Re: [dmarc-ietf] 3.2.6 The meaning of non-existence (Sample Data)
If you prefix _domainkey to those names and do a lookup, several of them return NOERROR which suggests they have DKIM keys. Hm... one of them returns NXDOMAIN even though there is a DMARC record below. ale@pcale:~/tmp$ dig mail.foodnetwork.com ;; ->>HEADER<<- opcode: QUERY, status: NXDOMAIN, id: 64715 Sigh. That's just wrong. I'll see if I can find someone who can fix it. R's, John PS: pretty please can we not even think about changing the spec to work around other people's bugs ___ dmarc mailing list dmarc@ietf.org https://www.ietf.org/mailman/listinfo/dmarc
Re: [dmarc-ietf] 3.2.6 The meaning of non-existence (Sample Data)
On Mon 20/Dec/2021 20:59:45 +0100 John Levine wrote: It appears that Alessandro Vesely said: On Mon 20/Dec/2021 12:53:12 +0100 Douglas Foster wrote: I am not doing any root domain lookups. If that is part of the proposed algorithm, somebody needs to document it. I am simply looking for a resource record matching the FROM domain name. Oops, yes, you're right. Dunno why I looked up their org domain, probably lack of caffeine... Those 10 domains are non-existing under 3.2.6. Only 4 of them return NXDOMAIN. If you prefix _domainkey to those names and do a lookup, several of them return NOERROR which suggests they have DKIM keys. Hm... one of them returns NXDOMAIN even though there is a DMARC record below. ale@pcale:~/tmp$ dig mail.foodnetwork.com ; <<>> DiG 9.16.15-Debian <<>> mail.foodnetwork.com ;; global options: +cmd ;; Got answer: ;; ->>HEADER<<- opcode: QUERY, status: NXDOMAIN, id: 64715 ;; flags: qr rd ra; QUERY: 1, ANSWER: 0, AUTHORITY: 1, ADDITIONAL: 1 ;; OPT PSEUDOSECTION: ; EDNS: version: 0, flags:; udp: 1232 ; COOKIE: d56c07ed27e795d3010061c1bc09cea5581e05ff08ab (good) ;; QUESTION SECTION: ;mail.foodnetwork.com. IN A ;; AUTHORITY SECTION: foodnetwork.com.875 IN SOA ns-298.awsdns-37.com. awsdns-hostmaster.amazon.com. 1 7200 900 1209600 86400 ale@pcale:~/tmp$ dig _dmarc.mail.foodnetwork.com txt ; <<>> DiG 9.16.15-Debian <<>> _dmarc.mail.foodnetwork.com txt ;; global options: +cmd ;; Got answer: ;; ->>HEADER<<- opcode: QUERY, status: NOERROR, id: 32999 ;; flags: qr rd ra; QUERY: 1, ANSWER: 1, AUTHORITY: 0, ADDITIONAL: 1 ;; OPT PSEUDOSECTION: ; EDNS: version: 0, flags:; udp: 1232 ; COOKIE: 00298aa760a08142010061c1bc0e9fc192e7b9269cf7 (good) ;; QUESTION SECTION: ;_dmarc.mail.foodnetwork.com. IN TXT ;; ANSWER SECTION: _dmarc.mail.foodnetwork.com. 300 IN TXT "v=DMARC1; p=reject; fo=1; ri=3600; rua=mailto:discov...@rua.agari.com; ruf=mailto:discov...@ruf.agari.com"; For the umpteenth time, there is a DNS definition of non-existent which is the only one you get to use in the IETF. The definition in Section 3.2.6 is different. Can we stop wasting time with this fruitless argument, please? Yes. Best Ale -- ___ dmarc mailing list dmarc@ietf.org https://www.ietf.org/mailman/listinfo/dmarc
Re: [dmarc-ietf] 3.2.6 The meaning of non-existence (Sample Data)
Using an NXDOMAIN test, I have these occurrences in the last 24 hours: - 1 nxdomain that is from a legitimate sender, - 1 that is NXDOMAIN because the ESP misspelled the client organization name, and - 1 that is a bogus NDR. So the NXDOMAIN test will identify fewer problems, but will create fewer problems also. Caveat: Consider that my email stream is modest. Internet scale multiplies everything by many orders of magnitude. Does anyone have data from a bigger message stream? Doug On Sun, Dec 19, 2021 at 6:42 PM Douglas Foster < dougfoster.emailstanda...@gmail.com> wrote: > Here are some results based on 3025 messages, involving 1253 unique > RFC5322.From domains, collected over less than 24 hours. These results > are collected AFTER excluding messages from blacklisted sources and sources > with SPF=NXDOMAIN, so a high percentage is not spam. > > I detected 52 messages, from 10 unique domains, which failed the MX/A > test. I do not test on because I do not accept mail using IPv6. > > All 10 could produce DMARC PASS based on relaxed alignment, although I > have not evaluated whether they publish a DMARC policy. I simply evaluate > SPF and DKIM based on relaxed alignment for all incoming messages. > > All 10 domains had RFC5321.MailFrom and RFC5322.From domains that were > different. > > 7 of 10 had DMARC PASS based on both SPF and DKIM: > bc.qvcemail.com > doctors-digest.com > email.nutricia-na.com > mail.foodnetwork.com > mail.medscape.org > mktg.daily-harvest.com > email3.reachmd.com > > 1 of 10 had DMARC PASS based on SPF alignment only: > mg.homedepot.com > > 2 of 10 had DMARC PASS based on DKIM only: > info.extraspace.com > update.strava.com > > 1 of 10 had an SPF record on the RFC5322.From address. > email.nutricia-na.com > > Overall, this suggests to me that ESP messages will have trouble complying > with any NP criteria, and this may force us to use a weaker one, such as > NXDOMAIN only, even though my preference is a strong one. > > Doug Foster > ___ dmarc mailing list dmarc@ietf.org https://www.ietf.org/mailman/listinfo/dmarc
Re: [dmarc-ietf] 3.2.6 The meaning of non-existence (Sample Data)
It is not infrequent. Here is some more detailed statistics: msgs domains description Msg Fail Domain Fail 3,025 1,253 All messages1.72% 0.80% 1,098 296 Allowed msgs4.74% 3.38% 581 175 ESP messages8.95% 5.71% 52 10 MX/A Failures The percentages are relative to the failure counts. "ESP messages" counts any message where the From addresses are different. So 5.7% of third-party mailings use a From address that is not found or not easily found in DNS. This is not insignificant. John, I don't know what algorithm you are proposing. Please clarify. Doug On Mon, Dec 20, 2021 at 12:10 PM Alessandro Vesely wrote: > On Mon 20/Dec/2021 12:53:12 +0100 Douglas Foster wrote: > > I am not doing any root domain lookups. If that is part of the > proposed > > algorithm, somebody needs to document it. I am simply looking for a > resource > > record matching the FROM domain name. > > > Oops, yes, you're right. Dunno why I looked up their org domain, probably > lack > of caffeine... > > Those 10 domains are non-existing under 3.2.6. Only 4 of them return > NXDOMAIN. > > One of them, mail.foodnetwork.com, has its own DMARC record with a policy > different from that of its parent domain, which can be a reason to use a > subdomain. (Curiously, it is one of those returning NXDOMAIN.) > > For the other 9, all what I can think of is some kind of > misconfiguration. I > asked a few times why would one want to use a non-existing domain for the > From: > address, but got no answers. Anyway, your numbers show that it's not a > very > frequent setup. > > > Best > Ale > -- > > > > > > ___ dmarc mailing list dmarc@ietf.org https://www.ietf.org/mailman/listinfo/dmarc
Re: [dmarc-ietf] 3.2.6 The meaning of non-existence (Sample Data)
It appears that Alessandro Vesely said: >On Mon 20/Dec/2021 12:53:12 +0100 Douglas Foster wrote: >> I am not doing any root domain lookups. If that is part of the proposed >> algorithm, somebody needs to document it. I am simply looking for a >> resource >> record matching the FROM domain name. > > >Oops, yes, you're right. Dunno why I looked up their org domain, probably >lack >of caffeine... > >Those 10 domains are non-existing under 3.2.6. Only 4 of them return NXDOMAIN. If you prefix _domainkey to those names and do a lookup, several of them return NOERROR which suggests they have DKIM keys. For the umpteenth time, there is a DNS definition of non-existent which is the only one you get to use in the IETF. Can we stop wasting time with this fruitless argument, please? R's, John ___ dmarc mailing list dmarc@ietf.org https://www.ietf.org/mailman/listinfo/dmarc
Re: [dmarc-ietf] 3.2.6 The meaning of non-existence (Sample Data)
On Monday, December 20, 2021 12:10:31 PM EST Alessandro Vesely wrote: > On Mon 20/Dec/2021 12:53:12 +0100 Douglas Foster wrote: > > I am not doing any root domain lookups. If that is part of the proposed > > algorithm, somebody needs to document it. I am simply looking for a > > resource record matching the FROM domain name. > > Oops, yes, you're right. Dunno why I looked up their org domain, probably > lack of caffeine... > > Those 10 domains are non-existing under 3.2.6. Only 4 of them return > NXDOMAIN. > > One of them, mail.foodnetwork.com, has its own DMARC record with a policy > different from that of its parent domain, which can be a reason to use a > subdomain. (Curiously, it is one of those returning NXDOMAIN.) > > For the other 9, all what I can think of is some kind of misconfiguration. > I asked a few times why would one want to use a non-existing domain for the > From: address, but got no answers. Anyway, your numbers show that it's not > a very frequent setup. ... for legitimate mail. Scott K ___ dmarc mailing list dmarc@ietf.org https://www.ietf.org/mailman/listinfo/dmarc
Re: [dmarc-ietf] 3.2.6 The meaning of non-existence (Sample Data)
On Mon 20/Dec/2021 12:53:12 +0100 Douglas Foster wrote: I am not doing any root domain lookups. If that is part of the proposed algorithm, somebody needs to document it. I am simply looking for a resource record matching the FROM domain name. Oops, yes, you're right. Dunno why I looked up their org domain, probably lack of caffeine... Those 10 domains are non-existing under 3.2.6. Only 4 of them return NXDOMAIN. One of them, mail.foodnetwork.com, has its own DMARC record with a policy different from that of its parent domain, which can be a reason to use a subdomain. (Curiously, it is one of those returning NXDOMAIN.) For the other 9, all what I can think of is some kind of misconfiguration. I asked a few times why would one want to use a non-existing domain for the From: address, but got no answers. Anyway, your numbers show that it's not a very frequent setup. Best Ale -- ___ dmarc mailing list dmarc@ietf.org https://www.ietf.org/mailman/listinfo/dmarc
Re: [dmarc-ietf] 3.2.6 The meaning of non-existence (Sample Data)
I am not doing any root domain lookups. If that is part of the proposed algorithm, somebody needs to document it. I am simply looking for a resource record matching the FROM domain name. Because I am a Windows guy, I use the deprecated NSLOOKUP. I have done minimal work in DIG. I retested one of the names and confirmed the same results: > set type=MX > info.extraspace.com Server: G3100.myfiosgateway.com Address: 192.168.1.1 *** G3100.myfiosgateway.com can't find info.extraspace.com: Non-existent domain > set type=A > info.extraspace.com Server: G3100.myfiosgateway.com Address: 192.168.1.1 *** G3100.myfiosgateway.com can't find info.extraspace.com: Non-existent domain > Doug Foster On Mon, Dec 20, 2021 at 4:44 AM Alessandro Vesely wrote: > On Mon 20/Dec/2021 00:42:27 +0100 Douglas Foster wrote: > > > > I detected 52 messages, from 10 unique domains, which failed the MX/A > test. > > [...] > > > > 7 of 10 had DMARC PASS based on both SPF and DKIM: > > bc.qvcemail.com > > doctors-digest.com > > email.nutricia-na.com > > mail.foodnetwork.com > > mail.medscape.org > > mktg.daily-harvest.com > > email3.reachmd.com > > > > 1 of 10 had DMARC PASS based on SPF alignment only: > > mg.homedepot.com > > > > 2 of 10 had DMARC PASS based on DKIM only: > > info.extraspace.com > > update.strava.com > > > What do you mean by "failed the MX/A test"? Only doctors-digest.com > seems to be non-existent under 3.2.6. > > > ale@pcale:~/tmp$ for d in $doms mg.homedepot.com info.extraspace.com > update.strava.com; do r=$(get_root_domain $d|sed -rn 's/^ Root Domain: > *(.*)$/\1/p'); echo "$d -> $r"; dig +short $r; dig +short $r mx; echo; done > bc.qvcemail.com -> qvcemail.com > 167.140.19.203 > 100 smtp2.qvc.com. > 100 smtp3.qvc.com. > > doctors-digest.com -> doctors-digest.com > > email.nutricia-na.com -> nutricia-na.com > 52.36.54.191 > 20 mail3792.nutricianorthamerica.mkt4389.com. > 5 bounce.email.nutricia-na.com. > 10 reply.email.nutricia-na.com. > > mail.foodnetwork.com -> foodnetwork.com > 204.78.50.45 > 100 foodnetwork-com.mail.protection.outlook.com. > 1 aspmx.l.google.com. > 10 alt3.aspmx.l.google.com. > 5 alt1.aspmx.l.google.com. > 5 alt2.aspmx.l.google.com. > 10 alt4.aspmx.l.google.com. > > mail.medscape.org -> medscape.org > 104.18.27.226 > 104.18.26.226 > 10 alt3.aspmx.l.google.com. > 10 reply-mx.s6.exacttarget.com. > 1 aspmx.l.google.com. > 5 alt2.aspmx.l.google.com. > 5 alt1.aspmx.l.google.com. > 10 alt4.aspmx.l.google.com. > > mktg.daily-harvest.com -> daily-harvest.com > 104.18.1.9 > 104.18.0.9 > 10 alt4.aspmx.l.google.com. > 1 aspmx.l.google.com. > 10 alt3.aspmx.l.google.com. > 5 alt1.aspmx.l.google.com. > 5 alt2.aspmx.l.google.com. > > email3.reachmd.com -> reachmd.com > 34.195.222.240 > 34.233.81.108 > 10 mx1-us1.ppe-hosted.com. > 10 mx2-us1.ppe-hosted.com. > > mg.homedepot.com -> homedepot.com > 35.201.95.83 > 20 mx0a-000e6601.pphosted.com. > 10 mxb-000e6601.gslb.pphosted.com. > 10 mxa-000e6601.gslb.pphosted.com. > 20 mx0b-000e6601.pphosted.com. > > info.extraspace.com -> extraspace.com > 13.107.246.13 > 10 mxa-00257001.gslb.pphosted.com. > 10 mxb-00257001.gslb.pphosted.com. > > update.strava.com -> strava.com > 3.227.103.50 > 44.195.56.39 > 52.0.47.160 > 3.217.33.77 > 3.237.58.53 > 34.197.5.198 > 54.209.232.157 > 52.72.119.210 > 30 alt2.aspmx.l.google.com. > 50 aspmx3.googlemail.com. > 20 alt1.aspmx.l.google.com. > 10 aspmx.l.google.com. > 40 aspmx2.googlemail.com. > > > ___ dmarc mailing list dmarc@ietf.org https://www.ietf.org/mailman/listinfo/dmarc
Re: [dmarc-ietf] 3.2.6 The meaning of non-existence (Sample Data)
On Mon 20/Dec/2021 00:42:27 +0100 Douglas Foster wrote: I detected 52 messages, from 10 unique domains, which failed the MX/A test. [...] 7 of 10 had DMARC PASS based on both SPF and DKIM: bc.qvcemail.com doctors-digest.com email.nutricia-na.com mail.foodnetwork.com mail.medscape.org mktg.daily-harvest.com email3.reachmd.com 1 of 10 had DMARC PASS based on SPF alignment only: mg.homedepot.com 2 of 10 had DMARC PASS based on DKIM only: info.extraspace.com update.strava.com What do you mean by "failed the MX/A test"? Only doctors-digest.com seems to be non-existent under 3.2.6. ale@pcale:~/tmp$ for d in $doms mg.homedepot.com info.extraspace.com update.strava.com; do r=$(get_root_domain $d|sed -rn 's/^ Root Domain: *(.*)$/\1/p'); echo "$d -> $r"; dig +short $r; dig +short $r mx; echo; done bc.qvcemail.com -> qvcemail.com 167.140.19.203 100 smtp2.qvc.com. 100 smtp3.qvc.com. doctors-digest.com -> doctors-digest.com email.nutricia-na.com -> nutricia-na.com 52.36.54.191 20 mail3792.nutricianorthamerica.mkt4389.com. 5 bounce.email.nutricia-na.com. 10 reply.email.nutricia-na.com. mail.foodnetwork.com -> foodnetwork.com 204.78.50.45 100 foodnetwork-com.mail.protection.outlook.com. 1 aspmx.l.google.com. 10 alt3.aspmx.l.google.com. 5 alt1.aspmx.l.google.com. 5 alt2.aspmx.l.google.com. 10 alt4.aspmx.l.google.com. mail.medscape.org -> medscape.org 104.18.27.226 104.18.26.226 10 alt3.aspmx.l.google.com. 10 reply-mx.s6.exacttarget.com. 1 aspmx.l.google.com. 5 alt2.aspmx.l.google.com. 5 alt1.aspmx.l.google.com. 10 alt4.aspmx.l.google.com. mktg.daily-harvest.com -> daily-harvest.com 104.18.1.9 104.18.0.9 10 alt4.aspmx.l.google.com. 1 aspmx.l.google.com. 10 alt3.aspmx.l.google.com. 5 alt1.aspmx.l.google.com. 5 alt2.aspmx.l.google.com. email3.reachmd.com -> reachmd.com 34.195.222.240 34.233.81.108 10 mx1-us1.ppe-hosted.com. 10 mx2-us1.ppe-hosted.com. mg.homedepot.com -> homedepot.com 35.201.95.83 20 mx0a-000e6601.pphosted.com. 10 mxb-000e6601.gslb.pphosted.com. 10 mxa-000e6601.gslb.pphosted.com. 20 mx0b-000e6601.pphosted.com. info.extraspace.com -> extraspace.com 13.107.246.13 10 mxa-00257001.gslb.pphosted.com. 10 mxb-00257001.gslb.pphosted.com. update.strava.com -> strava.com 3.227.103.50 44.195.56.39 52.0.47.160 3.217.33.77 3.237.58.53 34.197.5.198 54.209.232.157 52.72.119.210 30 alt2.aspmx.l.google.com. 50 aspmx3.googlemail.com. 20 alt1.aspmx.l.google.com. 10 aspmx.l.google.com. 40 aspmx2.googlemail.com. ___ dmarc mailing list dmarc@ietf.org https://www.ietf.org/mailman/listinfo/dmarc
Re: [dmarc-ietf] 3.2.6 The meaning of non-existence (Sample Data)
Here are some results based on 3025 messages, involving 1253 unique RFC5322.From domains, collected over less than 24 hours. These results are collected AFTER excluding messages from blacklisted sources and sources with SPF=NXDOMAIN, so a high percentage is not spam. I detected 52 messages, from 10 unique domains, which failed the MX/A test. I do not test on because I do not accept mail using IPv6. All 10 could produce DMARC PASS based on relaxed alignment, although I have not evaluated whether they publish a DMARC policy. I simply evaluate SPF and DKIM based on relaxed alignment for all incoming messages. All 10 domains had RFC5321.MailFrom and RFC5322.From domains that were different. 7 of 10 had DMARC PASS based on both SPF and DKIM: bc.qvcemail.com doctors-digest.com email.nutricia-na.com mail.foodnetwork.com mail.medscape.org mktg.daily-harvest.com email3.reachmd.com 1 of 10 had DMARC PASS based on SPF alignment only: mg.homedepot.com 2 of 10 had DMARC PASS based on DKIM only: info.extraspace.com update.strava.com 1 of 10 had an SPF record on the RFC5322.From address. email.nutricia-na.com Overall, this suggests to me that ESP messages will have trouble complying with any NP criteria, and this may force us to use a weaker one, such as NXDOMAIN only, even though my preference is a strong one. Doug Foster ___ dmarc mailing list dmarc@ietf.org https://www.ietf.org/mailman/listinfo/dmarc