Help Start
Your AWS account
SendForever runs Amazon SES in your own AWS account. It reaches the account through an IAM role that a CloudFormation stack creates there. This page covers how to connect the account, who updates the stack, what the stack creates, what SendForever can and cannot do with it, the AWS account settings page, updating the stack, fixing a failed stack, disconnecting, and what AWS bills you.
What connecting does
Each team connects one AWS account in one Amazon SES region. You create a CloudFormation stack in that account from a link that SendForever fills in. The stack creates:
- An IAM role. Only one IAM user in SendForever's AWS account, the one SendForever runs as, can assume it, and only with an external ID unique to your team. Before version 32, the role trusted that whole account, so update a stack that is still on an older version.
- An S3 bucket, an Amazon Data Firehose stream, an AWS Glue table and an Athena workgroup. They store and query the activity log of every email, inside your account.
- An SNS topic and CloudWatch alarms that alert your team when your sending reputation drops.
- The roles and the schedule group that run your campaigns inside your account.
SendForever stores the role's ARN, never access keys to your account. Each session it opens with the role lasts at most an hour. Your contacts, templates and activity log stay in your AWS account, and Amazon SES bills that account directly for what you send.
Before you start
- Only team owners and admins can connect an AWS account. A member sees "Only team owners and admins can connect an AWS account."
- You need an AWS account. AWS asks for a card and a phone number when you create one, and a new account can take up to a day to be ready.
- The person who creates the stack needs permission to create CloudFormation stacks and IAM roles. The root user and administrators have it, and so does any other IAM user with those permissions.
- Each team needs its own AWS account. If the account already has a SendForever stack from another team, connect a different AWS account.
You connect the account on the first step of Getting started, or later from the team's Settings, under AWS account.
Say whether you already send with Amazon SES
The card first asks Do you already send email with Amazon SES? Until you answer, Connect with AWS opens nothing and the card says "Say whether you already send email with Amazon SES."
- No AWS account yet: the card shows Create an AWS account, which opens AWS's sign-up page. AWS asks for a card and a phone number, and a new account can take up to a day to be ready. Come back once it is ready and choose Not yet.
- Not yet: SendForever sets up Amazon SES in the region nearest your browser's time zone, of the regions every AWS account has from the start, never one AWS asks you to enable first. For example, Portugal, Spain and France get Europe (Paris), and Germany gets Europe (Frankfurt). A time zone that names no place, such as UTC, gets Europe (Ireland). The card does not ask you for a region.
- Yes: choose the region you already send from in Region you send from. Production access does not carry over to other regions, so any other region would start in the Amazon SES sandbox. The list shows each region where SendForever runs Amazon SES by name and code, for example "Europe (Ireland) — eu-west-1". A team that connected before finds its earlier region chosen. Effect on your current sending explains what connecting changes in an account that already sends (see below).
You can change your answer until the stack reports back. If AWS opens CloudFormation in another region, for example because your AWS account only allows one, create the stack there: SendForever moves the team to the region the stack was created in, as long as the team has no domains yet. Once the account is connected, it stays in its region.
Choose who updates the stack
New features may need a new version of the stack. Under Who updates the stack, SendForever comes chosen:
- SendForever: SendForever applies each new version when you ask, within limits it can never change. See From SendForever.
- By hand in AWS: SendForever can never change the stack. It shows the steps to apply each new version in AWS. See By hand.
The stack stores the choice in its StackUpdates parameter, which the link fills in as SendForever or ByHand, and enforces it. With ByHand, the StackUpdatesAccess policy only lets SendForever read the stack, so SendForever cannot prepare or apply a change to it. Every update SendForever applies leaves the parameter as it is. Only you can change it, by updating the stack in AWS (see Change who updates the stack).
Create the stack
- Answer whether you already send with Amazon SES, and change who updates the stack if you want to, as above.
- Click Connect with AWS. A new tab opens CloudFormation's quick-create page in the region, with the template and its parameters filled in. If AWS asks you to sign in, sign in to the AWS account you want to connect, as the root user or an administrator. If you use IAM Identity Center, sign in through your access portal, open the account there, then click Connect with AWS again.
- Leave every parameter as it is. At the bottom, tick "I acknowledge that AWS CloudFormation might create IAM resources with custom names" and choose "Create stack".
- Come back to SendForever. AWS takes a few minutes to create the stack, and the page connects on its own about a minute after that. If it takes longer, the page says "We have not heard from the stack yet. If it was created, paste its role below."
The stack is named sendforever. Its parameters TrustedAccountId, TrustedUserName, ExternalId, CallbackUrl and CallbackToken each say "Do not change". StackUpdates holds your choice of who updates the stack. When the stack is created, a small Lambda function reports the new role to SendForever. The page then shows Verifying access to your AWS account… while it checks that the role works. A new role can take up to two minutes to become usable, and the page keeps checking on its own. Once AWS accepts the role, the account shows as connected.
If the page does not connect
While SendForever waits for the stack, the page shows Troubleshooting:
- Landed on an AWS sign-in page? Sign in to the account you want to connect, as the root user or an administrator, then choose Connect with AWS again. Any other IAM user must be allowed to create CloudFormation stacks and IAM roles.
- The stack failed or rolled back? In CloudFormation, the stack's "Events" tab names the resource that failed and why. Delete the stack, then choose Connect with AWS again.
- AWS says the stack already exists? The account already has a stack named sendforever. The page shows your team's ExternalId, with a copy button. If the existing stack's "ExternalId" parameter matches, the stack belongs to this team. Delete it in CloudFormation and create it again. If it does not match, another team may still use it, so connect a different AWS account.
- AWS says a role or policy already exists? IAM allows each of the stack's fixed names only once per AWS account, across all regions. If several names conflict, the account has another sendforever stack, possibly in another region. Connect a different AWS account, or delete that stack if nothing uses it. If only sendforever-sending-boundary conflicts, see If only the sending boundary already exists below.
- Stack created, but nothing happened? Open the stack in CloudFormation and go to the Outputs tab. Copy the RoleArn value, paste it in Role ARN and click Verify role.
If SendForever has not heard from the stack after a few minutes, the page says so and opens the Role ARN field for you. After an hour with no word from the stack, Getting started shows the step as not started again. Click Connect with AWS again, or paste the role if the stack exists.
If you create the stack in a different region and the team already has domains, the connection fails and names both regions. Delete that stack and create it again from the link on the page, signed in to the right region. The same happens in a region where SendForever does not run Amazon SES.
If only the sending boundary already exists
Each sending user that SendForever creates has the sendforever-sending-boundary policy as its permissions boundary. IAM does not delete a policy that is in use, so the policy can outlive the stack. A new stack then fails on it with "Resource name conflict". The page shows these commands too.
-
In the AWS console, open CloudShell and run this command. It removes the policy from the sending users and deletes it. Their keys keep working, and their own policy still only lets them send as their domain.
P=arn:aws:iam::$(aws sts get-caller-identity --query Account --output text):policy/sendforever-sending-boundary; for U in $(aws iam list-entities-for-policy --policy-arn $P --policy-usage-filter PermissionsBoundary --query 'PolicyUsers[].UserName' --output text); do aws iam delete-user-permissions-boundary --user-name "$U"; done; for V in $(aws iam list-policy-versions --policy-arn $P --query 'Versions[?!IsDefaultVersion].VersionId' --output text); do aws iam delete-policy-version --policy-arn $P --version-id "$V"; done; aws iam delete-policy --policy-arn $P -
Delete the failed stack in CloudFormation, choose Connect with AWS again and wait until the stack is created. The new stack creates the policy again.
-
Run this command to give the policy back to every sending user under /sendforever/. Until you run it, the AWS account page lists the sending users that are missing it, with the same command and Check again:
P=arn:aws:iam::$(aws sts get-caller-identity --query Account --output text):policy/sendforever-sending-boundary; for U in $(aws iam list-users --path-prefix /sendforever/ --query 'Users[].UserName' --output text); do aws iam put-user-permissions-boundary --user-name "$U" --permissions-boundary $P; done
What the stack creates
These resources are listed on the stack's Resources tab in CloudFormation. Some have fixed names. CloudFormation names the others from the stack's name and their logical ID, such as sendforever-AccessRole-….
| Resource | Name in your account | What it is for |
|---|---|---|
| IAM role | sendforever-AccessRole-… | The role SendForever assumes. |
| IAM managed policies | SesAccess, ActivityLogAccess, CampaignsAccess, TrackingDomainAccess, ContactsAccess, StackUpdatesAccess (logical IDs), and the policy SendingUsersAccess | What the access role may do. |
| IAM managed policies | sendforever-ceiling, sendforever-ceiling-services | The ceilings: the most that the access role, and the roles AWS services act as, can ever do. Each role of the stack has one as its permissions boundary. SendForever can never change them. |
| IAM managed policy | sendforever-sending-boundary | The permissions boundary of every sending user SendForever creates. It allows sending email, and nothing more. Before version 32, it also left room for adding contacts to an Amazon SES contact list, which audiences no longer use. |
| S3 bucket | DataBucket (logical ID), named in the DataBucketName output | The activity log, campaign files, contact files to import, the images in your emails and your BIMI logo. |
| Amazon Data Firehose stream | sendforever-events | Carries Amazon SES email events to the bucket. |
| IAM roles | sendforever-events-publish, and EventsDeliveryRole (logical ID) | Let Amazon SES write to the stream, and the stream write to the bucket. |
| AWS Glue database and table | sendforever, with the table events | Describe the activity log for Athena. |
| Athena workgroup | sendforever | Runs SendForever's queries of the activity log. |
| SNS topic | sendforever-alerts | Receives reputation alerts. SendForever subscribes to it. |
| CloudWatch alarms | sendforever-bounce-rate, sendforever-complaint-rate | Fire when the account's bounce rate reaches 5% or its complaint rate reaches 0.1%. |
| EventBridge rules | sendforever-tenant-status, sendforever-tenant-findings | Tell the topic when Amazon SES pauses an isolated domain or finds a problem with one. |
| IAM roles | sendforever-campaigns, sendforever-campaign-schedules | What a campaign workflow may do, and what lets EventBridge Scheduler start one. |
| IAM policies | EventsDeliveryAccess, EventsPublishAccess, CampaignWorkflowsAccess, CampaignSchedulesAccess, ContactWorkflowsAccess, PreferencesFunctionAccess (logical IDs) | What each of those roles may do. They are kept separate from the roles, so an update never changes a role itself. |
| EventBridge Scheduler schedule group | sendforever | Holds scheduled campaigns. |
| IAM role | sendforever-contacts | What a contacts workflow may do: maintain the contacts table, read contact files to import, add unsubscribes to the activity log and post subscription changes to the domain's webhook. |
| Lambda function | sendforever-preferences, with its role, its log group and a function URL | Serves the page where people unsubscribe and choose their audiences, and the API your app uses to add people, on each domain's links host. It only answers requests that come through a links host, and hands each one to a workflow. |
| Lambda function | CallbackFunction (logical ID), with its role | Reports the role to SendForever when the stack is created, updated or deleted. |
The stack's Outputs are RoleArn, DataBucketName, PreferencesUrl, SendingPermissionsBoundaryArn, StackUpdates and TemplateVersion.
The bucket is encrypted and blocks all public access. It stays in your account when the stack is deleted. It deletes its own old files: the activity log and campaign results after 90 days, contact files to import after 7 days, records Firehose could not deliver after 30 days, and query results after a day.
As you use SendForever, it creates more resources in your account through the role. They are named after sendforever too, and they are not part of the stack:
- A configuration set for each domain, named sendforever- followed by the domain with dashes.
- A DynamoDB table for your contacts and audiences.
- Step Functions state machines for campaigns, subscriptions, the page where people unsubscribe, the subscriptions API, contact imports and audience counts.
- A CloudFront distribution for images and logos.
- Certificates and a distribution for each links host.
- IAM users under the path /sendforever/ for sending credentials.
- An SNS topic for each domain's webhook.
- A Mail Manager archive for kept copies.
- Amazon SES tenants for isolated domains.
What SendForever can and cannot do
The public page How SendForever reaches your AWS account, at sendforever.net/security, is written for whoever approves the access. The card that connects the account links to it as How access works. It links to the CloudFormation template, so you can read the template before you create the stack. AWS also lists everything the stack creates before you confirm.
The role can:
| Area | What it can do |
|---|---|
| Amazon SES | Set up and check domains, configuration sets, templates, the suppression list and tenants. Change the account settings you turn on in SendForever. Send email for your tests and campaigns. |
| Activity log | Query the events of your email in the S3 bucket the stack creates, with Athena. |
| Campaigns | Run the Step Functions state machines and EventBridge Scheduler schedules named sendforever that send your campaigns, from your account. |
| Contacts | Create and use one DynamoDB table, sendforever-contacts, for your contacts and audiences. Run the workflows that change subscriptions. It can empty the table, which managing contacts requires, but it can never delete the table itself. |
| The stack | Read the stack that created the role, and apply new versions of it that SendForever publishes. It only applies SendForever's own template, which it checks first, and only within the ceiling the stack puts on every role it creates. With StackUpdates set to ByHand, it can only read the stack. |
| Images and links | Create CloudFront distributions and ACM certificates for the images in your email, your BIMI logo and your links hosts. Each one gets the tag sendforever:workspace as it is made. From version 29, the role changes or deletes only the ones that carry that tag. From version 26, delete an image you remove from the bucket. |
| Alerts | Manage the SNS topics and CloudWatch alarms named sendforever. |
| Sending credentials | Create IAM users under the path /sendforever/. A permissions boundary limits each one to sending email. |
The role cannot:
- Reach other AWS services such as EC2 or RDS, DynamoDB tables other than sendforever-contacts, stacks other than its own, S3 buckets other than the one the stack creates, or anything in another AWS account.
- Create IAM roles, or change who may assume them or their permissions boundaries. Create IAM users outside /sendforever/, or with more than their boundary allows. From version 23, the one exception is Amazon SES's own service-linked role,
AWSServiceRoleForAmazonSES. The role can create it the first time a feature such as kept copies needs it. AWS makes this role once per account, and only Amazon SES can use it. - From version 29, change or delete a CloudFront distribution or an ACM certificate it did not make, or put its tag on one of yours. AWS only lets it tag what it is making, or what already carries the tag.
- Read or change anything in Route 53, from version 28. Version 27 let it add, change or remove TXT, CNAME, MX and CAA records in your hosted zones, so update a stack that is still on it. From version 30, the ceiling of your stack has no room for Route 53 either, so no later version can give it back.
- Go beyond its ceiling, or change it. If a new version of the stack needs more, you apply it yourself in the AWS console.
- Let another AWS account send as your domains, receive your email, or reach your bucket, your table or the alerts topic.
AWS does not let a role be limited in two places. Its Amazon SES permissions cover the whole account in the region, including the domains your app sends from, though SendForever only touches the domains you add to it. Before version 29, its CloudFront and ACM permissions also reached every distribution and certificate in the account, so update a stack that is still on an older version. From version 30, the ceiling holds the same limit: only what carries the role's tag. You apply version 30 by hand, because it changes the ceiling.
What the ceiling cannot stop: within it, a version SendForever publishes can do anything the role could already do. For example, it could send what SendForever can read to an address of its choosing. Also, the role is removed with the stack only because the template does not keep it. SendForever's templates never keep it, and after you delete the stack you can check that no role named sendforever-AccessRole-… is left.
SendForever can read who you sent email to and the subject line, from the activity log in your account. It reads the content of your email only in your templates, and in the copies Mail Manager keeps for a domain where you turn on kept copies. It reads them when someone on your team opens a page that shows them, and does not store them.
If you already send with Amazon SES
What stays as it is:
- Your app keeps sending as it does today, with the same domains and credentials. SendForever stops nothing you already send.
- Connecting changes nothing in Amazon SES. A domain only changes once you add it to SendForever, and only that domain.
- The stack adds its own resources beside yours, all named sendforever.
What changes, and when:
- A domain you add that is already in Amazon SES keeps its configuration set and also sends its events to your activity log. SendForever sets its MAIL FROM domain only if it has none.
- If you apply the recommended configuration set from the domain's page, the domain gets a set of its own as its default. Its old set stays as it was, and its open and click tracking and kept copies carry over. Calls from your app that name a configuration set keep using that set.
- Your Amazon SES contact list, if you have one, stays as it is. Audiences keep their people in a separate table in your account, sendforever-contacts.
- Templates your app sends stay in Amazon SES and also show in SendForever, under Not in a domain until you add them to one. Saving one in SendForever changes what your app sends.
- The suppression list, Deliverability insights and the request for production access only change when you choose them here.
The AWS account page
Open the team's Settings and choose AWS account. Everyone on the team can see it, but only owners and admins can change anything.
For a connected account, the Connected account card shows:
- AWS account ID, SES region and Access role.
- Stack version: the template version your stack last reported. It shows Latest, or the newer version you can update to, such as Version 16 available. It says Unknown if the stack never reported a version.
- Stack updates: who applies new versions of the stack, SendForever or By hand in AWS, as read from the stack. Owners and admins see Change, which shows the steps to change it in AWS (see below).
- Production access: Enabled, or Sandbox while Amazon SES only delivers to verified addresses. See Leaving the SES sandbox.
- Sending: Enabled, or Paused by AWS.
- Standing with AWS: Healthy, or Under review with what Amazon SES reported.
- Suppression list: On, or an Enable button. While it is off, addresses that bounce or complain still get emailed. See Suppressions.
- Last verified: when SendForever last confirmed that the role works.
The standing and suppression list rows come from the latest check of any of your domains, so they appear once a domain has been checked.
Below the card:
- A sandbox callout with Request production access, while the account is in the sandbox.
- Update your AWS stack, when a newer template is offered (see below).
- Alerts: whether the alerts topic is Connected to SendForever, the bounce and complaint alarms, and who the alerts are emailed to. See Alerts.
- Deliverability insights: Amazon SES's Virtual Deliverability Manager. It is one switch for the whole account, because Amazon SES charges for it on every email the account sends. Click Turn on, for $0.07 per 1,000 emails, or Turn off.
- What AWS bills you: an estimate of your monthly AWS bill (see below).
- Verify again, which checks the role again, and Switch AWS account.
When a team has just connected and has no domain yet, the page shows Connected. Next: add the domain you send from with Add a domain.
Update the stack
When SendForever offers a newer version of the template, team owners and admins see Your AWS stack needs an update at the top of every page. Getting started and the AWS account page show it too, whatever step you are on. It says which version your stack is on and which version is available. The update adds what newer features need, such as the activity log, alerts, templates and campaigns. Your domains, credentials and sending stay as they are. Features that need the newer stack say so until you update it, and nothing is blocked while you wait.
From SendForever
If your team chose SendForever under Stack updates and your stack is on version 17 or later, you update it from SendForever itself: click Update the stack. SendForever has AWS prepare the update from its template, checks that the template is the one it expects, and applies it within the ceiling that every role of the stack has. While AWS works, which usually takes a few minutes, it shows Updating your AWS stack to version …. It opens the page again once the update is done. If AWS cannot apply the update, AWS puts the stack back as it was and SendForever says why, with Try again, Open the stack in AWS and Dismiss.
By hand
You update the stack by hand when it is set to By hand in AWS, when it runs with an IAM role of its own, or when a new version changes what SendForever can never change by itself: version 23 needed more than the ceiling allowed, version 30 narrows it, and version 32 narrows it again, keeps sending users to sending email and lets only SendForever's own IAM user assume the role. Update the stack on the banner opens the AWS account page at the Update your AWS stack card. If SendForever normally updates your team's stack, Stack updates says that this one version is yours to apply, and SendForever applies the versions after it again. The card shows these steps:
- Sign in to the AWS console with the account named on the card. Its number shows in the top right corner.
- Click Open the update in AWS on the card. It opens the update of your stack in CloudFormation with the new template already chosen. Check that the Amazon S3 URL matches the Template URL shown on the card, and choose "Next".
- Check that CallbackUrl matches the address shown on the card. Keep every other parameter as it is, and choose "Next". When SendForever applies an update itself, it sets CallbackUrl on its own.
- Leave the stack options as they are, with no IAM role, and tick "I acknowledge that AWS CloudFormation might create IAM resources with custom names". On the last page, choose "Submit". If you give the stack an IAM role, it must be updated by hand from then on.
- Once the stack shows
UPDATE_COMPLETE, come back and click Check the update. SendForever verifies the role again and reads the new version from the stack.
If AWS does not open the update, or the card says Open the stack in AWS because SendForever does not know your stack yet, find it yourself in place of step 2: search for "CloudFormation" at the top of the console, choose the region named on the card in the top right corner, and under "Stacks" choose the stack named on the card. Then choose "Update stack", "Make a direct update", "Replace existing template" and "Amazon S3 URL", paste the Template URL, and choose "Next".
Members see "Ask a team owner or admin to update it."
Change who updates the stack
Only you can change this, and only in AWS. SendForever could never give itself back a permission the stack took away, and every update it applies leaves StackUpdates as it is. On the AWS account page, owners and admins click Change beside Stack updates to see the steps:
- Sign in to the AWS console with the account named in the steps.
- Open CloudFormation in the stack's region.
- Under "Stacks", choose the stack named in the steps, then "Update stack" and "Make a direct update".
- Choose "Use existing template" and "Next".
- Set StackUpdates to
ByHandorSendForever. Check that CallbackUrl matches the address shown, keep every other parameter as it is, and choose "Next". - Leave the stack options as they are, with no IAM role, and tick "I acknowledge that AWS CloudFormation might create IAM resources with custom names". On the last page, choose "Submit".
- Once the stack shows
UPDATE_COMPLETE, it tells SendForever on its own. If the page does not show the change, click Check the change.
When the connection fails
If SendForever cannot use the role, pages show We could not access your AWS account with AWS's explanation. Owners and admins also get Fix connection or Try again. The message names these common causes:
- The stack was deleted, created again, or created from another team. Compare the stack's ExternalId parameter with your team's, which the message shows. If they differ, the stack belongs to another team or an earlier connection. Connect a different AWS account, unless nothing uses that stack. If they match, or the stack is gone, connect again.
- The pasted role is not the one the stack creates. SendForever can only use a role named sendforever-AccessRole-… from a stack still named sendforever. Copy the RoleArn from the stack's Outputs tab, or delete a renamed stack and create it again from the page.
- An explicit deny in your account. The message quotes it as AWS wrote it.
If the problem is on SendForever's side, the message says so. Try again later.
Revoke access or switch accounts
To revoke SendForever's access, delete the sendforever stack in CloudFormation. Access ends at once. The app then shows The CloudFormation stack was deleted: "SendForever no longer has access to this AWS account. Connect again to resume." The S3 bucket with your activity log stays in your account, and the notice names it. A stack you create later makes a new bucket and leaves the old one as it is. If you no longer want the old bucket, open it in Amazon S3 and choose Empty, then Delete.
The resources SendForever created through the role, listed above, are not part of the stack, so they stay too. They include the sending users. While any sending user is left, IAM keeps the sendforever-sending-boundary policy, so the stack deletion can stop on it. A new stack in that account then needs the steps in If only the sending boundary already exists, above.
To connect a different AWS account:
- Remove the team's domains first. They live in the account you are leaving, so Switch AWS account refuses while the team has any. See Domains.
- On the AWS account page, click Switch AWS account and confirm.
- Connect the account you want to send from.
Switching only makes SendForever forget the old account. It changes nothing in AWS, so delete the sendforever stack there yourself when you no longer need it.
While the team has domains, it stays on the account they live in, even after its stack was deleted. Connecting again then works only with a stack in that same account. A stack or role from any other account fails, and the page names the account to sign in to.
What AWS bills you
SendForever's plan is a one-time payment. AWS bills your own account for Amazon SES. The What AWS bills you card on the AWS account page estimates one month of cost. It starts from what your account sent over the last 30 days (or 10,000 emails when it cannot tell) and an average email size of 50 KB. Change either number, or switch options on and off, to see the cost before you change anything. The card uses Amazon SES's pay-as-you-go prices before tax:
- Sending: $0.10 per 1,000 emails, and $0.12 per GB they weigh.
- Deliverability insights: $0.07 per 1,000 emails, on every email the account sends.
- Isolation from other domains: $0.005 per 1,000 emails an isolated domain sends.
- Copies of sent emails: $2 per GB archived, and $0.19 per GB a month for the three months they are kept.
New domains start isolated, and you turn on copies per domain, but the card counts both on every email. The activity log, the webhook, alerts and campaigns cost cents at most, and mostly fall within AWS's free tier. The card does not count dedicated IPs.
Common questions
Which IAM permissions do I need to create the stack?
You need permission to create CloudFormation stacks and IAM roles. The root user and administrators have it, and any other IAM user with those permissions can create the stack too.
Which region should I choose?
If you already send with Amazon SES, answer Yes and choose the region you send from, because production access does not carry over to other regions. Otherwise answer Not yet, and SendForever uses the region nearest your browser's time zone.
Can I change the region later?
Not on a connected account. To send from another region, remove the team's domains, use Switch AWS account, delete the old sendforever stack, and connect again in the new region. The old stack must go because some of its IAM roles and policies have fixed names, and IAM allows each one once per account. The new region starts in the Amazon SES sandbox. A domain removed from SendForever leaves only its sending user behind. If only the sending boundary already exists explains how to handle that.
The stack was created but SendForever still waits. What do I do?
Open the stack in CloudFormation and go to the Outputs tab. Paste the RoleArn value under Stack created, but nothing happened?, then click Verify role. The page points to that box if it hears nothing from the stack for a few minutes.
The stack failed or rolled back. What now?
Open the stack's "Events" tab in CloudFormation to see which resource failed and why. Fix the cause, delete the stack, and choose Connect with AWS again.
Will connecting break what my app already sends with Amazon SES?
No. Your app keeps sending with the same domains and credentials, and connecting changes nothing in Amazon SES. A domain only changes once you add it to SendForever. Templates are shared in Amazon SES, though, so saving a template your app uses changes what your app sends.
Does SendForever store access keys to my account?
No. It stores the role's ARN, your account ID and your region. Each session it opens with the role lasts at most an hour and needs your team's external ID. For sending credentials, it stores only the access key ID, never the secret.
How do I remove SendForever's access?
Delete the sendforever stack in CloudFormation. Access ends at once, and the app shows that the stack was deleted. The bucket with your activity log stays in your account.
Why does the page say to update my AWS stack?
Your stack was created from an older template. Some features, such as alerts or Deliverability insights, need what the current version creates. If your stack is on version 17 or later and Stack updates is set to SendForever, click Update the stack and SendForever updates it. Otherwise, follow the steps on the Update your AWS stack card. Your domains and sending stay as they are.
Can I keep SendForever from changing my stack?
Yes. Choose By hand in AWS when you connect, or later set the stack's StackUpdates parameter to ByHand (see Change who updates the stack). The stack then gives SendForever's role no permission to change it. In IAM, the StackUpdatesAccess policy only allows reading the stack. SendForever shows you the steps to apply each new version yourself.
Why can't I click Switch AWS account?
Switching is blocked while the team has domains, because they live in the account you are leaving. Remove them first. Only owners and admins can switch.
Still stuck?