This archive is retained to ensure existing URLs remain functional. It will not contain any emails sent to this mailing list after July 1, 2024. For all messages, including those sent before and after this date, please visit the new location of the archive at https://mailman.ripe.net/archives/list/[email protected]/
[dns-resolver-tf] DNS Resolver Recommendations
- Previous message (by thread): [dns-resolver-tf] Fwd: [ripe-list] DNS Resolver Recommendations Published (ripe-823)
- Next message (by thread): [dns-resolver-tf] DNS Resolver Recommendations
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Shane Kerr
shane at time-travellers.org
Wed May 1 17:09:21 CEST 2024
Hank, On 01/05/2024 15.27, Hank Nussbacher wrote: > > Under the section discussing Ingress Filtering you failed to discuss the > issue of fragment filtering. > > A very common and powerful DDoS attack is UDP fragment attack: > > https://ddos-guard.net/en/terms/ddos-attack-types/udp-fragmentation-flood > > The common thing many ISPs as well as enterprises do to mitigate the > attack is to block all fragments which on most servers has almost no > effect. But on DNS and VPN servers, blocking fragments is fatal and > therefore a warning needs to be put into the doc that UDP fragments > should *never* be blocked to DNS servers - even when under fragment > attack. See: > > https://puck.nether.net/pipermail/cisco-nsp/2023-December/108992.html > > for further details. Thanks for this! As mentioned in the thread there, using fragmentation avoidance should limit the need for fragments, which means blocking them should be basically okay. Fragmentation in DNS and how to avoid is is discussed in some detail in this IETF draft, which is referenced in the DNS Resovler Recommendations document: https://datatracker.ietf.org/doc/draft-ietf-dnsop-avoid-fragmentation/ Clients of resolvers will basically never send any large packets; although it is theoretically possible to build a valid query larger than 1232 bytes, in practice this is never seen. So no fragmented packets will arrive from there. Responses from authority servers should respect the EDNS0 buffer size and not fragment, although I suppose it is possible for some networks to have a smaller MTU than 1280 and want to fragment replies. In practice this should never happen either. So I think the right answer is to tune your DNS to avoid fragments, and then you can block them at will. 😄 IMHO, fragments in general are a badly designed and terribly insecure feature of IPv4 which was made worse when dragged into IPv6 and then made worse by removing the ability to fragment in the network itself. Blocking them seems like a good idea! Cheers, -- Shane -------------- next part -------------- A non-text attachment was scrubbed... Name: OpenPGP_0x3732979CF967B306.asc Type: application/pgp-keys Size: 11519 bytes Desc: OpenPGP public key URL: </ripe/mail/archives/dns-resolver-tf/attachments/20240501/c448ecbe/attachment-0001.bin> -------------- next part -------------- A non-text attachment was scrubbed... Name: OpenPGP_signature.asc Type: application/pgp-signature Size: 840 bytes Desc: OpenPGP digital signature URL: </ripe/mail/archives/dns-resolver-tf/attachments/20240501/c448ecbe/attachment-0001.sig>
- Previous message (by thread): [dns-resolver-tf] Fwd: [ripe-list] DNS Resolver Recommendations Published (ripe-823)
- Next message (by thread): [dns-resolver-tf] DNS Resolver Recommendations
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]