I'm currently trying to get my head around an issue with our Shop's email setup. So far we were using G Suite for all our emails related to our domain abc.com.
# The History: #
Our WooCommerce Shop was utilising the wp_mail() PHP-Mail functionality to send transactional emails (e.g. Order received, Account created, ..) by our Webhoster. As we were facing more and more emails being send to our customers SPAM folders, we decided to move to a SMTP provider.
# The new Setup: #
We installed the WP SMTP Mail Plugin for Wordpress and chose Sendgrid as SMTP provider. We followed the instructions and were able to connect all three: Our WooCommerce Shop, WP Mail SMTP and Sendgrid.
# The problem: #
Now that we have the new setup running, things seem to be somehow messed up. I read that it is not allowed / not good practice to mix the MX-record setting for abc.com. But this is exactly what I did - what I had to do - to make both, Sendgrid and G Suite function.
Despite the fact that Sendgrid lists outbound emails from shop#abc.com in its activity dashboard, the actual email is signed by Gmail's default DKIM domain key: d=*.gappssmtp.com, which shouldn't be the case, right?
# The objective: #
Use Sendgrid to send outbound emails from shop#abc.com
Use G Suite to receive inbound emails to shop#abc.com
Use G Suite to send outbound emails from sarah#abc.com, ...
# Open Questions: #
Which MX-record shall I set for my top-level domain "abc.com" to make the setup work?
How should the SPF/DKIM/DMARC setup look like for this use case?
Should shop#abc.com still be an active G Suite group / Mailing list or must this be removed?
Bonus: Is it possible to send outbound emails from "shop#abc.com" from both Sendgrid and G Suite?
Bonus: Is it correct to perform a Single Sender Verification with Sendgrid for shop#abc.com?
# Additional information: #
DNS/MX-records required by G Suite:
Which DNS-record shall I set for my top-level domain "abc.com" to make the setup work?
I assume you are talking about MX records here. MX records tell the outside world which server incoming email should be delivered to, so you probably want the G-Suite settings here. Note that if you have more than one server listed (in the screenshot you have 5) delivery will be attempted first to the server with the lowest priority value and will stop whenever a delivery is successful. If 2 records have the same priority, which one gets tried first is up to the sender.
How should the SPF/DKIM/DMARC setup look like for this use case?
DMARC: This is a TXT record at _dmarc.yourdomain.com. It tells the recipient weather all, some, or none of the outgoing emails should be SPF/DKIM authenticated, what should happen to unauthenticated email (p=reject or p=quarantine), and optionally who to tell about authentication failures. In your case I would just use the one from Sendgrid, since both the one from Sendgrid and the one from G-Suite probably specify that all emails should be authenticated, and Sendgrid is the only one that really needs to get notifications for failures.
You can ask for reports to be sent to multiple addresses, but everyone on that list will get reports about all failed deliveries (G-Suite will get notifications about Sendgrid and Sendgrid will get notifications about G-Suite) and the spec says the people you are emailing don't have to send responses to anyone beyond the first 2.
SPF: This is a anti-spam tool that tells email recipients which email servers are the real email servers for your domain. You can only have one SPF record, but you can merge them. Understanding the syntax will make merging SPF records a lot easier, but essentially do this:
Remove v=spf1 from the beginning of both records
Remove -all or ~all from both records
Put both records together (separated by a space)
Remove duplicate entries (for example both records are likely to contain a and mx)
consider +foo and foo to be duplicates
Put v=spf1 on the beginning of your new combined record (there should be a space between it and the rest of the record)
Put ~all on the end of your new combined record (there should be a space between it and the rest of the record) (you can use -all if you want to be more aggressive in not allowing emails through from servers impersonating your domain rather than just sending them to spam)
DKIM: These records are cryptographic keys that can be used to sign emails. You can have more than one of these, and each one has a unique name. When G-suite sends an email it includes a signature using it's key and also specifies that the signature should be checked against the key named google. If your domain is example.com this key should be at google._domainkey.example.com. Sendgrid's key will be named something else. Include both keys as separate records.
Should shop#abc.com still be an active G Suite group / Mailing list or must this be removed?
I would keep it. You will still be able to send mail from it if you want, but more importantly it will catch any replies customers send to your automated emails.
Is it possible to send outbound emails from "shop#abc.com" from both Sendgrid and G Suite?
Yes
Is it correct to perform a Single Sender Verification with Sendgrid for shop#abc.com?
Sorry, I can't help you there. Everything up to this point has been generic advice about using 2 email providers, but that seems to be a Sendgrid specific thing.
Related
I'm currently testing some complicated functionality in a web app that involves multiple accounts within the app; each with its own unique email address. I have a few temp gmail accounts that I'm using for this purpose, but keep having to update email addresses in the database to something else for some accounts in order to reuse email addresses. It gets meesy very quickly.
My question is; is there a way I can setup an email server on my VPS (Ubuntu 16.04) that is able to accept all email addresses on a given domain without me having to define each one in advance? Ie; I'd like to be able to make up email addresses as I go; test1#mydomain.com, test2#mydomain.com, etc, and have the server receive AND store any incoming emails to those addresses. Emails include account verifications, notifications, etc, and I need to know what is being sent to each participant at each point in each workflow as I test.
Ideally I'd like some way to view all of these emails on the domain regardless of address. Perhaps setting up an alias using some kind of wildcard so that all incoming email go to a single address that I can configure in my email client on my dev machine.
Consider rewriting envelope recipient x#subdomain.example.net to handler+x#example.net
AFAIR sendmail, postfix and exim support such rewrites. AFAIR all three support passing +detail as parameter to procmail script (to ~handler/.procmailrc as $1).
WARNING: test handling message to multiple special recipients.
Post which specific SMTP/MTA server do you use if you want more hints.
Gmail also supports plussed addresses BUT you get reliable hint about original recipient only for message to single special/plussed recipient.
After a lot of research and messing about with different configurations, I found the best solution for me was PostFix with Dovecot.
I was able to set up virtual mailboxes in PostFix with virtual_mailbox_maps pointing to a file that contained this; #mydomain.com mydomain.com/catchall/
This allows mail to any email address within the domain to go to one mailbox. The mydomain.com/catchall/ is a location on the server where the mailbox is located. By convention, the domain is used in the path. Useful if you are hosting email on multiple domains. Dovecot supports variables in the config which allows you to point it to the correct mailbox.
I also found that the PostFix main.cf file must have mydestination = localhost for this configuration to work.
the following posts helped me getting this up and running;
https://www.digitalocean.com/community/tutorials/how-to-set-up-a-postfix-e-mail-server-with-dovecot
https://www.digitalocean.com/community/tutorials/how-to-install-and-setup-postfix-on-ubuntu-14-04
I'm building an app to allow my users to send email. I want the email to originate from their domain. Currently, email is sent on behalf of my Mandrill account with their name/email used for the From header. It "works" but most of the mail is not delivering as best as I think it could.
The way I see it, one option is to use a service like Mandrill, Mailgun, Sendgrid, etc and have my users update there TXT records to verify their domains, thus allowing me to send on behalf of my users. Is that correct?
I'm wondering if another option would be to collect SMTP credentials, and then send the message via SMTP for my user, thereby preventing my user from having to log in and update their TXT records before using my app to send messages. I think it would be far easier to simply add SMTP credentials. Is this possible?
"The way I see it, one option is to use a service like Mandrill, Mailgun, Sendgrid, etc and have my users update there TXT records to verify their domains, thus allowing me to send on behalf of my users. Is that correct?"
Correct you'll want them to minimally have an SPF record that says the service you use is allowed to send email for the domain. I.e. TXT v=spf1 +a +mx inlcude:sendgrid.net ~all
"I'm wondering if another option would be to collect SMTP credentials, and then send the message via SMTP for my user, thereby preventing my user from having to log in and update their TXT records before using my app to send messages. I think it would be far easier to simply add SMTP credentials. Is this possible?"
Not really. They'll need to make sure their DNS records minimally have a valid SPF (TXT) record, otherwise the major email providers and players will either drop their messages or mark them as SPAM/junk.
I'm trying to understand a few concepts around sendgrid, whitelabeling and different servers that I plan to deploy the same sendgrid account in.
So my questions are:
1) Is whitelabelling purely for masking the via sendgrid.me and will I have any issues if I dont use it with my current setup(i.e. same account on several domains)
2) How does sendgrid deal with messages that have a "From" email that doesnt match the domain the email is sent from? Cause I read that it would silently drop them but instead I see that emails do get delivered however the statistics in sendgrid's dashboard are not being updated.
3) Upon creating a demo account I was asked to provide the domain from which the emails will be triggered but since I want to deploy this in several different domains will I need multiple accounts or is there an alternative option when you go for a paid plan?
Mike
Yes - whitelabelling will replace the sendgrid.me with your own domain.
You can also setup multiple domains inside SendGrid and assign each to a subuser. That will get you one SendGrid account, with multiple whitelabelled domains and separate sender reputation for each.
SendGrid will attempt delivery of whatever you ask it to send -- if you send an email with a different from domain than the signed sending domain, it's up to the receiving mail server to decide whether to block, flag as spam, or allow the email. Different receiving domains will behave differently.
It's generally best practice to always have your from domain match your signed sender domain.
Once you have one domain setup, you can setup additional domains using SendGrid subusers -- more info here.
If you have a complicated multi-domain setup, you might want to check out a templating API, like sendwithus, for making things easier to manage. They'll integrate directly with your SendGrid subusers on your behalf.
I have a PHP web app that is going to send out about 1,000 emails. I would love to test the performance beforehand. Is there any kind of service that provides dummy email addresses to send to, for this kind of testing? I can't find anything that's not just a general bulk-email service. The key here is I just want dummy addresses to send to.
If you have the ability to just purchase a domain name from a hosting service, I know at least 1&1 gives you like 2500 email addresses per domain so you could literally spam yourself to death and not worry about any other 3rd party. You can pick up a domain name for like
When you say "test the performance", do you mean you want to know about your deliverability rates, or how your emails look?
Deliverability Rates
This is entirely dependent on your SMTP server and the reputation of the IP that it will be sending from along with your domain's SPF records and the content of your email. To maximize this, I would recommend using a marketing email service such as MailChimp or MadMimi.
Appearance of Emails
You could always just send yourself a test email to see how it looks. An alternative is to use a service like PostageApp that has a built in template designer that has both an easy email preview function and a test send email function.
(Full Disclosure: I am the Product Manager of PostageApp.)
If you use "Post Hoc" you can send email to an unlimited number of email addresses. Post Hoc acts like an SMTP server, and receives the email messages that you are sending, but it does not forward them on anywhere. You do not need to set up any email inboxes ahead of time, so there is no problem if you have 1000 different unique email addresses. They do not need to be from a single domain -- you can use any email address you want. It stores the email messages received so that you can inspect them if necessary. You would run it locally so that there is no concern about network problems, and it is very low overhead since it does almost no processing of the email. This way, the performance measure will be mostly the sending side processing. Best of all, it is open source and freely available:
Find it on GitHub: GitHub for Post Hoc
Also see the blog post: PostHoc: Testing Apps that Send Email
Given multi-tenant application, How are vendors implementing email notifications from an email account setup and programming perspective:
Sending emails could come from a generic account: eg notifications#VendorName.com or noreply#VendorName.com, this seems reasonable considering reply addresses and lilnks can be contained within the email contents.
Receiving Emails: How would an application receive email, for instance; to generate support tickets or assign comments in an email to a project/task. I have seen ID's within the subject and some reply to addresses containing the account name eg: notifications#AccountName.VendorName.com
I realise one can programatically connect to a pop3 server and receive emails and look for the IDs with the subject, but is there a way of setting up and receiving email to a single pop3 account from multiple sub-host name email addresses (not sure on terminology there) eg: noreply#AccountName1.VendorName.com or noreply#AccountName2.VendorName.com and check the Account Name from the address? (similar to checking subdomains on a URL)
Any practices, experience, comments or sughestions?
(not sure its relevant, but using C# asp.net-mvc and services etc)
For sending notification emails, we have a notification send to address associated with each account and simply send from our domain to that address. Our from address is monitored and replies end up in the CSR work queue.
For inbound emails, we use FogBugz (from the makers of Stack Overflow) for case tracking. That accepts new cases via email (e.g. cases#mycompany.com). Tickets are auto-created from the email. My only complaint there is that the customer needs to check an obscure link for case updates (no "my cases" web portal, but maybe that will come out in an upcoming version of FogBugz).
We have a custom field in FogBugz to indicate the customer the ticket is from. We could theoretically write a plugin to FogBugz that auto-assigns that using the senders domain, but I guess the CSR's haven't complained loudly enough yet :-)
We (at muHive) are an inbound email/social conversations management product. If you are looking at a handling inbound email or social media conversations from customers, we have an impressive toolset.
For our own outbound needs, the simplest way is to use an Email sending API. Don't bother with SMTP sending by yourself. We use Amazon SES and have also tried Sendgrid which gave us additional benefits like delivery status and email parsing.
There are two ways in which you can handle multiple accounts to a catch all email address. If your target system can differentiate between different customers and assign tasks to the correct representatives based on either the content/sender, ask all your customers to send an email to support#company.com.
As you rightly said, you could also create *accountName_support#company.com* email addresses and use different accounts on whatever CRM/Support solution use to manage these emails.
Another approach is to have your customers send you an email to support#company.com and you use a rule based system (like muHive) to forward these mails to the appropriate account executives based on the customer/account who sent the mail.