Alerting_v2: Dispatcher notification policy#252758
Alerting_v2: Dispatcher notification policy#252758kdelemme merged 63 commits intoelastic:alerting_v2from
Conversation
|
/ci |
|
/ci |
|
/ci |
cnasikas
left a comment
There was a problem hiding this comment.
LGMT! I am almost done with the review. I will continue tomorrow with the most complex step, and I will post a last review.
x-pack/platform/plugins/shared/alerting_v2/server/lib/dispatcher/tokens.ts
Outdated
Show resolved
Hide resolved
x-pack/platform/plugins/shared/alerting_v2/server/lib/dispatcher/tokens.ts
Outdated
Show resolved
Hide resolved
...ugins/shared/alerting_v2/server/lib/notification_policy_client/notification_policy_client.ts
Outdated
Show resolved
Hide resolved
| ); | ||
|
|
||
| bind(RulesSavedObjectService).toSelf().inRequestScope(); | ||
| bind(RulesSavedObjectServiceInternalToken) |
There was a problem hiding this comment.
What about (pseudo code)? Could that work?
bind(RulesSavedObjectServiceScopedToken).toSelf().inRequestScope();
bind(RulesSavedObjectServiceInternalToken).toSelf().inSingletonScope();
bind(NotificationPolicySavedObjectServiceScopedToken)
.toDynamicValue(({ get }) => {
const soClient = get(RulesSavedObjectServiceScopedToken);
const spaces = get(PluginStart<AlertingServerStartDependencies['spaces']>('spaces'));
return new NotificationPolicySavedObjectService(soClient, spaces);
})
.inRequestScope();
bind(NotificationPolicySavedObjectServiceInternalToken)
.toDynamicValue(({ get }) => {
const soClient = get(RulesSavedObjectServiceInternalToken);
const spaces = get(PluginStart<AlertingServerStartDependencies['spaces']>('spaces'));
return new NotificationPolicySavedObjectService(soClient, spaces);
})
.inSingletonScope();
If this will not work, we can follow the same example as EsServiceInternalToken and EsServiceScopedToken for the saved object client.
There was a problem hiding this comment.
Why would the NotificationPolicy SO service uses the Rules SO Service? This part is not clear to me.
As for renaming the token, you're suggesting to follow XXXInternalToken (using internal dependencies) and XXXScopedToken (using request scoped dependencies)?
There was a problem hiding this comment.
Why would the NotificationPolicy SO service uses the Rules SO Service? This part is not clear to me.
Yes, sorry, it was a mistake by me. What I had in mind was something like this but feel free to ignore:
bind(InternalSavedObjectsClientFactory)
.toResolvedValue(
(savedObjects) => (includedHiddenTypes) => {
const client = savedObjects.createInternalRepository(includedHiddenTypes);
return () => client;
},
[CoreStart('savedObjects')]
)
.inSingletonScope();
bind(RulesSavedObjectService).toSelf().inRequestScope();
bind(RulesSavedObjectServiceScopedToken).toService(RulesSavedObjectService);
bind(RulesSavedObjectServiceInternalToken)
.toResolvedValue(
(internalFactory, spaces) =>
new RulesSavedObjectService(internalFactory([RULE_SAVED_OBJECT_TYPE]), spaces),
[
InternalSavedObjectsClientFactory,
PluginStart<AlertingServerStartDependencies['spaces']>('spaces'),
]
)
.inSingletonScope();
bind(NotificationPolicySavedObjectService).toSelf().inRequestScope();
bind(NotificationPolicySavedObjectServiceScopedToken).toService(
NotificationPolicySavedObjectService
);
bind(NotificationPolicySavedObjectServiceInternalToken)
.toResolvedValue(
(internalFactory, spaces) =>
new NotificationPolicySavedObjectService(
internalFactory([NOTIFICATION_POLICY_SAVED_OBJECT_TYPE]),
spaces
),
[
InternalSavedObjectsClientFactory,
PluginStart<AlertingServerStartDependencies['spaces']>('spaces'),
]
)
.inSingletonScope();
x-pack/platform/plugins/shared/alerting_v2/server/setup/bind_services.ts
Outdated
Show resolved
Hide resolved
x-pack/platform/plugins/shared/alerting_v2/server/lib/dispatcher/steps/apply_throttling_step.ts
Outdated
Show resolved
Hide resolved
.../platform/plugins/shared/alerting_v2/server/lib/dispatcher/steps/fetch_policies_step.test.ts
Outdated
Show resolved
Hide resolved
x-pack/platform/plugins/shared/alerting_v2/server/lib/dispatcher/steps/fetch_rules_step.test.ts
Outdated
Show resolved
Hide resolved
x-pack/platform/plugins/shared/alerting_v2/server/lib/dispatcher/dispatcher.test.ts
Outdated
Show resolved
Hide resolved
x-pack/platform/plugins/shared/alerting_v2/server/lib/dispatcher/execution_pipeline.ts
Outdated
Show resolved
Hide resolved
x-pack/platform/plugins/shared/alerting_v2/server/lib/dispatcher/steps/fetch_rules_step.test.ts
Outdated
Show resolved
Hide resolved
x-pack/platform/plugins/shared/alerting_v2/server/lib/dispatcher/steps/apply_throttling_step.ts
Show resolved
Hide resolved
x-pack/platform/plugins/shared/alerting_v2/server/lib/dispatcher/steps/record_actions_step.ts
Outdated
Show resolved
Hide resolved
x-pack/platform/plugins/shared/alerting_v2/server/lib/dispatcher/dispatcher.test.ts
Outdated
Show resolved
Hide resolved
x-pack/platform/plugins/shared/alerting_v2/server/lib/dispatcher/dispatcher.test.ts
Outdated
Show resolved
Hide resolved
| updatedAt: string; | ||
| } | ||
|
|
||
| export interface NotificationPolicy { |
There was a problem hiding this comment.
Could we infer this from the schema?
There was a problem hiding this comment.
This one I haven't done it because we use different casing, and the dispatcher don't really care of every fields like createdAt, etc.. But maybe we'll revisit later, and just use the NotificationPolicyResponse as is.
| ); | ||
|
|
||
| bind(RulesSavedObjectService).toSelf().inRequestScope(); | ||
| bind(RulesSavedObjectServiceInternalToken) |
There was a problem hiding this comment.
Why would the NotificationPolicy SO service uses the Rules SO Service? This part is not clear to me.
Yes, sorry, it was a mistake by me. What I had in mind was something like this but feel free to ignore:
bind(InternalSavedObjectsClientFactory)
.toResolvedValue(
(savedObjects) => (includedHiddenTypes) => {
const client = savedObjects.createInternalRepository(includedHiddenTypes);
return () => client;
},
[CoreStart('savedObjects')]
)
.inSingletonScope();
bind(RulesSavedObjectService).toSelf().inRequestScope();
bind(RulesSavedObjectServiceScopedToken).toService(RulesSavedObjectService);
bind(RulesSavedObjectServiceInternalToken)
.toResolvedValue(
(internalFactory, spaces) =>
new RulesSavedObjectService(internalFactory([RULE_SAVED_OBJECT_TYPE]), spaces),
[
InternalSavedObjectsClientFactory,
PluginStart<AlertingServerStartDependencies['spaces']>('spaces'),
]
)
.inSingletonScope();
bind(NotificationPolicySavedObjectService).toSelf().inRequestScope();
bind(NotificationPolicySavedObjectServiceScopedToken).toService(
NotificationPolicySavedObjectService
);
bind(NotificationPolicySavedObjectServiceInternalToken)
.toResolvedValue(
(internalFactory, spaces) =>
new NotificationPolicySavedObjectService(
internalFactory([NOTIFICATION_POLICY_SAVED_OBJECT_TYPE]),
spaces
),
[
InternalSavedObjectsClientFactory,
PluginStart<AlertingServerStartDependencies['spaces']>('spaces'),
]
)
.inSingletonScope();
💔 Build Failed
Failed CI StepsMetrics [docs]
History
cc @kdelemme |
## Summary ### Key capabilities - **ES|QL-native rule evaluation** — Rules are defined as ES|QL queries with optional WHERE clause conditions, evaluated on a configurable schedule - **Alert lifecycle management** — Full episode tracking with pending → active → recovering → inactive state transitions, including configurable alert delay (consecutive breaches / duration) - **Event-driven architecture** — Alert events and actions are stored in dedicated data streams (`.alerting-events`, `.alerting-actions`) with ES|QL views for querying - **Notification dispatch pipeline** — A multi-step dispatcher that matches alert episodes to notification policies, handles throttling/suppression, and triggers Kibana Workflows using encrypted API keys - **Notification policies** — CRUD APIs and UI for creating notification policies with KQL-based rule matching, workflow integration, and API key management - **Rule authoring UI** — A shared rule form package (`@kbn/alerting-v2-rule-form`) usable standalone or embedded in Discover, with ES|QL editor, WHERE clause condition editing, recovery configuration, and live query preview - **Rule management UI** — Full rule list with pagination, enable/disable, clone, edit, and delete operations - **APM instrumentation** — Middleware and decorators for tracing rule execution and client operations ### Architecture highlights - **InversifyJS DI** — All services use constructor injection with typed tokens, scoped per-request or singleton as appropriate - **Pipeline pattern** — Rule executor and dispatcher use composable step-based pipelines - **Saved Objects** — Rules stored as hidden saved objects; notification policies stored as encrypted saved objects (for API key protection) - **Feature privileges** — Dedicated Kibana feature with read/all privileges for RBAC --- ## Contained PRs <details> <summary><strong>Core Engine & Plugin Init</strong> (12 PRs)</summary> - #247283 — Init alerting v2 plugin (@cnasikas) - #247452 — Add the alerting v2 feature privileges (@cnasikas) - #247673 — Director (@cnasikas) - #248306 — Create basic services (@cnasikas) - #248696 — Initialize all resources (@cnasikas) - #250023 — Schema package (@cnasikas) - #250010 — YML Editor (@cnasikas) - #251064 — Remove index.mode: lookup for RnA alert indices (@cnasikas) - #251707 — Simplify task registration pattern (@kdelemme) - #251876 — Dedicated user service (@cnasikas) - #252073 — Use `kbn/data-streams` in alerting_v2 (@cnasikas) - #255120 — Update alerting-v2 owner to new rna project team (@cnasikas) </details> <details> <summary><strong>Rule Execution Pipeline</strong> (12 PRs)</summary> - #247472 — Add alerting v2 Rule Executor (@darnautov) - #248285 — Alerting v2 rule HTTP APIs (@darnautov) - #248728 — Add basic alert actions route (@darnautov) - #250161 — Refactor rule executor to use a pipeline pattern (@darnautov) - #252292 — Implement the CountTimeframeStrategy for the director (@cnasikas) - #252544 — Add support of streaming in the rule executor (@darnautov) - #252754 — Update rule attributes (@kdelemme) - #253355 — Add getRules client method (@kdelemme) - #253668 — Make evaluation.query.condition optional (@kdelemme) - #254031 — Add recovery event generation to rule execution pipeline (@kdelemme) - #255968 — ES|QL views (@adcoelho) - #256697 — Create episodes ES|QL view (@adcoelho) </details> <details> <summary><strong>Alert Suppression & Episodes</strong> (3 PRs)</summary> - #252174 — Alert suppression (@kdelemme) - #256486 — Fix suppression query (@kdelemme) - #256527 — Store 'unmatched' action for unmatched alert episodes (@kdelemme) </details> <details> <summary><strong>Dispatcher & Notification Engine</strong> (6 PRs)</summary> - #250822 — Alerting v2 dispatcher (@kdelemme) - #251529 — Use query service in dispatcher (@kdelemme) - #251679 — Dispatcher task (@kdelemme) - #252758 — Dispatcher notification policy (@kdelemme) - #255332 — Wait for resources before scheduling dispatcher task (@kdelemme) - #256536 — Use stored encrypted API keys from Notification Policy in dispatcher step (@kdelemme) </details> <details> <summary><strong>Notification Policies (Server)</strong> (4 PRs)</summary> - #251336 — Introduce notification policy CRUD APIs and client (@cnasikas) - #253134 — Update notification policy (@cnasikas) - #254808 — Store API key owner on Notification Policy (@kdelemme) - #256940 — Make notification policies global with optional rule-label scoping (@kdelemme) </details> <details> <summary><strong>Notification Policies UI</strong> (1 PR)</summary> - #255599 — Add notification policies UI and Storybook form story (@adcoelho) </details> <details> <summary><strong>Rule Authoring UI</strong> (13 PRs)</summary> - #250961 — Add create rule flyout in Discover (@adcoelho) - #255111 — Add activation configuration fields to alerting V2 rule form (@yiannisnikolopoulos) - #255427 — Rule form: provide services via context (@dominiqueclarke) - #255876 — MVP rule form, Split evaluation condition, and Recovery configuration (@dominiqueclarke) - #256260 — Foundational rule list (@dominiqueclarke) - #256756 — Wire up edit flow (@dominiqueclarke) - #256801 — Move consecutive breaches max to shared constants (@yiannisnikolopoulos) - #256818 — Preview query and design parity (@dominiqueclarke) - #256938 — Allow clearing number inputs in state transition fields (@yiannisnikolopoulos) - #257017 — Add enable/disable and clone rule to rule list (@dominiqueclarke) - #257246 — Remove all React.FC (@dominiqueclarke) - #257415 — Rule form - fix test (@dominiqueclarke) - #257454 — Block comma key in number input component (@yiannisnikolopoulos) </details> <details> <summary><strong>API Documentation & Schema</strong> (2 PRs)</summary> - #254901 — Rename indexes for alert events and actions (@adcoelho) - #255810 — OAS for alert action routes (@adcoelho) </details> <details> <summary><strong>Observability & Monitoring</strong> (3 PRs)</summary> - #254925 — Add ApmMiddleware to the rule executor (@adcoelho) - #255115 — Add the withAPM decorator and apply it to the rules_client (@adcoelho) - #255999 — Fix linting problem in apm middleware (@adcoelho) </details> <details> <summary><strong>CI & Maintenance</strong> (2 PRs)</summary> - #257409 — Refactor SO services to use inversify DI for client initialization (@darnautov) - Fix alerting-v2-schema jest config (@darnautov) </details> --- --------- Co-authored-by: Dima Arnautov <dmitrii.arnautov@elastic.co> Co-authored-by: kibanamachine <42973632+kibanamachine@users.noreply.github.com> Co-authored-by: Kevin Delemme <kevin.delemme@elastic.co> Co-authored-by: Mike Côté <mikecote@users.noreply.github.com> Co-authored-by: Antonio <antonio.coelho@elastic.co> Co-authored-by: Kevin Delemme <kdelemme@gmail.com> Co-authored-by: Dominique Clarke <dominique.clarke@elastic.co> Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com> Co-authored-by: Jason Rhodes <jason.rhodes@elastic.co> Co-authored-by: Yiannis Nikolopoulos <yiannis.nikolopoulos@elastic.co> Co-authored-by: Alexi Doak <109488926+doakalexi@users.noreply.github.com> Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com> Co-authored-by: Bailey Cash <bailey.cash@elastic.co> Co-authored-by: Anna Davydova <ana.davydova@elastic.co> Co-authored-by: Umberto Pepato <umbopepato@users.noreply.github.com> Co-authored-by: Jason Rhodes <jason.matthew.rhodes@gmail.com> Co-authored-by: Joana Cardoso <169058851+joana-cps@users.noreply.github.com>
## Summary ### Key capabilities - **ES|QL-native rule evaluation** — Rules are defined as ES|QL queries with optional WHERE clause conditions, evaluated on a configurable schedule - **Alert lifecycle management** — Full episode tracking with pending → active → recovering → inactive state transitions, including configurable alert delay (consecutive breaches / duration) - **Event-driven architecture** — Alert events and actions are stored in dedicated data streams (`.alerting-events`, `.alerting-actions`) with ES|QL views for querying - **Notification dispatch pipeline** — A multi-step dispatcher that matches alert episodes to notification policies, handles throttling/suppression, and triggers Kibana Workflows using encrypted API keys - **Notification policies** — CRUD APIs and UI for creating notification policies with KQL-based rule matching, workflow integration, and API key management - **Rule authoring UI** — A shared rule form package (`@kbn/alerting-v2-rule-form`) usable standalone or embedded in Discover, with ES|QL editor, WHERE clause condition editing, recovery configuration, and live query preview - **Rule management UI** — Full rule list with pagination, enable/disable, clone, edit, and delete operations - **APM instrumentation** — Middleware and decorators for tracing rule execution and client operations ### Architecture highlights - **InversifyJS DI** — All services use constructor injection with typed tokens, scoped per-request or singleton as appropriate - **Pipeline pattern** — Rule executor and dispatcher use composable step-based pipelines - **Saved Objects** — Rules stored as hidden saved objects; notification policies stored as encrypted saved objects (for API key protection) - **Feature privileges** — Dedicated Kibana feature with read/all privileges for RBAC --- ## Contained PRs <details> <summary><strong>Core Engine & Plugin Init</strong> (12 PRs)</summary> - elastic#247283 — Init alerting v2 plugin (@cnasikas) - elastic#247452 — Add the alerting v2 feature privileges (@cnasikas) - elastic#247673 — Director (@cnasikas) - elastic#248306 — Create basic services (@cnasikas) - elastic#248696 — Initialize all resources (@cnasikas) - elastic#250023 — Schema package (@cnasikas) - elastic#250010 — YML Editor (@cnasikas) - elastic#251064 — Remove index.mode: lookup for RnA alert indices (@cnasikas) - elastic#251707 — Simplify task registration pattern (@kdelemme) - elastic#251876 — Dedicated user service (@cnasikas) - elastic#252073 — Use `kbn/data-streams` in alerting_v2 (@cnasikas) - elastic#255120 — Update alerting-v2 owner to new rna project team (@cnasikas) </details> <details> <summary><strong>Rule Execution Pipeline</strong> (12 PRs)</summary> - elastic#247472 — Add alerting v2 Rule Executor (@darnautov) - elastic#248285 — Alerting v2 rule HTTP APIs (@darnautov) - elastic#248728 — Add basic alert actions route (@darnautov) - elastic#250161 — Refactor rule executor to use a pipeline pattern (@darnautov) - elastic#252292 — Implement the CountTimeframeStrategy for the director (@cnasikas) - elastic#252544 — Add support of streaming in the rule executor (@darnautov) - elastic#252754 — Update rule attributes (@kdelemme) - elastic#253355 — Add getRules client method (@kdelemme) - elastic#253668 — Make evaluation.query.condition optional (@kdelemme) - elastic#254031 — Add recovery event generation to rule execution pipeline (@kdelemme) - elastic#255968 — ES&elastic#124;QL views (@adcoelho) - elastic#256697 — Create episodes ES&elastic#124;QL view (@adcoelho) </details> <details> <summary><strong>Alert Suppression & Episodes</strong> (3 PRs)</summary> - elastic#252174 — Alert suppression (@kdelemme) - elastic#256486 — Fix suppression query (@kdelemme) - elastic#256527 — Store 'unmatched' action for unmatched alert episodes (@kdelemme) </details> <details> <summary><strong>Dispatcher & Notification Engine</strong> (6 PRs)</summary> - elastic#250822 — Alerting v2 dispatcher (@kdelemme) - elastic#251529 — Use query service in dispatcher (@kdelemme) - elastic#251679 — Dispatcher task (@kdelemme) - elastic#252758 — Dispatcher notification policy (@kdelemme) - elastic#255332 — Wait for resources before scheduling dispatcher task (@kdelemme) - elastic#256536 — Use stored encrypted API keys from Notification Policy in dispatcher step (@kdelemme) </details> <details> <summary><strong>Notification Policies (Server)</strong> (4 PRs)</summary> - elastic#251336 — Introduce notification policy CRUD APIs and client (@cnasikas) - elastic#253134 — Update notification policy (@cnasikas) - elastic#254808 — Store API key owner on Notification Policy (@kdelemme) - elastic#256940 — Make notification policies global with optional rule-label scoping (@kdelemme) </details> <details> <summary><strong>Notification Policies UI</strong> (1 PR)</summary> - elastic#255599 — Add notification policies UI and Storybook form story (@adcoelho) </details> <details> <summary><strong>Rule Authoring UI</strong> (13 PRs)</summary> - elastic#250961 — Add create rule flyout in Discover (@adcoelho) - elastic#255111 — Add activation configuration fields to alerting V2 rule form (@yiannisnikolopoulos) - elastic#255427 — Rule form: provide services via context (@dominiqueclarke) - elastic#255876 — MVP rule form, Split evaluation condition, and Recovery configuration (@dominiqueclarke) - elastic#256260 — Foundational rule list (@dominiqueclarke) - elastic#256756 — Wire up edit flow (@dominiqueclarke) - elastic#256801 — Move consecutive breaches max to shared constants (@yiannisnikolopoulos) - elastic#256818 — Preview query and design parity (@dominiqueclarke) - elastic#256938 — Allow clearing number inputs in state transition fields (@yiannisnikolopoulos) - elastic#257017 — Add enable/disable and clone rule to rule list (@dominiqueclarke) - elastic#257246 — Remove all React.FC (@dominiqueclarke) - elastic#257415 — Rule form - fix test (@dominiqueclarke) - elastic#257454 — Block comma key in number input component (@yiannisnikolopoulos) </details> <details> <summary><strong>API Documentation & Schema</strong> (2 PRs)</summary> - elastic#254901 — Rename indexes for alert events and actions (@adcoelho) - elastic#255810 — OAS for alert action routes (@adcoelho) </details> <details> <summary><strong>Observability & Monitoring</strong> (3 PRs)</summary> - elastic#254925 — Add ApmMiddleware to the rule executor (@adcoelho) - elastic#255115 — Add the withAPM decorator and apply it to the rules_client (@adcoelho) - elastic#255999 — Fix linting problem in apm middleware (@adcoelho) </details> <details> <summary><strong>CI & Maintenance</strong> (2 PRs)</summary> - elastic#257409 — Refactor SO services to use inversify DI for client initialization (@darnautov) - Fix alerting-v2-schema jest config (@darnautov) </details> --- --------- Co-authored-by: Dima Arnautov <dmitrii.arnautov@elastic.co> Co-authored-by: kibanamachine <42973632+kibanamachine@users.noreply.github.com> Co-authored-by: Kevin Delemme <kevin.delemme@elastic.co> Co-authored-by: Mike Côté <mikecote@users.noreply.github.com> Co-authored-by: Antonio <antonio.coelho@elastic.co> Co-authored-by: Kevin Delemme <kdelemme@gmail.com> Co-authored-by: Dominique Clarke <dominique.clarke@elastic.co> Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com> Co-authored-by: Jason Rhodes <jason.rhodes@elastic.co> Co-authored-by: Yiannis Nikolopoulos <yiannis.nikolopoulos@elastic.co> Co-authored-by: Alexi Doak <109488926+doakalexi@users.noreply.github.com> Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com> Co-authored-by: Bailey Cash <bailey.cash@elastic.co> Co-authored-by: Anna Davydova <ana.davydova@elastic.co> Co-authored-by: Umberto Pepato <umbopepato@users.noreply.github.com> Co-authored-by: Jason Rhodes <jason.matthew.rhodes@gmail.com> Co-authored-by: Joana Cardoso <169058851+joana-cps@users.noreply.github.com>
- **ES|QL-native rule evaluation** — Rules are defined as ES|QL queries with optional WHERE clause conditions, evaluated on a configurable schedule - **Alert lifecycle management** — Full episode tracking with pending → active → recovering → inactive state transitions, including configurable alert delay (consecutive breaches / duration) - **Event-driven architecture** — Alert events and actions are stored in dedicated data streams (`.alerting-events`, `.alerting-actions`) with ES|QL views for querying - **Notification dispatch pipeline** — A multi-step dispatcher that matches alert episodes to notification policies, handles throttling/suppression, and triggers Kibana Workflows using encrypted API keys - **Notification policies** — CRUD APIs and UI for creating notification policies with KQL-based rule matching, workflow integration, and API key management - **Rule authoring UI** — A shared rule form package (`@kbn/alerting-v2-rule-form`) usable standalone or embedded in Discover, with ES|QL editor, WHERE clause condition editing, recovery configuration, and live query preview - **Rule management UI** — Full rule list with pagination, enable/disable, clone, edit, and delete operations - **APM instrumentation** — Middleware and decorators for tracing rule execution and client operations - **InversifyJS DI** — All services use constructor injection with typed tokens, scoped per-request or singleton as appropriate - **Pipeline pattern** — Rule executor and dispatcher use composable step-based pipelines - **Saved Objects** — Rules stored as hidden saved objects; notification policies stored as encrypted saved objects (for API key protection) - **Feature privileges** — Dedicated Kibana feature with read/all privileges for RBAC --- <details> <summary><strong>Core Engine & Plugin Init</strong> (12 PRs)</summary> - elastic#247283 — Init alerting v2 plugin (@cnasikas) - elastic#247452 — Add the alerting v2 feature privileges (@cnasikas) - elastic#247673 — Director (@cnasikas) - elastic#248306 — Create basic services (@cnasikas) - elastic#248696 — Initialize all resources (@cnasikas) - elastic#250023 — Schema package (@cnasikas) - elastic#250010 — YML Editor (@cnasikas) - elastic#251064 — Remove index.mode: lookup for RnA alert indices (@cnasikas) - elastic#251707 — Simplify task registration pattern (@kdelemme) - elastic#251876 — Dedicated user service (@cnasikas) - elastic#252073 — Use `kbn/data-streams` in alerting_v2 (@cnasikas) - elastic#255120 — Update alerting-v2 owner to new rna project team (@cnasikas) </details> <details> <summary><strong>Rule Execution Pipeline</strong> (12 PRs)</summary> - elastic#247472 — Add alerting v2 Rule Executor (@darnautov) - elastic#248285 — Alerting v2 rule HTTP APIs (@darnautov) - elastic#248728 — Add basic alert actions route (@darnautov) - elastic#250161 — Refactor rule executor to use a pipeline pattern (@darnautov) - elastic#252292 — Implement the CountTimeframeStrategy for the director (@cnasikas) - elastic#252544 — Add support of streaming in the rule executor (@darnautov) - elastic#252754 — Update rule attributes (@kdelemme) - elastic#253355 — Add getRules client method (@kdelemme) - elastic#253668 — Make evaluation.query.condition optional (@kdelemme) - elastic#254031 — Add recovery event generation to rule execution pipeline (@kdelemme) - elastic#255968 — ES&elastic#124;QL views (@adcoelho) - elastic#256697 — Create episodes ES&elastic#124;QL view (@adcoelho) </details> <details> <summary><strong>Alert Suppression & Episodes</strong> (3 PRs)</summary> - elastic#252174 — Alert suppression (@kdelemme) - elastic#256486 — Fix suppression query (@kdelemme) - elastic#256527 — Store 'unmatched' action for unmatched alert episodes (@kdelemme) </details> <details> <summary><strong>Dispatcher & Notification Engine</strong> (6 PRs)</summary> - elastic#250822 — Alerting v2 dispatcher (@kdelemme) - elastic#251529 — Use query service in dispatcher (@kdelemme) - elastic#251679 — Dispatcher task (@kdelemme) - elastic#252758 — Dispatcher notification policy (@kdelemme) - elastic#255332 — Wait for resources before scheduling dispatcher task (@kdelemme) - elastic#256536 — Use stored encrypted API keys from Notification Policy in dispatcher step (@kdelemme) </details> <details> <summary><strong>Notification Policies (Server)</strong> (4 PRs)</summary> - elastic#251336 — Introduce notification policy CRUD APIs and client (@cnasikas) - elastic#253134 — Update notification policy (@cnasikas) - elastic#254808 — Store API key owner on Notification Policy (@kdelemme) - elastic#256940 — Make notification policies global with optional rule-label scoping (@kdelemme) </details> <details> <summary><strong>Notification Policies UI</strong> (1 PR)</summary> - elastic#255599 — Add notification policies UI and Storybook form story (@adcoelho) </details> <details> <summary><strong>Rule Authoring UI</strong> (13 PRs)</summary> - elastic#250961 — Add create rule flyout in Discover (@adcoelho) - elastic#255111 — Add activation configuration fields to alerting V2 rule form (@yiannisnikolopoulos) - elastic#255427 — Rule form: provide services via context (@dominiqueclarke) - elastic#255876 — MVP rule form, Split evaluation condition, and Recovery configuration (@dominiqueclarke) - elastic#256260 — Foundational rule list (@dominiqueclarke) - elastic#256756 — Wire up edit flow (@dominiqueclarke) - elastic#256801 — Move consecutive breaches max to shared constants (@yiannisnikolopoulos) - elastic#256818 — Preview query and design parity (@dominiqueclarke) - elastic#256938 — Allow clearing number inputs in state transition fields (@yiannisnikolopoulos) - elastic#257017 — Add enable/disable and clone rule to rule list (@dominiqueclarke) - elastic#257246 — Remove all React.FC (@dominiqueclarke) - elastic#257415 — Rule form - fix test (@dominiqueclarke) - elastic#257454 — Block comma key in number input component (@yiannisnikolopoulos) </details> <details> <summary><strong>API Documentation & Schema</strong> (2 PRs)</summary> - elastic#254901 — Rename indexes for alert events and actions (@adcoelho) - elastic#255810 — OAS for alert action routes (@adcoelho) </details> <details> <summary><strong>Observability & Monitoring</strong> (3 PRs)</summary> - elastic#254925 — Add ApmMiddleware to the rule executor (@adcoelho) - elastic#255115 — Add the withAPM decorator and apply it to the rules_client (@adcoelho) - elastic#255999 — Fix linting problem in apm middleware (@adcoelho) </details> <details> <summary><strong>CI & Maintenance</strong> (2 PRs)</summary> - elastic#257409 — Refactor SO services to use inversify DI for client initialization (@darnautov) - Fix alerting-v2-schema jest config (@darnautov) </details> --- --------- Co-authored-by: Dima Arnautov <dmitrii.arnautov@elastic.co> Co-authored-by: kibanamachine <42973632+kibanamachine@users.noreply.github.com> Co-authored-by: Kevin Delemme <kevin.delemme@elastic.co> Co-authored-by: Mike Côté <mikecote@users.noreply.github.com> Co-authored-by: Antonio <antonio.coelho@elastic.co> Co-authored-by: Kevin Delemme <kdelemme@gmail.com> Co-authored-by: Dominique Clarke <dominique.clarke@elastic.co> Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com> Co-authored-by: Jason Rhodes <jason.rhodes@elastic.co> Co-authored-by: Yiannis Nikolopoulos <yiannis.nikolopoulos@elastic.co> Co-authored-by: Alexi Doak <109488926+doakalexi@users.noreply.github.com> Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com> Co-authored-by: Bailey Cash <bailey.cash@elastic.co> Co-authored-by: Anna Davydova <ana.davydova@elastic.co> Co-authored-by: Umberto Pepato <umbopepato@users.noreply.github.com> Co-authored-by: Jason Rhodes <jason.matthew.rhodes@gmail.com> Co-authored-by: Joana Cardoso <169058851+joana-cps@users.noreply.github.com>
- **ES|QL-native rule evaluation** — Rules are defined as ES|QL queries with optional WHERE clause conditions, evaluated on a configurable schedule - **Alert lifecycle management** — Full episode tracking with pending → active → recovering → inactive state transitions, including configurable alert delay (consecutive breaches / duration) - **Event-driven architecture** — Alert events and actions are stored in dedicated data streams (`.alerting-events`, `.alerting-actions`) with ES|QL views for querying - **Notification dispatch pipeline** — A multi-step dispatcher that matches alert episodes to notification policies, handles throttling/suppression, and triggers Kibana Workflows using encrypted API keys - **Notification policies** — CRUD APIs and UI for creating notification policies with KQL-based rule matching, workflow integration, and API key management - **Rule authoring UI** — A shared rule form package (`@kbn/alerting-v2-rule-form`) usable standalone or embedded in Discover, with ES|QL editor, WHERE clause condition editing, recovery configuration, and live query preview - **Rule management UI** — Full rule list with pagination, enable/disable, clone, edit, and delete operations - **APM instrumentation** — Middleware and decorators for tracing rule execution and client operations - **InversifyJS DI** — All services use constructor injection with typed tokens, scoped per-request or singleton as appropriate - **Pipeline pattern** — Rule executor and dispatcher use composable step-based pipelines - **Saved Objects** — Rules stored as hidden saved objects; notification policies stored as encrypted saved objects (for API key protection) - **Feature privileges** — Dedicated Kibana feature with read/all privileges for RBAC --- <details> <summary><strong>Core Engine & Plugin Init</strong> (12 PRs)</summary> - elastic#247283 — Init alerting v2 plugin (@cnasikas) - elastic#247452 — Add the alerting v2 feature privileges (@cnasikas) - elastic#247673 — Director (@cnasikas) - elastic#248306 — Create basic services (@cnasikas) - elastic#248696 — Initialize all resources (@cnasikas) - elastic#250023 — Schema package (@cnasikas) - elastic#250010 — YML Editor (@cnasikas) - elastic#251064 — Remove index.mode: lookup for RnA alert indices (@cnasikas) - elastic#251707 — Simplify task registration pattern (@kdelemme) - elastic#251876 — Dedicated user service (@cnasikas) - elastic#252073 — Use `kbn/data-streams` in alerting_v2 (@cnasikas) - elastic#255120 — Update alerting-v2 owner to new rna project team (@cnasikas) </details> <details> <summary><strong>Rule Execution Pipeline</strong> (12 PRs)</summary> - elastic#247472 — Add alerting v2 Rule Executor (@darnautov) - elastic#248285 — Alerting v2 rule HTTP APIs (@darnautov) - elastic#248728 — Add basic alert actions route (@darnautov) - elastic#250161 — Refactor rule executor to use a pipeline pattern (@darnautov) - elastic#252292 — Implement the CountTimeframeStrategy for the director (@cnasikas) - elastic#252544 — Add support of streaming in the rule executor (@darnautov) - elastic#252754 — Update rule attributes (@kdelemme) - elastic#253355 — Add getRules client method (@kdelemme) - elastic#253668 — Make evaluation.query.condition optional (@kdelemme) - elastic#254031 — Add recovery event generation to rule execution pipeline (@kdelemme) - elastic#255968 — ES&elastic#124;QL views (@adcoelho) - elastic#256697 — Create episodes ES&elastic#124;QL view (@adcoelho) </details> <details> <summary><strong>Alert Suppression & Episodes</strong> (3 PRs)</summary> - elastic#252174 — Alert suppression (@kdelemme) - elastic#256486 — Fix suppression query (@kdelemme) - elastic#256527 — Store 'unmatched' action for unmatched alert episodes (@kdelemme) </details> <details> <summary><strong>Dispatcher & Notification Engine</strong> (6 PRs)</summary> - elastic#250822 — Alerting v2 dispatcher (@kdelemme) - elastic#251529 — Use query service in dispatcher (@kdelemme) - elastic#251679 — Dispatcher task (@kdelemme) - elastic#252758 — Dispatcher notification policy (@kdelemme) - elastic#255332 — Wait for resources before scheduling dispatcher task (@kdelemme) - elastic#256536 — Use stored encrypted API keys from Notification Policy in dispatcher step (@kdelemme) </details> <details> <summary><strong>Notification Policies (Server)</strong> (4 PRs)</summary> - elastic#251336 — Introduce notification policy CRUD APIs and client (@cnasikas) - elastic#253134 — Update notification policy (@cnasikas) - elastic#254808 — Store API key owner on Notification Policy (@kdelemme) - elastic#256940 — Make notification policies global with optional rule-label scoping (@kdelemme) </details> <details> <summary><strong>Notification Policies UI</strong> (1 PR)</summary> - elastic#255599 — Add notification policies UI and Storybook form story (@adcoelho) </details> <details> <summary><strong>Rule Authoring UI</strong> (13 PRs)</summary> - elastic#250961 — Add create rule flyout in Discover (@adcoelho) - elastic#255111 — Add activation configuration fields to alerting V2 rule form (@yiannisnikolopoulos) - elastic#255427 — Rule form: provide services via context (@dominiqueclarke) - elastic#255876 — MVP rule form, Split evaluation condition, and Recovery configuration (@dominiqueclarke) - elastic#256260 — Foundational rule list (@dominiqueclarke) - elastic#256756 — Wire up edit flow (@dominiqueclarke) - elastic#256801 — Move consecutive breaches max to shared constants (@yiannisnikolopoulos) - elastic#256818 — Preview query and design parity (@dominiqueclarke) - elastic#256938 — Allow clearing number inputs in state transition fields (@yiannisnikolopoulos) - elastic#257017 — Add enable/disable and clone rule to rule list (@dominiqueclarke) - elastic#257246 — Remove all React.FC (@dominiqueclarke) - elastic#257415 — Rule form - fix test (@dominiqueclarke) - elastic#257454 — Block comma key in number input component (@yiannisnikolopoulos) </details> <details> <summary><strong>API Documentation & Schema</strong> (2 PRs)</summary> - elastic#254901 — Rename indexes for alert events and actions (@adcoelho) - elastic#255810 — OAS for alert action routes (@adcoelho) </details> <details> <summary><strong>Observability & Monitoring</strong> (3 PRs)</summary> - elastic#254925 — Add ApmMiddleware to the rule executor (@adcoelho) - elastic#255115 — Add the withAPM decorator and apply it to the rules_client (@adcoelho) - elastic#255999 — Fix linting problem in apm middleware (@adcoelho) </details> <details> <summary><strong>CI & Maintenance</strong> (2 PRs)</summary> - elastic#257409 — Refactor SO services to use inversify DI for client initialization (@darnautov) - Fix alerting-v2-schema jest config (@darnautov) </details> --- --------- Co-authored-by: Dima Arnautov <dmitrii.arnautov@elastic.co> Co-authored-by: kibanamachine <42973632+kibanamachine@users.noreply.github.com> Co-authored-by: Kevin Delemme <kevin.delemme@elastic.co> Co-authored-by: Mike Côté <mikecote@users.noreply.github.com> Co-authored-by: Antonio <antonio.coelho@elastic.co> Co-authored-by: Kevin Delemme <kdelemme@gmail.com> Co-authored-by: Dominique Clarke <dominique.clarke@elastic.co> Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com> Co-authored-by: Jason Rhodes <jason.rhodes@elastic.co> Co-authored-by: Yiannis Nikolopoulos <yiannis.nikolopoulos@elastic.co> Co-authored-by: Alexi Doak <109488926+doakalexi@users.noreply.github.com> Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com> Co-authored-by: Bailey Cash <bailey.cash@elastic.co> Co-authored-by: Anna Davydova <ana.davydova@elastic.co> Co-authored-by: Umberto Pepato <umbopepato@users.noreply.github.com> Co-authored-by: Jason Rhodes <jason.matthew.rhodes@gmail.com> Co-authored-by: Joana Cardoso <169058851+joana-cps@users.noreply.github.com>
## Summary ### Key capabilities - **ES|QL-native rule evaluation** — Rules are defined as ES|QL queries with optional WHERE clause conditions, evaluated on a configurable schedule - **Alert lifecycle management** — Full episode tracking with pending → active → recovering → inactive state transitions, including configurable alert delay (consecutive breaches / duration) - **Event-driven architecture** — Alert events and actions are stored in dedicated data streams (`.alerting-events`, `.alerting-actions`) with ES|QL views for querying - **Notification dispatch pipeline** — A multi-step dispatcher that matches alert episodes to notification policies, handles throttling/suppression, and triggers Kibana Workflows using encrypted API keys - **Notification policies** — CRUD APIs and UI for creating notification policies with KQL-based rule matching, workflow integration, and API key management - **Rule authoring UI** — A shared rule form package (`@kbn/alerting-v2-rule-form`) usable standalone or embedded in Discover, with ES|QL editor, WHERE clause condition editing, recovery configuration, and live query preview - **Rule management UI** — Full rule list with pagination, enable/disable, clone, edit, and delete operations - **APM instrumentation** — Middleware and decorators for tracing rule execution and client operations ### Architecture highlights - **InversifyJS DI** — All services use constructor injection with typed tokens, scoped per-request or singleton as appropriate - **Pipeline pattern** — Rule executor and dispatcher use composable step-based pipelines - **Saved Objects** — Rules stored as hidden saved objects; notification policies stored as encrypted saved objects (for API key protection) - **Feature privileges** — Dedicated Kibana feature with read/all privileges for RBAC --- ## Contained PRs <details> <summary><strong>Core Engine & Plugin Init</strong> (12 PRs)</summary> - elastic#247283 — Init alerting v2 plugin (@cnasikas) - elastic#247452 — Add the alerting v2 feature privileges (@cnasikas) - elastic#247673 — Director (@cnasikas) - elastic#248306 — Create basic services (@cnasikas) - elastic#248696 — Initialize all resources (@cnasikas) - elastic#250023 — Schema package (@cnasikas) - elastic#250010 — YML Editor (@cnasikas) - elastic#251064 — Remove index.mode: lookup for RnA alert indices (@cnasikas) - elastic#251707 — Simplify task registration pattern (@kdelemme) - elastic#251876 — Dedicated user service (@cnasikas) - elastic#252073 — Use `kbn/data-streams` in alerting_v2 (@cnasikas) - elastic#255120 — Update alerting-v2 owner to new rna project team (@cnasikas) </details> <details> <summary><strong>Rule Execution Pipeline</strong> (12 PRs)</summary> - elastic#247472 — Add alerting v2 Rule Executor (@darnautov) - elastic#248285 — Alerting v2 rule HTTP APIs (@darnautov) - elastic#248728 — Add basic alert actions route (@darnautov) - elastic#250161 — Refactor rule executor to use a pipeline pattern (@darnautov) - elastic#252292 — Implement the CountTimeframeStrategy for the director (@cnasikas) - elastic#252544 — Add support of streaming in the rule executor (@darnautov) - elastic#252754 — Update rule attributes (@kdelemme) - elastic#253355 — Add getRules client method (@kdelemme) - elastic#253668 — Make evaluation.query.condition optional (@kdelemme) - elastic#254031 — Add recovery event generation to rule execution pipeline (@kdelemme) - elastic#255968 — ES&elastic#124;QL views (@adcoelho) - elastic#256697 — Create episodes ES&elastic#124;QL view (@adcoelho) </details> <details> <summary><strong>Alert Suppression & Episodes</strong> (3 PRs)</summary> - elastic#252174 — Alert suppression (@kdelemme) - elastic#256486 — Fix suppression query (@kdelemme) - elastic#256527 — Store 'unmatched' action for unmatched alert episodes (@kdelemme) </details> <details> <summary><strong>Dispatcher & Notification Engine</strong> (6 PRs)</summary> - elastic#250822 — Alerting v2 dispatcher (@kdelemme) - elastic#251529 — Use query service in dispatcher (@kdelemme) - elastic#251679 — Dispatcher task (@kdelemme) - elastic#252758 — Dispatcher notification policy (@kdelemme) - elastic#255332 — Wait for resources before scheduling dispatcher task (@kdelemme) - elastic#256536 — Use stored encrypted API keys from Notification Policy in dispatcher step (@kdelemme) </details> <details> <summary><strong>Notification Policies (Server)</strong> (4 PRs)</summary> - elastic#251336 — Introduce notification policy CRUD APIs and client (@cnasikas) - elastic#253134 — Update notification policy (@cnasikas) - elastic#254808 — Store API key owner on Notification Policy (@kdelemme) - elastic#256940 — Make notification policies global with optional rule-label scoping (@kdelemme) </details> <details> <summary><strong>Notification Policies UI</strong> (1 PR)</summary> - elastic#255599 — Add notification policies UI and Storybook form story (@adcoelho) </details> <details> <summary><strong>Rule Authoring UI</strong> (13 PRs)</summary> - elastic#250961 — Add create rule flyout in Discover (@adcoelho) - elastic#255111 — Add activation configuration fields to alerting V2 rule form (@yiannisnikolopoulos) - elastic#255427 — Rule form: provide services via context (@dominiqueclarke) - elastic#255876 — MVP rule form, Split evaluation condition, and Recovery configuration (@dominiqueclarke) - elastic#256260 — Foundational rule list (@dominiqueclarke) - elastic#256756 — Wire up edit flow (@dominiqueclarke) - elastic#256801 — Move consecutive breaches max to shared constants (@yiannisnikolopoulos) - elastic#256818 — Preview query and design parity (@dominiqueclarke) - elastic#256938 — Allow clearing number inputs in state transition fields (@yiannisnikolopoulos) - elastic#257017 — Add enable/disable and clone rule to rule list (@dominiqueclarke) - elastic#257246 — Remove all React.FC (@dominiqueclarke) - elastic#257415 — Rule form - fix test (@dominiqueclarke) - elastic#257454 — Block comma key in number input component (@yiannisnikolopoulos) </details> <details> <summary><strong>API Documentation & Schema</strong> (2 PRs)</summary> - elastic#254901 — Rename indexes for alert events and actions (@adcoelho) - elastic#255810 — OAS for alert action routes (@adcoelho) </details> <details> <summary><strong>Observability & Monitoring</strong> (3 PRs)</summary> - elastic#254925 — Add ApmMiddleware to the rule executor (@adcoelho) - elastic#255115 — Add the withAPM decorator and apply it to the rules_client (@adcoelho) - elastic#255999 — Fix linting problem in apm middleware (@adcoelho) </details> <details> <summary><strong>CI & Maintenance</strong> (2 PRs)</summary> - elastic#257409 — Refactor SO services to use inversify DI for client initialization (@darnautov) - Fix alerting-v2-schema jest config (@darnautov) </details> --- --------- Co-authored-by: Dima Arnautov <dmitrii.arnautov@elastic.co> Co-authored-by: kibanamachine <42973632+kibanamachine@users.noreply.github.com> Co-authored-by: Kevin Delemme <kevin.delemme@elastic.co> Co-authored-by: Mike Côté <mikecote@users.noreply.github.com> Co-authored-by: Antonio <antonio.coelho@elastic.co> Co-authored-by: Kevin Delemme <kdelemme@gmail.com> Co-authored-by: Dominique Clarke <dominique.clarke@elastic.co> Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com> Co-authored-by: Jason Rhodes <jason.rhodes@elastic.co> Co-authored-by: Yiannis Nikolopoulos <yiannis.nikolopoulos@elastic.co> Co-authored-by: Alexi Doak <109488926+doakalexi@users.noreply.github.com> Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com> Co-authored-by: Bailey Cash <bailey.cash@elastic.co> Co-authored-by: Anna Davydova <ana.davydova@elastic.co> Co-authored-by: Umberto Pepato <umbopepato@users.noreply.github.com> Co-authored-by: Jason Rhodes <jason.matthew.rhodes@gmail.com> Co-authored-by: Joana Cardoso <169058851+joana-cps@users.noreply.github.com>
Resolves https://github.com/elastic/rna-program/issues/117
Resolves https://github.com/elastic/rna-program/issues/118
Dispatcher and Notification Policy Architecture
The dispatcher is the asynchronous component that bridges rule execution (alert-event production) and notification delivery. It runs as a Task Manager task on a fixed 5-second polling interval, processes batches of up to 10,000 alert episodes (due to ES|QL limits), applies suppression and throttling logic, evaluates notification policies, and dispatches matched episodes to workflows. All decisions are recorded back to
.alerts-actionsso subsequent runs skip already-handled episodes.We leverage a pipeline pattern like the Rule Executor
Out of scope for this PR
Decision that can be revisited
Dispatcher pipeline
The dispatcher executes a 10-step pipeline on every tick. The steps are sequential -- each depends on the output of the previous one.
flowchart TD S1["1. Fetch alert episodes"] --> S2["2. Fetch suppressions"] S2 --> S3["3. Apply suppressions"] S3 --> S3a["suppressed episodes"] S3 --> S3b["dispatach-able episodes"] S3b --> S4["4. Fetch rules"] S4 --> S5["5. Fetch notification policies"] S5 --> S6["6. Evaluate matchers"] S6 --> S7["7. Build notification groups"] S7 --> S8["8. Apply throttling"] S8 --> S8a["throttled groups"] S8 --> S8b["dispatch groups"] S8b --> S9["9. Dispatch to workflows"] S9 --> S10["10. Record outcomes to .alerts-actions"] S3a --> S10 S8a --> S10Detailed steps
Step 1: Fetch alert episodes
Runs
getDispatchableAlertEventsQuery()via ES|QL against both.alerts-eventsand.alerts-actions. The query usesINLINE STATSto compute thelast_firedtimestamp per(rule_id, group_hash)from existingfire/suppressactions, then filters to only include alert events newer thanlast_fired. A lookback window ofpreviousStartedAt - 10 minutesis applied as a time range filter. Results are sorted bylast_event_timestampascending and capped at 10,000 rows.Output:
AlertEpisode[](up to 10,000).Step 2: Fetch suppressions
Using the alert episodes from step 1, runs
getAlertEpisodeSuppressionsQuery()against.alerts-actionsonly. This resolves the latest ack/unack, deactivate/activate, and snooze/unsnooze state per(rule_id, group_hash, episode_id).Output:
AlertEpisodeSuppression[].Step 3: Apply suppressions
Merges suppression results into the alert episodes. An episode is suppressed if any of the three suppression types apply:
(rule_id, group_hash, episode_id)last_ack_action == "ack"(rule_id, group_hash, episode_id)last_deactivate_action == "deactivate"(rule_id, group_hash)last_snooze_action == "snooze"ANDexpiry > episode.last_event_timestampSnooze is scoped to the entire series (all episodes under a
group_hash), while ack and deactivate are scoped to individual episodes.Output: Two arrays --
suppressed[](with reason) andactive[].Step 4: Fetch rules
Extracts unique
rule_idvalues from the active episodes and batch-fetches the rule objects.Output:
Map<string, Rule>keyed byrule_id.Step 5: Fetch notification policies
Extracts unique notification policy IDs from the fetched rules and batch-fetches the policy objects.
Output:
Map<string, NotificationPolicy>keyed bypolicy_id.Step 6: Evaluate matchers
For each active episode, iterates through the notification policies associated with its rule. A policy with an empty/undefined
matcheris a catch-all that matches every episode. When a matcher is defined, it is evaluated as a KQL expression against the episode object usingevaluateKql()from@kbn/eval-kql.Supported KQL features:
episode_status: activegroup_hash: critical*AND,OR,NOTdata.severity: criticaltimestamp >= now-1hAn episode can match multiple policies (from the same rule). Each match is handled independently -- the episode may be dispatched to multiple workflows.
Output:
MatchedPair[](episode + policy pairs).Step 7: Build notification groups
Groups matched pairs by
(rule_id, policyId, groupKey). ThenotificationGroupIdis anobjectHashof these three fields, used as a stable identifier for throttle tracking.When
groupByis[](current default), each episode dispatches individually -- the group key is set to{ groupHash, episodeId }, meaning oneNotificationGroupper episode.Episodes from different rules are never merged into the same group, even if they share the same policy.
Output:
NotificationGroup[].Step 8: Apply throttling
For each notification group, checks whether a notification was already sent within the policy's throttle interval. The dispatcher queries
.alerts-actionsforaction_type: "notified"records matching thenotification_group_id.Throttling is per
(rule_id, policy_id, group_key). The throttle interval defines the minimum gap between consecutive notifications for the same combination.The interval resets from the timestamp of the last dispatched notification (dispatch-time based), not from the first occurrence.
Output: Two arrays --
dispatch[]andthrottled[].Step 9: Dispatch to workflows
For each non-throttled notification group, invokes the target workflow with the group's episodes as payload. Not yet implemented.
Step 10: Record outcomes
Bulk-writes action documents to
.alerts-actionsto record what happened to each episode:suppressfor each episode from step 3 (ack/deactivate/snooze) -- with reasonsuppressfor each episode in every throttled group from step 8 -- reason:"suppressed by throttled policy {policyId}"firefor each episode in every dispatched group -- reason:"dispatched by policy {policyId}"notifiedfor each dispatched group itself -- used for throttle tracking on subsequent runs, keyed bynotification_group_idThe
fireandsuppressactions carrylast_series_event_timestampto feed back into the dispatcher query's fire filter, ensuring already-processed episodes are excluded on the next tick.Notification policy
A notification policy defines how and when to dispatch alert episodes to a workflow. Each rule references one or more notification policies via
notification_policies.Fields
idstringnamestringdescriptionstringmatcherstring?groupBystring[]data.*fields to group episodes;[]= one notification per episodethrottle.intervalstring?"1h","5m")workflowIdstringMatcher evaluation
The
matcheris a KQL (Kibana Query Language) expression evaluated against the episode's fields usingevaluateKql()from@kbn/eval-kql. This supports:episode_status: activegroup_hash: prod-*episode_status: active AND rule_id: rule-001data.severity: criticaltimestamp >= now-1hIf the matcher is empty or undefined, the policy acts as a catch-all and matches all episodes.
Grouping semantics
When
groupByis[], each episode becomes its ownNotificationGroupwith a group key of{ groupHash, episodeId }. WhengroupBycontains field names, episodes sharing the same values for those fields (within the same rule) are batched into one notification group.Grouping is always scoped per rule -- episodes from different rules are never merged.
Throttle semantics
Throttling is per
(rule_id, policy_id, group_key), tracked vianotifiedaction records in.alerts-actions. Thenotification_group_idis anobjectHashof these three fields, providing a stable key across dispatcher runs.The first time a group is seen (no prior
notifiedaction), it always fires. Subsequent occurrences within the throttle window are suppressed. The window resets from the dispatch time of the last successful notification.Outcome recording and
.alerts-actionsschemaThe dispatcher writes three action types to
.alerts-actions:action_type: "fire"
Written for each episode dispatched to at least one workflow. Used by the fire filter to avoid reprocessing.
{ "@timestamp": "2026-01-27T16:16:00.000Z", "actor": "system", "action_type": "fire", "rule_id": "rule-001", "group_hash": "rule-001-series-1", "last_series_event_timestamp": "2026-01-27T16:15:00.000Z", "source": "internal", "reason": "dispatched by policy policy-A" }action_type: "suppress"
Written for each episode suppressed by ack, deactivate, snooze, or throttle.
{ "@timestamp": "2026-01-27T16:16:00.000Z", "actor": "system", "action_type": "suppress", "rule_id": "rule-002", "group_hash": "rule-002-series-1", "last_series_event_timestamp": "2026-01-27T16:15:00.000Z", "source": "internal", "reason": "ack" }action_type: "notified"
Written once per dispatched notification group (not per episode). Used for throttle lookups on subsequent runs.
{ "@timestamp": "2026-01-27T16:16:00.000Z", "actor": "system", "action_type": "notified", "rule_id": "rule-001", "group_hash": "irrelevant", "last_series_event_timestamp": "2026-01-27T16:16:00.000Z", "notification_group_id": "abc123hash", "source": "internal", "reason": "notified by policy policy-A with throttle interval" }