Cloud platforms

Send AWS CloudWatch Alarm (via SNS) alerts to on-call with CallHeim.

CloudWatch alarm published to SNS → HTTPS subscription. CallHeim maps the payload, attaches repeats to the incident that is already open, and pages whoever is on call for the service it belongs to.

cloudwatchalarmsource names also accepted: cloudwatch-sns, cw-sns, sns-cloudwatch

AWS CloudWatch alarms, delivered through an SNS topic, become CallHeim alerts: repeats of the same alarm collapse onto one incident, and the service the integration is bound to pages whoever is on call for it. CallHeim includes a payload mapping for AWS CloudWatch’s webhook format. To forward through EventBridge instead of SNS, use the AWS CloudWatch (EventBridge) guide.

How the alert reaches on-call

  1. Step 1A CloudWatch alarm changes state and publishes to an SNS topic in your account.
  2. Step 2SNS delivers the notification as an HTTP(S) POST to your CallHeim ingest URL (the SNS envelope, with the alarm JSON stringified inside its Message field).
  3. Step 3CallHeim parses the inner Message, reads AlarmName as the identity signal, and maps NewStateValue to a severity.
  4. Step 4A repeat of the same alarm, account and region attaches to the incident that is already open for it, if that incident was triggered in the last 24 hours.
  5. Step 5The integration's bound service selects an escalation policy, which pages the on-call tier.

Setting it up

  1. 01

    Create a standard SNS topic in the same AWS account and region as the alarm.

    CloudWatch alarms can only publish to an SNS topic in the same account and region. Make it a standard topic: a FIFO topic cannot deliver to HTTPS.

  2. 02

    Subscribe your CallHeim ingest URL to the topic with protocol HTTPS.

    Create the CloudWatch Alarm (via SNS) integration in CallHeim first to get the exact ingest URL. Raw message delivery can be on or off: CallHeim reads the notification envelope, or the bare alarm message, either way.

  3. 03

    Confirm the subscription.

    SNS sends a SubscriptionConfirmation message to the endpoint first; CallHeim confirms the subscription automatically, usually within a few seconds of you creating it. If it stays "Pending confirmation", choose Request confirmation.

  4. 04

    Point the alarm's ALARM and OK actions at the same SNS topic.

    For the alarm (or a whole set of alarms), add a Notification with alarm state trigger "In alarm" that notifies the topic, and a second Notification with alarm state trigger "OK" that notifies the SAME topic. The OK notification is what resolves the incident. Without it the incident stays open until a person resolves it — see Recovery below.

  5. 05

    Leave the INSUFFICIENT_DATA action off this topic.

    That state counts as firing and would open an incident (at P3, see the field mapping); it never resolves one. Notify this topic on "In alarm" and "OK" only.

  6. 06

    (Alternative) Route through an EventBridge rule instead of SNS.

    AWS also lets you forward CloudWatch Alarm State Change events with an EventBridge rule and an API destination. That route has its own integration and guide, AWS CloudWatch (EventBridge): the rule sends the event exactly as EventBridge emits it, with no input transformer, so there is nothing to reshape.

Example payloadJSON
{
  "Type": "Notification",
  "MessageId": "8f5a1c2e-...",
  "TopicArn": "arn:aws:sns:us-east-1:111122223333:alarms",
  "Subject": "ALARM: prod-api-5xx",
  "Message": "{\"AlarmName\":\"prod-api-5xx\",\"AlarmDescription\":\"5xx errors elevated\",\"NewStateValue\":\"ALARM\",\"NewStateReason\":\"5xx > 10 for 2 datapoints\",\"Region\":\"us-east-1\",\"AWSAccountId\":\"111122223333\"}",
  "Timestamp": "2026-06-28T00:00:00Z"
}
Example payload in the SNS envelope; check AWS's own documentation for the full notification format.

What CallHeim reads from the payload

Payload fields and what each one maps to on the incident
Payload fieldMaps toNote
Message (inner AlarmName, or Subject)incident title
Message.NewStateValue (ALARM / OK / INSUFFICIENT_DATA)severityALARM→P2, INSUFFICIENT_DATA→P3, OK→P4.
Message.NewStateReason / AlarmDescription / Subjectincident body
Message.AlarmName + AWSAccountId + Regionidentity signalScopes the same alarm name in two different accounts or regions to separate incidents.

How CallHeim processes it

Routing

Each alert source is bound to a service, and the service’s escalation policy (or its team’s) sets who is paged.

Deduplication

A repeat of the same alert attaches to the incident that is still open for it in the same service, triggered or acknowledged, with no new incident and no new page. When the sender supplies its own dedup key (a PagerDuty dedup_key, an Opsgenie alias, a Datadog alert id, an Alertmanager fingerprint, or a dedupKey that is more than the alert title), this holds however late the repeat arrives. When the identity comes from the e-mail subject or the alert title, the repeat attaches only if the incident was triggered in the last 24 hours; after that it opens a new incident and pages. A repeat at P1 or P2 is not folded into an open P3, P4 or P5 incident: it opens a new incident and pages. Once an incident is resolved or closed, the next repeat opens a new incident. For this source the fingerprint is the alarm name scoped to the AWS account and region it came from, so the same alarm firing again on the same resource collapses onto the open incident. The alarm name is the identity, so the 24-hour limit applies: a repeat joins the open incident only if it was triggered in the last 24 hours.

Recovery

When the alarm returns to OK, CallHeim resolves the incident. INSUFFICIENT_DATA does not resolve it; it is recorded at P3. Checked with a real CloudWatch alarm and SNS HTTPS subscription on 25 September 2026. An alarm with no OK action never sends the OK, so its incident stays open until a person resolves it.

Troubleshooting

Common problems and how to fix them
ProblemFix
No alerts arrive at all.Check the SNS subscription status in the AWS console — an unconfirmed subscription (still "Pending confirmation") delivers nothing; choose Request confirmation. Check too that the topic is a standard topic: a FIFO topic cannot deliver to HTTPS.
The alert body is empty or unreadable.Something other than SNS is posting to the URL, or the alarm is publishing to a different topic than the one subscribed. CallHeim reads the SNS envelope shape shown in the sample, or the bare alarm message when raw message delivery is on.
An OK state does not close the incident.Check that the alarm's OK action notifies the same SNS topic, and that the OK comes from the same alarm, account and region as the ALARM that opened the incident.
An alarm with no data opens an incident.The alarm's INSUFFICIENT_DATA action notifies this topic. That state counts as firing and would open an incident: remove that action, and keep only the "In alarm" and "OK" notifications on this topic.
Two different AWS accounts' alarms with the same name collapse together.They shouldn't — the fingerprint includes AWSAccountId and Region. Check both alarms are actually publishing the account id in the inner Message (a hand-rolled EventBridge-to-SNS bridge can drop it).
Alerts stop for a while, then a new incident opens.The earlier incident was resolved or closed in the meantime (an OK transition resolves it), or it was triggered more than 24 hours ago (the alarm name is the identity, so the 24-hour limit applies), so the next ALARM opens a new incident by design — that is not a bug. While a recent incident is still open, a repeat attaches to it instead.

Security

SNS HTTPS delivery cannot add a custom header. The integration URL is the credential: keep it secret, and if it is exposed, disable the integration and create a new one. A signing proxy you run in front of the sender can add an X-CallHeim-Signature header; without one, do not set a signing secret, because every unsigned delivery would be rejected.

Vendor documentation checked

The thresholds it passes through

Repeat, sender-supplied dedup key
attaches, any age
Repeat, title or subject only
attaches, incident under 24 h old
Flap threshold
4 transitions / 600s
Title correlation
similarity ≥ 0.6, same source and service
Group window default
600s

All defaults are published. You can turn title correlation off or change its threshold, and set the window on your own noise rules; the dedup rule and the flap settings are fixed. A P1 or P2 repeat on an open P3 to P5 incident opens a new incident and pages. How alerts are processed →

CallHeim

Point AWS CloudWatch Alarm (via SNS) at CallHeim and see what it does with your alerts.

Start on the free trial, connect AWS CloudWatch Alarm (via SNS) and send a test alert to see the incident it opens.