If nodes drop and .watches / .triggered_watches shards are available after those shards were started a new cluster state update will come along that triggers the start watcher logic.
Original commit: elastic/x-pack-elasticsearch@af36f8b078
Before if timing permitted it a start and stop could be executed very close to each other time wise. This could make the manual stopped check seem to be valid, because the stop hadn't yet been performed.
Original commit: elastic/x-pack-elasticsearch@c1865c1069
This enables different constructs (primarily scripts) to set variables that can be access by subsequent constructs throughout the wathc execution. These variables are scoped to a single execution, that is, they are not persisted across multiple executions of the same watch.
Closeselastic/elasticsearch#589
Original commit: elastic/x-pack-elasticsearch@34223d1991
There are a lot of settigns which we don't need at all or are already
defined in the parent. For instance dependencies are not needed if
they are included in elasticsearch-parent.
This commit also removes the shading which we don't do anymore in
core and we include guava already.
Original commit: elastic/x-pack-elasticsearch@c4f951a751
Upgrade to Elasticsearch 2.0.0-SNAPSHOT
This moves the master branch to follow Elasticsearch 2.0.0-SNAPSHOT and fixes most problems that occurred during the upgrade. The remaining issues not yet fixed are:
* `HttpClient` and the `Account` used for Email support need to install security manager which is not supported by the elasticsearch security policy. This is not yet resolved an requires fundamental changes and/or a rule in the core policy file. See elastic/elasticsearch#597
* Due to changes to the way Time/Byte settings are parsed settings without a unit must be upgraded. See elastic/elasticsearch#598
* REST tests are currently disabled due to some limitations from Elasticsearch core that don't allow to run 3rd party REST tests. See elastic/elasticsearch#599
Watcher now also inherits the elaticsearch-parent pom file and all it's properties.
Original commit: elastic/x-pack-elasticsearch@1e03234e3e
The execution service will throw an exection if we acquire a watch lock or add a watch to the current executions if Watcher has stopped or is stopping. We should deal with this sitation by properly catching those failures. FYI if a watch managed to put itself to the current executions a shutdown will wait until a watch execution is finished.
Original commit: elastic/x-pack-elasticsearch@5692316072
This change allows the specification of a watch inline to the `_execute` API.
This watch id will not be persisted to the index and if record_execution is set to true it will result in an error.
The internal id `_anonymous_` will be used for the watch id and will be the watch id in the watch record.
Original commit: elastic/x-pack-elasticsearch@00e32c3838
Until now, if the transform failed (either on the watch or action level), an exception would be thrown and it would be captured globally on the watch execution and in the watch record message
This commit bring the error to the transform result
- A new `status` field was added to the transform result. Can either have `success` or `failure` values. When set to `failure` a `reason` field will hold the failure message.
- The `ExecutionService` changed to enable this functionality. Mainly, instead of relying on exception, during the execution the transform result status is checked and if failed the watch execution is aborted.
Original commit: elastic/x-pack-elasticsearch@65b7f51f00
Until now the email sanitization was fixed and could not be configured. That means that if ppl wanted a feature and our sanitization didn't support it, they were forced to disable sanitizaion all together.
In this commit a new `HtmlSanitizer` construct was introduced that is bound by Guice and can be configured via the settings.
Closeselastic/elasticsearch#586
Original commit: elastic/x-pack-elasticsearch@0081d1bf41
Until today connection and read timeout for http was not directly supported. This means that without setting oracle specific system properties at startup, calling a bad http service would by default hold the watch executing thread forever... is niet goed!!!
This commit introduces connection & read timeouts.
- Connection timeouts are timeouts for setting up the connection
- Read timeouts are timeouts waiting for data to be read
By default both timeouts are set to 10 seconds (overriding the default jdk to indefinite). It is possible to customize the default timeouts by settings `watcher.http.default_connection_timeout` and `watcher.http.default_read_timeout` settings).
It is also possible to override these defaults per http request, meaning, per webhook and http input configuration in the watch.
Original commit: elastic/x-pack-elasticsearch@224f50bc8b
Until now if a condition failed to execute (for whatever reason), an exception would be thrown and the watch would just abort. The problem with that approach is while the error message would have been logged in the watch record message, the result of the condition would have been lost.
This commit moves the condition execution error to the condition result (just like we have it today with actions and inputs).
- A `status` field was added to the condition result (can either be `success` or `failure`)
- A `reason` field is added to the condition result in case its status is `failure`
- If the condition result status is `failure`, the watch execution is aborted
Updated the rest APIs to verify the status & type of both the `input` and `condition` results on execution.
Original commit: elastic/x-pack-elasticsearch@dddca03ff5
Until now, if the input failed, an exception would be thrown and it would be captured globally on the watch execution and in the watch recod message. The problem with this approach is that the information about the input is lost. In this commit, the failure is returned as part of the input result.
- A new `status` field was added to the input result. Can either have `success` or `failure` values. When set to `failure` a `reason` field will be set with the error message.
- The `ExecutionService` changed to enable this functionality. Mainly, instead of relying on exception, during the execution the input result is checked for its status and the execution is aborted on failure. Also, the two places where the watch execution is handled were consolidated to a single method `execute(WatchExecutionContext)`.
- Also, the watch execution context id (which will end up being the `watch_record` id) was added the the context model (accessible via scripts and templates). This is done mainly for debugging purposes.
Original commit: elastic/x-pack-elasticsearch@e2567deada
There may be current executions still going on during stopping.
We should try to wait for the current executions to have completed.
Otherwise we can run into a situation where we didn't delete the watch from the .triggered_watches index,
but did insert into the history index. Upon start this can lead to DocumentAlreadyExistsException,
because we already stored the history record during shutdown...
Introduced `CurrentExecutions` to handle this synchronization.
(we always first store the watch record and then remove the triggered watch)
Original commit: elastic/x-pack-elasticsearch@89c4a8d8ad
This commit adds support for indexing multiple documents with the `index` action. This is done by introducing a special `_doc` field. During action execution, the `_doc` field will be looked up in the payload. If found, the value of the field will be considered as the document that needs to be indexed. If the value is an array of objects, each object in that array will be treated as a separate document and all the documents in the array will be bulk indexed.
This commit also changes the result of the action to hold `XContentSource` rather than a payload (to avoid Map creation explosions). Th `XContentSource` was also extended to support lists.
Original commit: elastic/x-pack-elasticsearch@86f454b029
Martijn van Groningen
uboness
Simon Willnauer
Simon Willnauer
Robert Muir