A monitoring cluster running Elasticsearch 7.0.0+ should never see `.monitoring-*-2-*` indices being created by the production cluster writing to it. So the Kibana Monitoring UI doesn't need to search against these indices any more.
* [Maps] support elasticsearch documents containing array of geometry values
* clean up
* use reduce instead of map and flatten
* replace map and flatten with reduce in geoShapeToGeometry
Plugins loaded by the `@kbn/interpreter` can sometimes setup global values while loading, like the regeneratorRuntime for instance, but the current plugin loading code is deleting every global that was added during plugin load. This changes the logic to only cleanup the `canvas` global after loading the canvas plugins.
resolves https://github.com/elastic/kibana/issues/27162
This adds logic so that the code will wait until any previous
attempt to send telemetry causes follow-up attempts to wait until
it has completed sending (in the unlikely event that it takes
longer than a minute, we don't want to waste time / resources).
* Fix small tipo
* [CCR] Add API to fetch ccr stats
* Add deserialization for recentAutoFollowErrors & autoFollowedClusters
* Add UI to display recent auto-follow pattern errors
* chore: remove asynchronous element selection
* chore: remove duplication between duplicateElement and rawDuplicateElement
* chore: renamed rawDuplicateElement to insertNodes (pure rename, not yet set based)
* perf: bulk node insertion
* chore: removed unnecessary element id duplications
* fix: supply empty array if preexisting workbook had none
* update layers on kibana global refresh interval
* store refresh in UI state
* clearer function names for refresh timer
* rename a bunch of things
* fix store_actions jest test
* add functional test to verify refresh config is loaded from saved object state
* add functional tests verifing layers re-fetch data on refresh timer
When cloning a job, the multi metric wizard would re-estimate the model memory limit and overwrite a possible customized limit. This PR fixes it by only setting the estimated limit when certain conditions apply: It only gets set to the estimation if the current limit is either the default value or the value of the previous estimation. That's our best guess if the value hasn't been customized. The check doesn't get it if the user intentionally for whatever reason (re)set the value to either the default or pervious estimate.
* Convert concat_stream_providers to jest
* Convert concat_stream to jest
* Convert intersperse_stream to jest
* Convert json_parse_stream and json_stringify_stream to jest
* Convert list_stream to jest
* Convert map_stream to jest
* Convert promise_from_streams to jest
* Convert reduce_stream to jest
* Convert replace_stream to jest
* Convert split_stream to jest
* Use jest built in error handling
Fixes a regression where the advanced job creation wizard wouldn't trigger the watch creation flyout in the jobs list.
The advanced job creation wizard is still based on angularjs and triggered broadcast events to trigger the watch creation, but with the new React based jobs list the listeners for those events were gone.
This PR fixes it by passing on the jobs list wrapping angular scope to be able to subscribe to the events from the React based JobsListView component.
Passing on angular's scope down to React components should be considered a workaround so we don't have to refactor all affected code to use another event system. Once the advanced job wizard gets ported to React too this should be revisited.
