Skip to main content

ankra alerts

Manage where Ankra delivers alerts and platform notifications. Destinations are the endpoints notifications are sent to: a webhook URL (Slack, Microsoft Teams, Discord, PagerDuty, or any custom receiver) or a Slack/Teams channel the Ankra bot posts to directly. Routes decide which notifications reach which destination, filtered by kind, severity, cluster, and source, so the whole alerting setup can live in scripts and CI next to the rest of your platform configuration.

ankra alerts destinations

Manage the organisation’s alert destinations. A destination is either a webhook URL or a channel-based Slack/Teams destination that carries a channel id instead of a URL (list the channels the Ankra bot can post to with ‘ankra alerts destinations channels’). Webhook URLs are shown masked on every read.

ankra alerts destinations channels

List the channels the Ankra bot can post to, for use with ‘ankra alerts destinations create --channel-id’. Both providers are shown unless --provider narrows it to one. A provider whose workspace or tenant is not connected to the organisation reads “not connected”; one whose bot service is not configured on this platform reads “not available”. Neither is an error.
Flags

ankra alerts destinations create

Create an alert destination. Pass --url for a webhook destination, or --channel-id for a channel-based Slack or Teams destination (find ids with ‘ankra alerts destinations channels’); a Teams channel also needs --teams-tenant-id. --type records which receiver the destination targets (slack, teams, discord, pagerduty, custom; default slack) and selects the default payload format; --template-file overrides the payload with your own template.
Flags

ankra alerts destinations delete

Delete an alert destination. Routes that pointed at it stop delivering; remove or re-point them with ‘ankra alerts routes’.
Flags

ankra alerts destinations get

Show one alert destination
Flags

ankra alerts destinations list

List the organisation’s alert destinations, 20 per page. Filter by name with --search and by state with --enabled or --disabled. The ID column is what routes and the other destination commands take.
Flags

ankra alerts destinations test

Send a sample notification to a stored destination and report whether the receiver accepted it. A failed delivery exits non-zero so CI can gate on it; the details are still printed (or emitted with -o json).
Flags

ankra alerts destinations test-url

Send a sample notification to an ad-hoc webhook URL, optionally with a custom payload template, without creating a destination. A failed delivery exits non-zero.
Flags

ankra alerts destinations update

Update an alert destination. Only the flags you pass are changed; the rest keep their current values.
Flags

ankra alerts ingest-credentials

Ingest credentials are the per-credential secrets an external Alertmanager or log monitor uses to deliver alerts into Ankra. A credential is pinned to one cluster (scope cluster), carries the platform’s own alerts with no pin (scope platform), or does both from a self-hosting cluster (scope mixed). The notifier posts to <platform>/webhooks/alerts/<credential-id> with the token as a Bearer Authorization header (or an X-Ankra-Alert-Token header). The token is minted by create and shown that once; it cannot be read back. A rebind moves the pin or the scope on the existing credential, so the notifier keeps delivering; a delete revokes the credential.

ankra alerts ingest-credentials create

Mint an ingest credential for an external Alertmanager or log monitor. The output carries the credential id, the webhook URL the notifier posts to, and the token. The token is shown this once and cannot be read back: store it in the notifier’s configuration (or a secret) straight away. --cluster pins the credential to a cluster, by name or id. --scope is cluster (the default), platform (the platform’s own alerts, never pinned) or mixed (both, always pinned). --token-file writes the token to a new file with mode 0600 and leaves it out of the output, -o json included, so automation never prints the secret. The file must not exist yet. Alertmanager delivers with a webhook receiver:
Flags

ankra alerts ingest-credentials delete

Delete an alert ingest credential. Its token is revoked: every delivery the notifier makes with it from then on is refused. Asks first; --yes skips the question, and without it a non-interactive run is cancelled.
Flags

ankra alerts ingest-credentials list

List the organisation’s alert ingest credentials with the cluster each one is pinned to. A pin whose cluster is deleted, archived, slated for deletion or gone shows as BROKEN: the notifier still delivers, but its alerts have no cluster to land on until the credential is rebound.
Flags

ankra alerts ingest-credentials rebind

Rebind an ingest credential: pin it to another cluster, unpin it, or change its scope. The token is untouched, so the notifier keeps delivering through the same credential. Pass --cluster with a cluster name to pin, --unpin to clear the pin, and --scope with cluster, platform or mixed. A platform-scoped credential carries no pin; a mixed one must keep one.
Flags

ankra alerts routes

Manage the organisation’s notification routes. A route sends notifications matching every filter it sets (kind, severity, cluster, source) to one destination; a route with no filters matches everything. Routes are evaluated in ascending priority, mode “exclude” withholds matches instead of delivering them, and --stop-on-match ends the walk at that route so lower-priority routes never see the notification.

ankra alerts routes create

Create a notification route to a destination. Every filter is optional; a route with none matches every notification. Kinds include execution_failed, resource_deployment_failed, resource_health_degraded, alert_trigger_fired, gitops_sync_failed, agent_offline, security_new_severe_cves, and the other platform notification kinds; severities are critical, warning, and info.
--application limits a route to one application’s notifications (pipeline runs, image gates, previews); --reference-pattern to the branches, tags or pull requests a glob matches. “main” matches pushes to main and never a pull request targeting main; “pull/” matches every pull request; “release/” one path segment and “release/**” any depth; “refs/tags/v*” tags. A route belongs to a project and delivers only that project’s notifications (those about its tickets, applications, pipeline runs, clusters and the rest). With neither --project nor --organisation-wide the route is created in your current project (‘ankra project current’; a project-pinned token always uses its pin). The global --project <id|slug|name> creates it in that project. --organisation-wide creates a route that delivers every project’s notifications and the organisation’s own (billing and the like); it needs alerts.manage at organisation scope. Move an existing route with ‘ankra project move --kind notification_route --id <route-id> --to <project>’.
Flags

ankra alerts routes delete

Delete a notification route
Flags

ankra alerts routes list

List notification routes
Flags

ankra alerts routes preview

Resolve a hypothetical notification against the organisation’s routing rules and, with --alert-id, that alert’s own destinations, and print the destinations it would reach with the reason each one matched. Nothing is delivered and nothing is changed - this is a dry run. Pass --alert-id whenever you preview an alert firing. A destination reached by both the alert’s own list and a routing rule is delivered to once, and without the alert id the preview cannot tell you which rule that affects.
--project resolves the notification as one of that project’s: only routes of that project and organisation-wide routes take part in the walk.
Flags

ankra alerts routes test

Queue a sample notification through a route’s destination. Delivery is asynchronous: the command returns the delivery id once the sample is queued, not once the receiver has accepted it.
Flags

ankra alerts routes update

Update a notification route. Only the flags you pass are changed; the rest keep their current values.
Flags

ankra alerts rules

Manage the organisation’s alert definitions: the rules Ankra evaluates against your clusters and opens incidents from. Reading needs alerts.read. Creating, editing and deleting a definition needs alerts.manage; muting one (disable) and re-enabling it needs alerts.write, the same as acknowledging or snoozing a firing.

ankra alerts rules create

Create an alert definition. The file carries the platform’s create body, in YAML or JSON: name, description, enabled, rules (each with name, resource_type, conditions, severity and the clusters it applies to) and the integration_names of the destinations to notify. Pass -f - to read stdin.
Flags

ankra alerts rules delete

Delete an alert definition and its rules. Its incident history stops being evaluated; to stop it notifying while keeping it, use ‘ankra alerts rules disable’ instead.
Flags

ankra alerts rules disable

Mute an alert definition; its history is kept
Flags

ankra alerts rules enable

Re-enable a muted alert definition
Flags

ankra alerts rules get

Show one alert definition
Flags

ankra alerts rules list

List the organisation’s alert definitions, 20 per page. Filter by name with --search, by state with --enabled or --disabled, and by the rules’ --severity, --resource-type or --cluster-id.
Flags

ankra alerts rules update

Change an alert definition. The file names only the members to change (name, description, enabled, rules, integration_names); the rest are kept. A rules member replaces the definition’s whole rule set.
Flags

ankra alerts triggers

Act on one firing (trigger) of an alert definition. Both need alerts.write.

ankra alerts triggers acknowledge

Acknowledge a firing
Flags

ankra alerts triggers snooze

Snooze a firing for a number of minutes
Flags