S Simplewala

Mailboxes, webmail and aliases

Day-to-day email operation: creating addresses, reading them, forwarding them, and keeping them inside their storage allowance. For the DNS records and deliverability side, see Email hosting.

Receiving works as soon as DNS is in place. Sending needs a relay. Your server hands outgoing mail to an SMTP relay rather than delivering it itself. Until one is configured, outgoing mail is held, not lost — why, and how to set one up.

Creating a mailbox

You need a verified domain and a server with the email platform installed.

  1. Portal → Emails → Create Email Account.
  2. Pick the domain, choose the part before the @, set a password of at least 12 characters, set a quota.
  3. Create. The mailbox is provisioned on the server in 10–30 seconds and moves from Pending to Active.

Connecting a mail client

SettingValue
IMAP servermail.yourdomain.com
IMAP port993, SSL/TLS
SMTP servermail.yourdomain.com
SMTP port587, STARTTLS
Usernamethe full address, not just the part before the @
Passwordthe mailbox password

IMAP only — POP3 is not offered. IMAP keeps the mailbox on the server, which is what makes the same mail readable from a phone, a laptop and webmail at once, and what makes it possible to back up.

Webmail

Roundcube is available in a browser with no client to configure. The URL is on the email account's detail page and normally looks like https://mail.yourdomain.com/webmail/. Sign in with the full address as the username.

If sign-in fails, the usual causes are the username entered without the domain, or a password that was changed in the portal but not remembered. Reset it at Emails → the account → Change Password.

Roundcube signs you out after a period of inactivity, and suspicious mail is filed into Junk automatically — worth checking when something expected has not arrived.

What happens when you press Send

Your server accepts the message and hands it to an SMTP relay. It does not deliver to the internet itself. That is deliberate: servers share IP reputation, and one bad sender on an address would get every other sender on it blocked.

So every address needs a relay, added in the portal under Server → Email → Sending: your own — SendGrid, Bird, Amazon SES, Mailgun, Postmark, SMTP2GO, or any SMTP relay you already pay for. Simplewala configures and verifies it, and charges nothing for sending through it.

A relay of our own is coming. Until it is here, bringing your own key is how an address sends — there is no sending plan to buy, and nothing in the portal will ask you for one.

Until a relay is in place, outgoing mail is held on your server rather than delivered. It is not lost, and it goes out the moment a relay is configured. Webmail says so when you press Send, the portal says so on the mailbox, and a notice lands in the sender's own inbox. Background on why this is how email works now: when port 25 is blocked.

Having a relay is necessary and not sufficient: the relay also has to be told which domains you may send as. That is the most common reason a correctly configured relay still refuses everything — see why a working relay can still refuse every message.

Why a working relay can still refuse every message

A relay makes two separate grants, and having one is not having the other. The API key answers who are you. A sender identity answers whose mail may you send — and without it you get:

550 The from address does not match a verified Sender Identity

This is not a SendGrid quirk. It is the only thing standing between a relay and being an open relay with a billing account attached, and SendGrid, Amazon SES, Mailgun, Postmark, Bird and SMTP2GO all enforce it. It is also why a key can pass every check — valid, correct permissions, a delivered test message — and still carry none of your mail.

The fix: authorise the domain

Server → Email → Sending → Sender identity. Publish the records the provider asks for (normally two or three DKIM CNAMEs), and from then on any address on that domain may send — including mailboxes you create next month. On SendGrid and on Mailgun-with-an-API-key the platform can create the authorisation for you; on SES and Postmark it gives you the console steps, because those APIs need a credential the platform deliberately does not hold.

The records appear on the DNS tab alongside MX and SPF, and they are counted separately and more bluntly — a missing MX record degrades delivery, a missing relay DKIM record stops outbound entirely.

The portal tells you the one thing to do

The Email tab carries a card naming the single next action for outbound, and nothing else:

  1. Add a relay — nothing can send until one exists.
  2. Authorise this domain — one click on SendGrid and on Mailgun with an API key; the console steps for the rest.
  3. Publish these DNS records — listed with a live/not-live column, checked against real DNS with Check DNS.
  4. Confirm — for providers the platform cannot poll, Mark as added at the relay once their console shows the domain verified.

Checking DNS and marking it added are different things on purpose. Records resolving means the records resolve; the provider still has to validate them. Only mark it added when the provider says so — if you mark it early, your addresses go back to sending as themselves and are refused with a 550 instead of going out via the fallback.

While that is outstanding: send via

Sometimes the DNS is somewhere you cannot reach. For those addresses there is a second option: send as an address the relay has authorised, and keep replies coming back to the real mailbox.

The recipient seesReplies go toNeeds
As itself alice@yourbusiness.com the same address yourbusiness.com authorised at the relay
Via "Alice (alice@yourbusiness.com)" <send@authorised.com> alice@yourbusiness.com any authorised address on that relay
Neither nothing — the relay refuses it — —

This happens automatically. While your domain is not authorised, addresses that would otherwise be refused are switched to send via an address that is — so your mail goes out rather than failing. The card says which addresses and via what; it is never silent.

And it reverses itself. The moment the domain is authorised — by a DNS check passing, by the provider verifying, or by you marking it added — those addresses go back to sending as themselves. You do not have to remember to undo it.

You can also set it per address under Where each address sends from → Change, or use Check every address to see the best route available to every mailbox at once.

What sending via costs you

It is a fallback, not a default, and the trade is worth stating plainly.

  • Your mail is delivered instead of refused. That is the entire point, and it is not a small thing.
  • The recipient sees a different address. Your real one is in the display name and in Reply-To, so a reply still reaches you, but the From is not yours.
  • Your domain builds no sending reputation. Authentication passes — for the authorised domain, not yours. Over months, that matters.

Which is why authorising your own domain is always the better answer where it is possible, and why the platform will never switch an address to via without being asked.

Reply-To on its own

Routing replies somewhere else has nothing to do with relays, and you can set it independently: leave the From alone and point Replies go to at a shared inbox. A common use is a noreply@ address whose replies land with the support team rather than in a mailbox nobody reads.

Both settings follow the same cascade as the relay choice — set them on the account, on a domain, or on one address, and the most specific wins.

How you find out a message did not send

A message in your Sent folder means the server accepted it. That is not the same as delivering it, and the difference used to be invisible. It no longer is — there are three places the truth shows up, and they say different things on purpose.

A notice in your own inbox

If a message is held — the server took it but has no relay that can carry it — you get a message in your inbox naming what was held, why, and what to do about it. Held mail is waiting, not lost; it goes out as soon as the address has a working relay.

If a message is delayed — delivery was attempted and failed — Postfix sends you a standard delay notice after about an hour, the same one any mail server sends. That one means delivery is being retried and will bounce if it keeps failing.

The Outbox panel

Server → Email → Sending → Outbox lists both, separately, because the fixes differ:

MeansWhat fixes it
Held The platform stopped it deliberately — the address has no relay that could carry it. Give the address a relay. The queue releases in one click.
Could not be delivered Delivery was attempted and failed. It is being retried and will bounce at the end. Fix the cause. Releasing held mail does not affect these.

Before you build the mail server

The readiness check on the deployment page now asks whether a relay exists at all, alongside RAM, ports and MX conflicts. A mail server with no relay can receive and can never send — that is a legitimate thing to build, and the check makes sure it is a choice rather than a surprise.

Sending from the portal

Each mailbox page has a Send a Message box, and it is the only place on the platform that can tell you what happened to a message while you are still looking at it. Webmail and your mail client both submit over SMTP and then file a copy — neither of them hears what the relay eventually did. When the portal sends, it hands the message to your server itself and reads the server's answer back:

Queued as 4A2B3C5D. Your copy is in Outbox and moves to Sent
once the relay confirms delivery.

If the address has no relay that can carry its mail, nothing is queued. You are told so on the spot, with what to add, and the message you typed stays in the box so you can send it once a relay is in place.

It is plain text, up to 25 recipients and 128 KB, rate limited to 60 messages an hour. For attachments, formatting or bulk mail, use webmail or a mail client.

Recent Sends

Underneath it, every recent message from that address and where each one got to: delivered, queued, retrying, bounced or held, with the queue reference your server assigned. It is read out of the mailbox itself every time you open it, so it cannot disagree with your mail — and it is the same state your mail client sees, because it is stored on the message rather than in our records. We keep no copy of your mail and no log of who you wrote to.

Aliases, forwarders and catch-alls

Three ways to receive mail at an address without paying for a second mailbox.

Delivers toUse for
Alias An existing mailbox on your domain. info@ landing in John's inbox. No separate storage, no separate password.
Forwarder An address outside your domain. Business mail arriving in someone's personal inbox.
Catch-all A mailbox you designate. Anything sent to an address that does not exist, instead of bouncing.

Create them at Emails → Aliases → Add Alias; they take effect immediately.

Forwarding has one caveat worth knowing before you rely on it: the receiving provider sees mail arriving from your server that was authored elsewhere, which is exactly what SPF is designed to distrust. Forwarded mail is therefore more likely to be filtered than mail sent directly. It is a property of email forwarding generally, not of this platform.

There is no hard limit on aliases — but if you are creating a dozen for a dozen different people, individual mailboxes are cleaner and each person gets their own password.

Autoresponders

Emails → the account → Autoresponder. Set a subject, a message, an active toggle and optionally a start and end date — the end date turns it off for you.

Two behaviours worth knowing so the results are not surprising:

  • Once per sender per day. Somebody who emails you five times gets one auto-reply, not five.
  • It will not reply to mailing lists or automated mail. It checks List-ID and similar headers first, which is what stops two autoresponders talking to each other forever.

The original message still arrives normally. The auto-reply is an extra, not a substitute.

Storage and quotas

Every mailbox has a quota. It counts everything on the server — inbox, sent, drafts, every folder, and attachments, which are almost always the reason a mailbox fills up. The default is 1 GB.

Usage is on the account's detail page in the portal, and in Roundcube under Settings. At 100% of quota, incoming mail is rejected with a bounce telling the sender the mailbox is full, and sending stops until space is freed.

Freeing space

  1. Delete large attachments — sort by size in your client.
  2. Empty Trash. Deleted mail stays there and still counts.
  3. Empty Junk, for the same reason.

The reported figure catches up within a few hours.

Raising the quota

Emails → the account → Change Quota, in megabytes, effective immediately. The quota is a limit, not an allocation — it cannot exceed what the server's disk can actually hold, so check Storage before raising several mailboxes at once.

Email backups

Mail data is backed up automatically — every mailbox, folder and message for the domain, as compressed archives, kept for the retention period (30 days by default). Nothing needs enabling. Emails → Backups lists them with date, size and status.

To restore, contact support with what you need back and roughly when it existed. One thing to be clear about when you ask: restoring replaces the current state of the mailbox, so mail received since the backup was taken can be lost. Say so if that matters and it can be planned around.

Backups coverBackups do not cover
All messages in every mailbox on the domainMail deleted before the last backup ran
Folder structureContacts and calendars
Every mailbox on the domain at onceMail that only ever existed on your device

Next