90 likes | 240 Views
DKIM Reporting. Murray S. Kucherawy <msk@sendmail.com>. The Problems to Solve. Senders want to know when their brands are being violated Mail being forged as coming from the sender Unsigned when DKIM policy is “all” or “strict”
E N D
DKIM Reporting Murray S. Kucherawy <msk@sendmail.com>
The Problems to Solve • Senders want to know when their brands are being violated • Mail being forged as coming from the sender • Unsigned when DKIM policy is “all” or “strict” • Unauthorized third-party signatures when DKIM policy is “strict” • Forged signatures, replay attacks • Compromised or revoked keys • Senders want to know when their mail is getting munged enroute • Such as by mailing list managers
The Problems to Solve • Senders want to know if what they’re doing doesn’t work • New hashes, algorithms that are not yet widely deployed • Implementors want to get back forensic evidence when verification failures occur • Especially important at the Interop event • Participants were asked to make sure their autoresponders returned forensic data on failures • Need to get back canonical forms seen at the verifier • To compare what the verifier got to what the signer signed • Pinpoints in-transit modifications
Issue(s) • Volume of mail generated by this can be big, of course • But the people making the request appear to know that and are willing to take the risk • Proposal will have to discuss this, possibly using deliberately scary language
Requirements • Based on consensus on ietf-dkim • Senders need to be able to indicate that they want to get these reports • Maybe break them down by type: unsigned, verification failures, expired signatures, syntax errors, general policy failures, other kinds? • Senders need to be able to specify where the reports should be sent • Senders need to be able to indicate what format(s) they want • ARF? INCH? Something else?
Proposal • A new draft which defines several extensions • Adds “r=“ tag to keys in DKIM base • Presence indicates reports are desired for failures involving that particular key; value indicates to where they should be mailed • Adds “r=“ tag to DKIM SSP • Presence indicates reports are desired for unsigned mail and policy failures; value indicates to where they should be mailed
Proposal • …also: • Adds “rf=“ tag to keys in DKIM base • Enumerates acceptable report formats (e.g. ARF, INCH, something else) • Maybe also need a way to request reports about specific types of failures that are of interest (see earlier slide) • Adds “rf=“ tag to DKIM SSP • Same deal
Proposal • …also: • Updates the current (-03) ARF draft, adding fields to the message/feedback-report format needed to provide DKIM verification failure forensics • Canonicalized headers • Maybe the canonicalized body • Maybe base64-encoded to ensure no changes in transit • Maybe what was retrieved in the key and/or SSP record as well, in case they change
Draft • draft-kucherawy-dkim-reporting • Ignore the one that’s up there • More of a placeholder • Was an idea under development • Needs examples • Wait for -01 • Will post it soon • Should probably become a DKIM WG draft after that