Audit History

Account audit event coverage, authentication, pagination, outcomes, and recovery. Draft release documentation.

Production availability · November 18, 2026

Audit export on cloud.synaccess.com is awaiting activation. You can prepare your CMP role, credentials and collector now. Begin polling after Synaccess confirms that capture and export are active.

Use GET /v1/audit-events to retrieve account security and device-operation history for Splunk or another destination. Events are accessible for 45 days from capture in CMP.

Once export is active, authorized users can inspect the same feed in CMP under Logs → Activity Log → View audit history.

For credential setup, polling and checkpoint examples, see Send CMP audit events to Splunk. The endpoint reference contains the request and response schema.

Authentication and account scope

Use HTTPS with a valid CMP session or personal access token (PAT):

Authorization: Bearer <CMP_SESSION_OR_PAT>

Account owners and enterprise administrators have audit access by default. Other enterprise users need Read account audit history on the role assigned directly to their user. A device-group role does not grant account-wide audit access. The user and account must remain active and eligible for access; permissions are checked on each request.

An account means one CMP enterprise, or an individual CMP account outside an enterprise. Authorized enterprise users read their enterprise's history across its users and PDUs. CMP determines the account from the credential. Requests do not include an organization selector.

A PAT inherits its user's current permissions, including any other account or device permissions. Use an integration user with the minimum permissions needed. The cursor is independent of PAT rotation; a replacement credential must still have audit access to the same account.

How collection works

  1. Make the first request with since set to an available UTC capture time.
  2. Forward the returned events to your destination.
  3. Save next_cursor after the destination accepts the entire page.
  4. Continue with that cursor. Fetch another page immediately when has_more is true; otherwise resume at your configured polling interval.

Events become available after CMP's audit save commits. They are returned in committed export sequence order, which can differ from when the actions occurred. A delayed save can appear on a later page with an earlier occurred_at or recorded_at. Keep using the cursor as your checkpoint. Retrying a cursor can replay event IDs, so handle duplicates by id.

Request parameters

ParameterMeaning
sinceRequired on the first request. Inclusive RFC 3339 UTC time using Z or +00:00. Compared with recorded_at. Must be within available coverage and no later than the current server time.
cursorThe previous response's next_cursor. Treat it as opaque and URL-encode it. Use either since or cursor, never both.
limitMaximum events per page: 1–1,000, default 500. The server may return fewer events because of its response-size limit.
GET /v1/audit-events?since=2026-11-16T00:00:00Z&limit=500
Authorization: Bearer <CMP_SESSION_OR_PAT>
GET /v1/audit-events?cursor=<URL_ENCODED_NEXT_CURSOR>&limit=500
Authorization: Bearer <CMP_SESSION_OR_PAT>

Choose an actual start time within your available coverage; the date above is illustrative. Event filters, offset pagination, and a request body are not supported.

Example response: an outlet reboot

This example shows the request and result together. They may arrive on separate pages in normal collection.

{
  "events": [
    {
      "id": "8b85e919-e4c9-4d30-b44b-fc68e57a8b6f",
      "schema_version": 1,
      "organization_id": "enterprise:42",
      "actor": {
        "type": "user",
        "id": "482",
        "attribution": "authenticated",
        "display_name": "Example Operator"
      },
      "source": {
        "auth_method": "pat",
        "credential_id": "pat:91",
        "ip_address": "192.0.2.10"
      },
      "target": {
        "type": "outlet",
        "id": "123456-2",
        "device_id": "9441604d-a2d7-4871-b875-e52ec257f5b1",
        "device_serial": "123456",
        "outlet_number": 2
      },
      "action": "reboot",
      "correlation_id": "215ec6ad-3255-4443-ae65-21050da46296",
      "type": "outlet.command_requested",
      "occurred_at": "2026-11-16T12:00:00.123Z",
      "recorded_at": "2026-11-16T12:00:00.130Z",
      "outcome": "accepted",
      "details": {}
    },
    {
      "id": "a2a0ca12-175a-4210-badc-61c1fa2a39c1",
      "schema_version": 1,
      "organization_id": "enterprise:42",
      "actor": {
        "type": "user",
        "id": "482",
        "attribution": "authenticated",
        "display_name": "Example Operator"
      },
      "source": {
        "auth_method": "pat",
        "credential_id": "pat:91",
        "ip_address": "192.0.2.10"
      },
      "target": {
        "type": "outlet",
        "id": "123456-2",
        "device_id": "9441604d-a2d7-4871-b875-e52ec257f5b1",
        "device_serial": "123456",
        "outlet_number": 2
      },
      "action": "reboot",
      "correlation_id": "215ec6ad-3255-4443-ae65-21050da46296",
      "type": "outlet.command_result",
      "occurred_at": "2026-11-16T12:00:01.456Z",
      "recorded_at": "2026-11-16T12:00:01.460Z",
      "outcome": "device_acknowledged",
      "details": {
        "device_response_status": 200
      }
    }
  ],
  "next_cursor": "<opaque cursor returned by CMP>",
  "has_more": false,
  "coverage_started_at": "2026-11-16T00:00:00.000Z",
  "earliest_available_at": "2026-11-16T00:00:00.000Z",
  "coverage_gaps": []
}

The two events have different id values and share a correlation_id. accepted records the request; device_acknowledged records the PDU's successful response. Acknowledgment alone does not establish that the attached equipment completed a reboot.

Response fieldMeaning
eventsThis page's committed audit records. An empty array means no matching saved events are available after the current position.
next_cursorBookmark for the next request. Save it even when events is empty, after completing delivery of the page.
has_moretrue when another matching page was available at the time of the read. false means caught up for now.
coverage_started_atWhen capture began for this audit feed. Earlier activity is not backfilled.
earliest_available_atEarliest supported start time, accounting for rollout and the 45-day API window.
coverage_gapsPossible capture-failure intervals CMP was able to identify. An empty array does not prove complete capture.

For an empty page, continue from its next_cursor; do not restart with the current time.

Event fields and outcomes

Event fieldMeaning
idStable UUID for this audit record. Use for duplicate detection.
schema_versionEvent envelope version; initially 1.
typeEvent name from the catalog below. Clients should tolerate future additional fields and event types.
occurred_atWhen the action or observation occurred, as known to CMP.
recorded_atDatabase-assigned capture timestamp for CMP's audit write. It can precede transaction commit/visibility and governs since filtering and the 45-day API access window.
organization_idAccount scope, such as enterprise:42 or account:482. Treat the identifier as opaque.
actorInitiating user or system when known. Includes attribution; unavailable identity is represented as unknown.
sourceAvailable authentication method, credential identifier, and observed IP address. A proxy address may be reported. Credential values are excluded.
targetAffected resource type/identifier, with device UUID, serial, or outlet number when applicable. An identifier can be null when unavailable.
actionOperation, such as reboot or update, when applicable.
outcomeStrongest result CMP could establish, when applicable.
correlation_idLinks a configuration or command request with its result; null when not applicable.
detailsApproved event-specific fields, including change markers, safe values, response status, or reason_code.
OutcomeMeaning
acceptedCMP received the request or queued the work.
device_acknowledgedThe PDU reported success.
confirmedOperation-specific evidence exists, such as a committed account change or verified configuration readback.
rejectedA negative response or a decision not to send the operation was established.
failedCMP established a failure.
unknownCMP could not establish the outcome, for example after interruption or timeout.

Pairing/unpairing records describe the response observed for the dispatched operation. A device acknowledgment alone is insufficient for confirmed completion. Pending or uncertain responses remain unknown; later ownership changes are not inferred as completion. Records stay associated with the initiating account, and audit processing never repeats the device command.

An unmatched request can remain pending while a queued rename waits for its PDU. CMP can record a recovered unknown result without sending the command again. Audit result events describe observations; they do not instruct an integration to retry device commands.

details.changes contains approved field names and selected safe values. Secret changes are marked as redacted. details.unlisted_fields and details.changes_truncated indicate omitted detail. Passwords, token values/hashes, raw assertions, and sensitive configuration values are excluded.

For bulk/group commands dispatched as individual device requests, your integration may send the same UUID in X-CMP-Parent-Operation-Id on those command requests. Captured records include it as details.parent_operation_id.

Event catalog

Event typeTrigger
auth.login_succeededCMP password or SAML/OIDC sign-in succeeded.
auth.login_failedA sign-in attempt failed and CMP could associate it with the account.
user.createdA CMP user was created through signup, enterprise provisioning, or SSO provisioning.
user.updatedA user profile, password, or account setting changed.
user.enabledA change made the user eligible for account access.
user.disabledA change removed the user's eligibility for account access.
user.deletedA CMP user was deleted.
user.role_changedA user's direct role assignment or administrative privileges changed.
user.access_changedA user's direct PDU/group access assignments changed.
role.createdAn account role was created.
role.permissions_changedA role's permissions or details changed.
role.deletedAn account role was deleted; its assignments were removed.
group.role_changedA device group's role assignment changed.
group.createdA CMP group was created, including an empty group.
group.updatedA CMP group name, description, or other group setting changed.
group.deletedA CMP group was deleted.
group.member_addedA device, outlet, or user was added to a CMP group. The record identifies the group and member.
group.member_removedA device, outlet, or user was removed from a CMP group. Bulk edits record actual differences; deleting a group emits its group deletion event.
enterprise.security_policy_changedEnterprise SAML/OIDC, SSO requirements, provisioning, approval, or email-verification policy changed. Secret values are excluded.
notification_rule.createdA CMP notification rule and its target/event/destination configuration was created.
notification_rule.updatedA CMP notification rule was changed, enabled, or disabled. Destination secrets are excluded.
notification_rule.deletedA CMP notification rule was deleted.
api_credential.createdA personal access token was created.
api_credential.enabledA personal access token was enabled.
api_credential.disabledA personal access token was disabled.
api_credential.revokedA personal access token was deleted/revoked.
device.configuration_requestedCMP received a supported configuration change or queued rename.
device.configuration_resultCMP established an outcome for a supported configuration change.
outlet.command_requestedCMP received an on, off, or reboot command for an outlet.
outlet.command_resultCMP established an outcome for an outlet command.
device.onlineCMP observed the device transition online.
device.offlineCMP observed or verified an offline transition.
device.pairing_requestedCMP initiated a pairing operation for the account.
device.pairing_resultCMP recorded the observed pairing response: device acknowledgment or an unknown outcome. Later ownership changes are not inferred as completion.
device.unpairing_requestedCMP initiated an unpairing operation for the account.
device.unpairing_resultCMP recorded the observed unpairing response. Confirmation requires completion evidence tied to the operation; uncertain outcomes remain unknown.

CMP group lifecycle and membership changes, enterprise sign-in policy changes, notification-rule changes, and pairing/unpairing are included as listed above. Group records distinguish cloud groups from PDU-local outlet groups.

Supported configuration operations include CMP configuration and names; outlet, inlet and bank settings; outlet groups; device users; Wi-Fi, VLANs, static routes and WAN health; controller restart; and device event/action definitions.

The feed excludes general telemetry, device provisioning logins, identity-provider failures that never reach CMP, firmware updates, arbitrary CLI, and actions executed entirely on a PDU. An offline event means CMP considers a device unreachable; it does not establish power loss.

History, limits, and recovery

Events are accessible for 45 days from recorded_at. Coverage begins when capture is enabled; earlier activity is not reconstructed. Your destination can retain its own copy longer.

Audit capture failures allow valid customer operations to continue. CMP reports possible intervals it could identify in coverage_gaps; an interval can combine incidents and does not identify exact missing events.

The starting read budgets are 60 requests per minute per account and 30 per principal within that account, shared across that principal's credentials. Honor Retry-After on 429. Polling once per minute is a starting point for a caught-up integration; processing, collection, network, and destination latency also affect arrival time.

StatusAction
400Correct an invalid query, timestamp, limit, or cursor.
401Check the session/PAT; it may be invalid, expired, disabled, or revoked.
403Check current account access and the direct user-role permission. A cursor from another account is also rejected.
410Requested history cannot resume safely. Record the gap and choose a new since using the error's coverage boundaries.
429Wait for Retry-After, then retry without advancing the checkpoint.
503Retry with backoff without advancing the checkpoint. Honor Retry-After when present.

Errors contain error.code, error.message, and request_id. A 410 error also includes coverage_started_at and earliest_available_at inside error. Contact support with the request ID when investigating a persistent failure.