When I send an email from my Apache JAMES SMTP server to my Gmail address, Gmail receives the email but apparently not via TLS encryption. Gmail shows the show red lock icon saying '[my server] did not encrypt this message.' The email headers also do not indicate receipt via TLS.
I have turned on TLS for my Apache JAMES SMTP server. Here is the relevant part from my smtpserver.xml:
<tls socketTLS="false" startTLS="true">
<keystore>[my-jks-file]</keystore>
<secret>[my-jks-password]</secret>
<provider>org.bouncycastle.jce.provider.BouncyCastleProvider</provider>
</tls>
I have also tried using
<tls socketTLS="true" startTLS="false">
instead but it did not fix the problem.
When I try testing my Apache JAMES email address using https://www.checktls.com/TestReceiver the report says everything (including TLS) passed. Testing my server domain using https://mxtoolbox.com/diagnostic.aspx also says TLS passed.
Why are the emails not being encrypted?
I too am having the same problem, trying to get emails pushed from email clients through Apache James onto Gmail, and Gmail is reporting that they have not been securely received.
From what I understand looking at Apache James documentation and source code, it is the RemoteDelivery mailet that is in charge of sending out emails to recipent email servers. https://james.apache.org/server/3/dev-provided-mailets.html
The doc and RemoteDelivery.java source code says RemoteDelivery contains the two properties:
startTLS: a Boolean (true/false) indicating whether the STARTTLS command (if supported by the server) to switch the connection to a TLS-protected connection before issuing any login commands. Default is false.
sslEnable: a Boolean (true/false) indicating whether to use SSL to connect and use the SSL port unless explicitly overridden. Default is false.
startTLS seems like the key here, where when James is negotiating a connection to Gmail's servers it should try to upgrade an insecure connection to a secure one if Gmail allows it.
I'm running Apache James 2.3.2 and in the /path/to/James/apps/james/SAR.INF/conf.xml file, under the Remove delivery section:
<mailet match="All" class="RemoteDelivery">
I tried setting setting properties to true
<startTLS>true</startTLS>
<sslEnable>true</sslEnable>
However James is still not sending secure emails to Gmail. So I'm out of ideas.
FYI in Apache James version 3.4 the RemoteDelivery configurations appears to be relocated to the conf/mailcontainer.xml file.
SocketTLS and startTLS only affect the encryption method of James inbound mail, and the encryption of mail sent from James, which affects the encryption of the gmail red lock, requires additional configuration. The test result mail is passed in Gmail: "security: Standard encryption (TLS)". My configuration(Apache James version:2.3.2) is like this:
<mail.smtp.port>25</mail.smtp.port>
<mail.smtp.starttls.enable>true</mail.smtp.starttls.enable>
<mail.smtp.auth>false</mail.smtp.auth>
<mail.smtps.auth>false</mail.smtps.auth>
<mail.smtp.socketFactory.port>25</mail.smtp.socketFactory.port>
<mail.smtp.socketFactory.fallback>false</mail.smtp.socketFactory.fallback>
<mail.smtps.ssl.protocols>TLSv1.2</mail.smtps.ssl.protocols>
<mail.transport.protocol>smtp</mail.transport.protocol>
<mail.smtps.port>465</mail.smtps.port>
<mail.smtps.ssl.enable>true</mail.smtps.ssl.enable>
<mail.smtps.starttls.enable>false</mail.smtps.starttls.enable>
<mail.smtps.socketFactory.port>465</mail.smtps.socketFactory.port>
<mail.smtps.socketFactory.class>javax.net.ssl.SSLSocketFactory</mail.smtps.socketFactory.class>
<mail.smtps.socketFactory.fallback>false</mail.smtps.socketFactory.fallback>
Just like setting properties in the java code in Christos's answer: Java mail cannot connect to smtp using tls or ssl.
According to above answers and/or comments, if anyone gets this problem in Apache James 3.7.0, you can try to seach for RemoteDelivery in mailetcontainer.xml and add:
<startTLS>true</startTLS>
Additionally, I currently do not know if it is necessary to add <sslEnable>true</sslEnable> while outgoing emails are encrypted and Gmail knows this without this setting. If you know this should be set up, you can explain why for anyone who needs to know. I do not set up this now because sending emails with <sslEnable>true</sslEnable> seems to be slower than this setting is removed (I feel this).
Related
Issue:
I have a working production process for initiating remote signing via the DocuSign REST API. For the most part, it works great. However, for the initial 20 contracts we've sent out, a few clients have not received their emails from DocuSign requesting their signature. I can confirm that DocuSign reports that they successfully SENT them to the signers even though clients have reported never receiving them (despite looking through spam/junk/deleted). In these few circumstances we're stuck at a dead-end because if they don't receive that email for whatever reason, we have to revert back to manual contracts.
Having read through the following article:
https://support.docusign.com/en/articles/Why-aren-t-my-signers-receiving-DocuSign-Notification-emails
It's clear that this is a common issue and the only options we have is to either send it to another email address, have the client reach out to their ISP and figure out why it might be rejected (This is not a good / professional option), or resort to a manual contract.
One thought I had...
was to use the API to obtain the link for the current signer that DocuSign provides in the emails they send and give my app's users an alternative option to send our branded email containing this link to the remote signing page from our email service since we have zero reputation issues.
Edit based on Inbar's reply
After reading through Larry's Blog as provided by Inbar in the comments below, my idea above would need to have my app send an email to the signer with a unique URL back to my App (so that this URL is Long-Lived / limited only based on business rules) which, when clicked, would make a request to EnvelopeViews:createRecipient (time-limited / must be used in 5 mins) to retrieve a url and redirect the user to the Remote Signing page which is governed by DocuSign's session policy.Should my client need to go away and click the link in their email again, the Long Lived ==> Time Limited & redirect would repeat starting a new session.
Support seems to be better here on SO than via ticketing, so I'm hoping a DocuSign rep/guru has some ideas on what I can do to handle these scenarios.
First off, yes there are cases where emails from DocuSign are rejected as SPAM by the ISP or by the email provider. These will not even show in the spam folder because they reject them before they send them to the end-user and we do need to work with these to fix the issue. Especially if these users will be receiving more than one contract etc. For that - you will need to contact support since we need to know the email address etc.
Your idea is possible, but you'll have to read Larry's blog here. This is because the URLs are typically very short lived and expire in 5 min so sending the regular URLs over email will not work very well.
Some emails sent by our sever go to spam for certain recipients. E.g when sent to #outlook.com email addresses.
I have been testing our emails using https://www.mail-tester.com
SpamAssassin gives a score of -0.1 for the issue of 'HEADER_FROM_DIFFERENT_DOMAINS From and EnvelopeFrom 2nd level mail domains are different'
This is the only issue reported. Everything else, such as SPF and DKIM, passes. There is no documentation on their website for this issue and I don't understand what it means or how to fix it.
The email is sent using PHPMailer via AWS SES.
I solved this by completing the setup of 'MAIL FROM' in AWS SES. This set the mail from header in emails to be my domain name.
"HEADER_FROM_DIFFERENT_DOMAINS" and mail-tester.com now gives me a score of 10/10.
Note that for some email providers it took a few weeks before we were trusted and emails stopped going to spam.
It means that the envelope sender address (which is used at the SMTP level) is different to the address used in the From header. This is very common, but can be a problem if you try to implement DMARC alignment. For example, you might send a message with a from header containing user#example.com, but your envelope sender might be bouncehandler#mail.example.com. You should be able to see this in the Return path header of a received message. Whether you can change this depends on exactly how you're sending your message, but in PHPMailer the envelope sender defaults to the from address, and you can override it by setting the Sender property.
A -0.1 penalty is unlikely to be the entire cause of your mail being sent to spam.
Its outlooks rubbish filtering system. They have "AI" rules that look at the sending ip address for reputation. They score you on user reportsand lots of other bits they will not tell you about. Make sure you have SPF, DMARC, DKIM, and sign up for their JMRP and SDNS they will tell you. But it still is a game of cat and mouse. Its a slippery slope and even Microsoft trap their own mail to their own outlook users. PITA, to be honest and luckily we managed to get a mitigation to the issue. However some users in different domains still complain of email going to JUNK. Go figure. I hate having to work on issues with Outlook.com. They themselves send out spam and have the audacity to block well configured SMTP senders.
I wish you luck. You will need it.
I have 2 Windows 2008 R2 boxes running in Microsoft Azure. My ASP.NET 4.0 site (let's imagine it's running at "example.com") has a standard Contact Us form.
When a user sends a Contact Us message, I use System.Net.Mail and SmtpDeliveryMethod.Network to deliver mail to an IIS6 SMTP server running on each box, which sends the mail to a Google Apps "enquiries#example.com" account, using the email address the user entered into the Contact Us form as the "From" address.
This was working beautifully for a year until I checked it today, and found this error in a .BDP file in the \Badmail folder:
550-5.7.1 Our system has detected an unusual rate of unsolicited mail originating from your IP address. To protect our users from spam, mail sent from your IP address has been blocked. Please visit http://www.google.com/mail/help/bulk_mail.html to review our Bulk Email Senders Guidelines.
Obviously Google upped their anti-spam strategies in the last 6 months - last time it worked was Feb 2013 (yeah, we don't get much mail luckily... yet).
I've read the Bulk Senders Guidelines linked above, but they're not really suited to my use case. My case is not sending emails from our server to users of our site (I simply use the Gmail API and send from our enquiries#example.com for that), but rather to collect users' enquiries so that we can easily respond by clicking Reply in that inbox.
I am looking for the easiest solution here. In response to the ones in Google's Bulk Senders Guidelines:
Use a consistent IP address to send bulk mail: I already do, doesn't seem to help
Reverse DNS: Godaddy, my domain and DNS provider doesn't seem to support them: http://support.godaddy.com/groups/domains-management-and-services/forum/topic/how-do-i-setup-reverse-dns/ Anyone know if there's a way?
Use the same address in the 'From:' header on every bulk mail you send: This is totally not my use case. I'll have different From headers in every email
SPF record: I think this only works if I am sending From ...#example.com every time. Is that right? My feeling is SPF doesn't help me here. Would love someone to enlighten me.
DKIM: This looks hellishly complicated, but I'll pursue it if someone thinks it can work in this case. Specifically is it OK that the From address doesn't match the "signing domain"? Anyone got any good "how to" links? And will this be sufficient for Google to un-blacklist me?
Sendgrid: Azure's preferred mail sending app. This means signing up, code changes, testing, and unknowns like "does Sendgrid allow any From address? It's non-trivial, and I'd like to avoid this, but again, will go there if it's what people think is the sanest option.
As a general answer to your questions, sending email on behalf of many different domains from one IP (e.g. example.net, example.org, and ex.co from 10.0.0.1) is generally seen as spammy behavior (and therefor not recommended).
Your points 1-5 only apply if you're sending from one domain. rDNS, SPF, and DKIM only improve delivery for one IP to one domain (in a generally 1:1) relationship.
Generally, the best way to avoid getting marked as spam in a situation like this is to set the From email as a consistent one that you actually control (e.g. enquery-sender#example.com), and then setting the Reply-To as the entered address (e.g. enquirer#someprovider.com). This way you consistently send from one domain, while still getting the benefit of replies going to the message originator (for example LinkedIn does it this way). Doing this will allow you to setup rDNS, SPF, and DKIM with benefit.
That said, if you decide that you don't want to use the recommended Reply-To method, you can use SendGrid to send from any arbitrary domain. It should not require any significant code change (just switching your current SMTP credentials to SendGrid's).
Disclaimer: I am a SendGrid employee.
I'm new to SMTP techniques so with just a few days work with it, I have this question - Can a mail-sending program send mails under my email?
It seems like I can put any email in to from field and if I put my friend's email there, I can disguise him!? This sound strange since I believe that being some one else is not that ease at all.
Please guide me if you have experiences on this.
Any email client can send an email using ANY email address as the from field.
That said, a lot of receiving mail servers are configured to do various tests to ensure that the email is coming from a real mail server.
For a list of those techniques used by receiving mail servers go here: http://en.wikipedia.org/wiki/Anti-spam_techniques Specifically the "automated techniques for e-mail administrators" section.
A short list from there is:
1. Reverse DNS - tests the IP address of the sender to ensure that the IP is listed as an MX record for the domain.
2. FCrDNS - receiving mail server will attempt to do an SMTP HELO or EHLO command back to the sender's IP. This makes sure the sender is an SMTP server.
3. Disallow Dynamic IPs - receiving server will test the sender's IP to see if it is DHCP'd. If so, then the mail is marked as spam.
Point is, you can send email as being originated from any email address. However, there is a huge possibility that the email will simply be deleted by the receiving server.
Actually, you can do what you say with an email client like Thunderbird where you can set the From address freely.
Anyway, you will need a SMTP server that will accept this any domain address (your provider SMTP will probably do).
That's also why some SMTP servers (like Gmail's one for instance) force the user to get athenticated and displays his user adress (it will sometimes display on behalf of when sending from another e-mail address. This is very well explained on Google site.
Regards,
Max
How do you stop the "Default SMTP Virtual Server" from sending bounce messages for email addresses that you don't have?
i.e. I'm using IIS' SMTP server to handle my email and if an email is sent unknown at mydomain.com a bounce email with 'address not known' (or something like that) is sent back to the sender. I want it to silently fail.
I found this article which has a script you can run to configure a catch-all account on your server. All emails which would generate a NDR will instead be directed to this account. Sorry, I haven't tested it. The article above has been removed here it is via the WayBack Machine
Basically the short answer to your question is no.
On another note, if you don't want to spend any money, or have no budget, and want a better email system, try something like Smarter Mail which you can use for free up to 10 users. I am sure there are others out there, but I have used Smarter Mail in the past successfully.
This isn't an IIS failure. The SMTP server receiving the message is looking for a valid email address, and when it doesn't find one, sends an email back to your email address saying that there isn't one there. The only way to have it silently fail is by putting the from address as a bogus email like no-reply#company.com, etc.
From an SMTP point of view, a better way to handle this is to reject the RCPT request at some point during the SMTP transaction. This way, your server isn't responsible for sending any blowback to the alleged sender.
I don't know how to configure IIS to do this specifically, but you certainly can with Postfix (which is what I use).