Closed ietf-svn-bot closed 3 years ago
@seth@sethblank.com changed status from new
to assigned
@seth@sethblank.com removed owner (was draft-ietf-dmarc-rfc7601bis@ietf.org
)
@seth@sethblank.com changed component from rfc7601bis
to dmarc-bis
@todd.herr@valimail.com set owner to todd.herr@valimail.com
@todd.herr@valimail.com changed status from assigned
to started
@todd.herr@valimail.com commented
Valimail data on April 13, 2021:
76830 DMARC policy records inspected
0 specifying message chunking
@todd.herr@valimail.com commented
Pushed to github and merged with main branch.
@todd.herr@valimail.com changed status from started
to infoneeded
@todd.herr@valimail.com changed status from infoneeded
to assigned
@todd.herr@valimail.com commented
Related to #50 and #71
@todd.herr@valimail.com changed status from assigned
to infoneeded
@mail@wander.science commented
From 152k DMARC records in tranco toplist:
3612 use ! in rua tag.
3053 of them are all pointing to mailto:dmarc@mailinblue.com!10m.
@mail@wander.science commented
Note that dmarcbis-01 still contains the syntax in the ABNF (6.4 Formal Definition).
@vesely@tana.it changed status from infoneeded
to assigned
@vesely@tana.it commented
Out of 119,920 domains, I found 26 different size limits:
MariaDB [mail]> select count(*) as c, dmarc_rua from domain where dmarc_rua rlike '!' group by dmarc_rua order by c desc; | # c | # dmarc_rua |
---|---|---|
105 | dmarc@mailinblue.com!10m |
|
2 | verisign@rua.agari.com,mailreports@verisign.com!10m |
|
1 | postmaster@cpcostruzioni.it!10m |
|
1 | dmarc-rua@t-2.net!1m |
|
1 | authfail@arissystem.com!8m |
|
1 | administrator@sunnyday.sk!1m |
|
1 | mx-dmarc@cpanel.net!10m |
|
1 | dmarc-rua@csirt.tfl.gov.uk!30m,dmarc-rua@dmarc.service.gov.uk |
|
1 | admin@dreamtilt.com.au!10m |
|
1 | dmarc@zumbi.com.ar!10m |
|
1 | dmarc@zaspy.com!10m |
|
1 | postmaster@pompo.co!2m |
|
1 | antispam@ac-orleans-tours.fr!5m |
|
1 | dmarc@mailinblue.com!10m,dmarc_rua@emaildefense.proofpoint.com |
|
1 | info@charismaedu.hk!10m |
|
1 | postmaster@rodier.me!10240 |
|
1 | DMARC-Admin@csiro.au!10m |
|
1 | dmarc-rua@siol.net!1m |
|
1 | dmarcagregate@firstbank.com!20m |
|
1 | aggrep@zorpia.com!10m |
|
1 | sam@gnubies.com!10m |
|
1 | admin@hellohotels.ro!30m |
|
1 | dmarc-rua@alerts.stux.fr!10m |
|
1 | dmarc@alliancecom.net!10m |
|
1 | abuse-dkim-rua@scissor.com!10m |
|
1 | dmarc_agg@cfigroup.com!10m |
26 rows in set (0.083 sec) —edited.
Besides, from the spec it is not clear at all from the spec whether a report should be sent in chunks (like Brandon said google does) rather than omitting to send if the size limit is exceeded.
@todd.herr@valimail.com changed status from assigned
to infoneeded
@smj@crash.com changed status from infoneeded
to assigned
@smj@crash.com commented
Looking at the ~2.9MM valid DMARC records from the dataset Farsight provided to DMARC.org through 1Q2021 with some quick and dirty regular expressions...
There were:
So that's a lot of current records that are trying to specify a size, but begs the question of whether any report generators implement the feature.
@todd.herr@valimail.com commented
Consensus from both the 27 May 2021 Interim (https://datatracker.ietf.org/doc/minutes-interim-2021-dmarc-01-202105270900/) and from discussion on the working group mailing list (https://mailarchive.ietf.org/arch/msg/dmarc/QgEEM1E8tMPXQ_oIer38RPxlCDU/) is to remove the ability to request message size chunking.
@todd.herr@valimail.com changed status from assigned
to closed
@todd.herr@valimail.com set resolution to fixed-consensus
keyword_nit_tag-update
owner:todd.herr@valimail.com
resolution_fixed-consensus
type_defect
| by seth@sethblank.comA dmarc-uri allows for message chunking (see https://tools.ietf.org/html/rfc7489#section-6.4) based on a maximum size specified in the rua/ruf URI.
No one uses this chunking mechanism, and further many dmarc validation systems have not implemented it, meaning no reports get sent at all to any domain that does specify a size limit. This has been a known bug in a major implementation for 5+ years, and no domain owner has noticed yet, because no one uses the feature. The size limit should be removed.
Issue migrated from trac:53 at 2022-01-24 16:17:02 +0000