specification", STD 13, RFC 1035, November 1987. 335 [RFC5966] Bellis, R., "DNS Transport over TCP - Implementation 336 Requirements", RFC 5966, August 2010. 338 [RFC6891] Damas, J., Graff, M., and P. Vixie, "Extension Mechanisms 339 for DNS (EDNS(0))", STD 75, RFC 6891, April 2013. 341 Author's Address 343 M. Andrews 344 Internet Systems Consortium 345 950 Charter Street 346 Redwood City, CA 94063 347 US 349 Email: marka@isc.org