Need help in understanding auth digest-md5 and realm
Hello dovecot community,
I've setup dovecot and need a bit help in understanding the auth mechanism digest-md5 and realm
in 10-auth.conf I got
auth_mechanisms = plain login digest-md5 cram-md5 apop #auth_realms = #auth_default_realm =
So i got empty realms.
Auth normally works fine and clients can auth with mechanism digest-md5 and I see the following log entries:
dovecot: auth: Debug:
sql(user@temizbau.de,46.85.229.153,<klUjO3FcTy8uVeWZ>): Generating
DIGEST-MD5 from user 'user@temizbau.de', password 'xxxx'
dovecot: auth: Debug:
sql(user@gruene-wiesentheid.de,87.168.26.5,
But sometimes clients get a password mismatch and I the see the following log entries:
dovecot: auth: Debug:
sql(user@temizbau.de,80.187.103.15,<adzhAnVclmxQu2cP>): Generating
DIGEST-MD5 from user 'user@temizbau.de@mail.beckspaced.com', password 'xxxx'
dovecot: auth: Debug:
sql(user@thansadet.com,87.218.86.165,<LWItYHVc6r1X2lal>): Generating
DIGEST-MD5 from user 'user@thansadet.com@mail.beckspaced.com', password
'xxxxxxxxxx'
dovecot: auth: Debug:
sql(user@plaa-thansadetresort.com,110.164.127.146,
when there's a password mismatch I see a different user string for generating the digest-md5 hash. i suppose users use a different mail client and the mail client does things differently?
How can I fix this password mismatch thing?
Do i just need to set an auth_realms of some random string in the 10-auth.conifig Or does the auth_realms need to be a host name? Domain name of some sort?
For the moment I just removed the digest-md5 mechanism ... Or could I just simply not offer that mechanism?
If someone could shed some light on this I would be more than grateful ;)
Thanks & greetings Becki
On October 27, 2017 at 6:00 PM Admin Beckspaced admin@beckspaced.com wrote:
Hello dovecot community,
I've setup dovecot and need a bit help in understanding the auth mechanism digest-md5 and realm
in 10-auth.conf I got
auth_mechanisms = plain login digest-md5 cram-md5 apop #auth_realms = #auth_default_realm =
So i got empty realms.
Auth normally works fine and clients can auth with mechanism digest-md5 and I see the following log entries:
dovecot: auth: Debug: sql(user@temizbau.de,46.85.229.153,<klUjO3FcTy8uVeWZ>): Generating DIGEST-MD5 from user 'user@temizbau.de', password 'xxxx' dovecot: auth: Debug: sql(user@gruene-wiesentheid.de,87.168.26.5,
): Generating DIGEST-MD5 from user 'user@gruene-wiesentheid.de@', password 'xxxxxxxxxx' dovecot: auth: Debug: sql(user@vitaler-genuss.de,81.209.203.170,<tzxyT3FcT9RR0cuq>): Generating DIGEST-MD5 from user 'user@vitaler-genuss.de', password 'xxxxxxxxxxx' But sometimes clients get a password mismatch and I the see the following log entries:
dovecot: auth: Debug: sql(user@temizbau.de,80.187.103.15,<adzhAnVclmxQu2cP>): Generating DIGEST-MD5 from user 'user@temizbau.de@mail.beckspaced.com', password 'xxxx' dovecot: auth: Debug: sql(user@thansadet.com,87.218.86.165,<LWItYHVc6r1X2lal>): Generating DIGEST-MD5 from user 'user@thansadet.com@mail.beckspaced.com', password 'xxxxxxxxxx' dovecot: auth: Debug: sql(user@plaa-thansadetresort.com,110.164.127.146,
): Generating DIGEST-MD5 from user 'user@plaa-thansadetresort.com@imap.beckspaced.com', password 'xxxxxxxxxx' when there's a password mismatch I see a different user string for generating the digest-md5 hash. i suppose users use a different mail client and the mail client does things differently?
How can I fix this password mismatch thing?
Do i just need to set an auth_realms of some random string in the 10-auth.conifig Or does the auth_realms need to be a host name? Domain name of some sort?
For the moment I just removed the digest-md5 mechanism ... Or could I just simply not offer that mechanism?
If someone could shed some light on this I would be more than grateful ;)
Thanks & greetings Becki
We actually discovered that Android has a bug with DIGEST-MD5, which Google refuses to fix. Also DIGEST-MD5/CRAM-MD5 etc are not really good idea with SSL anyways.
Aki
On 27.10.2017 20:35, Aki Tuomi wrote:
On October 27, 2017 at 6:00 PM Admin Beckspaced admin@beckspaced.com wrote:
Hello dovecot community, ...
If someone could shed some light on this I would be more than grateful ;)
Thanks & greetings Becki We actually discovered that Android has a bug with DIGEST-MD5, which Google refuses to fix. Also DIGEST-MD5/CRAM-MD5 etc are not really good idea with SSL anyways.
Aki
Hello Aki, thanks for your reply ... so if there's a bug which Google won't fix it's perhaps best to not offer digest-md5? what do you mean by it's not a good idea to use DIGEST-MD5/CRAM-MD5 with SSL?
Thanks & Greetings Becki
Am 28.10.2017 um 08:30 schrieb Admin Beckspaced:
On 27.10.2017 20:35, Aki Tuomi wrote:
On October 27, 2017 at 6:00 PM Admin Beckspaced admin@beckspaced.com wrote:
Hello dovecot community, ...
If someone could shed some light on this I would be more than grateful ;)
Thanks & greetings Becki We actually discovered that Android has a bug with DIGEST-MD5, which Google refuses to fix. Also DIGEST-MD5/CRAM-MD5 etc are not really good idea with SSL anyways.
Aki
Hello Aki, thanks for your reply ... so if there's a bug which Google won't fix it's perhaps best to not offer digest-md5? what do you mean by it's not a good idea to use DIGEST-MD5/CRAM-MD5 with SSL?
Those methods encrypt the password itself which was a good thing back in the days when most connections were unencrypted. The disadvantage is that they require the password to be saved in cleartext.
If you can enforce an encrypted connection it is better to use PLAIN/LOGIN and save the passwords as hashes (preferably with salts).
-- Alex JOST
On 28.10.2017 12:24, Alex JOST wrote:
Am 28.10.2017 um 08:30 schrieb Admin Beckspaced:
On 27.10.2017 20:35, Aki Tuomi wrote:
On October 27, 2017 at 6:00 PM Admin Beckspaced admin@beckspaced.com wrote:
Hello dovecot community, ...
If someone could shed some light on this I would be more than grateful ;)
Thanks & greetings Becki We actually discovered that Android has a bug with DIGEST-MD5, which Google refuses to fix. Also DIGEST-MD5/CRAM-MD5 etc are not really good idea with SSL anyways.
Aki
Hello Aki, thanks for your reply ... so if there's a bug which Google won't fix it's perhaps best to not offer digest-md5? what do you mean by it's not a good idea to use DIGEST-MD5/CRAM-MD5 with SSL?
Those methods encrypt the password itself which was a good thing back in the days when most connections were unencrypted. The disadvantage is that they require the password to be saved in cleartext.
If you can enforce an encrypted connection it is better to use PLAIN/LOGIN and save the passwords as hashes (preferably with salts).
Thanks for your explanation ;)
On Fri, 27 Oct 2017 21:35:16 +0300 (EEST), Aki Tuomi stated:
We actually discovered that Android has a bug with DIGEST-MD5, which Google refuses to fix. Also DIGEST-MD5/CRAM-MD5 etc are not really good idea with SSL anyways
Could you actually describe what that bug is? I actually know someone at Google and they might be able to get it investigated and perhaps corrected. The more info you could supply, the better.
Thanks :)
-- Jerry
On October 28, 2017 at 11:37 AM Jerry jerry@seibercom.net wrote:
On Fri, 27 Oct 2017 21:35:16 +0300 (EEST), Aki Tuomi stated:
We actually discovered that Android has a bug with DIGEST-MD5, which Google refuses to fix. Also DIGEST-MD5/CRAM-MD5 etc are not really good idea with SSL anyways
Could you actually describe what that bug is? I actually know someone at Google and they might be able to get it investigated and perhaps corrected. The more info you could supply, the better.
Thanks :)
-- Jerry
The issue is https://issuetracker.google.com/issues/36996387, and exactly what happens is bit unknown. From our point of view, Android sends all other values correctly except final hash when using digest-md5.
Aki
participants (4)
-
Admin Beckspaced
-
Aki Tuomi
-
Alex JOST
-
Jerry