Release notes

7.13.4-25.3

Fixed a join cache issue with Elastic Stack Security’s templated queries when they include placeholders such as {{ _user: user.name }}.
Fixed an issue in the siren_dataspaces query that could prevent users from seeing documents in Siren Investigate when hitting the join cache.
Removed the execution of a SQL query when validating datasources; the connection is now validated by checking that the Avatica TLS Server instance is reachable over the network.
Removed the autoCommit setting from Avatica connections. If required, autoCommit can be forced to have specific values in the Avatica TLS Server configuration.
Fixed a rare issue where a cluster was unable to initialize a Federate job due to a closed RootBufferManager on a node.

The minimal JDK required for Federate plugin is set to JDK 11. The recommendation is to use the bundled JDK of Elasticsearch platform.

Upgrades to Elasticsearch 7.13.3
Improves the capabilities of a join on a runtime field. A join query involving one or more runtime fields now supports using the HASH_JOIN or BROADCAST_JOIN algorithms. The INDEX_JOIN algorithm only supports a join with a runtime field on the right side.
Fixes broken links in PDF documentation.
Improves the selection of shards for the creation of point in time to maximize the use of query cache.
Removes the Volcano planner for selecting the optimal join implementation.

Deprecates property siren.io.netty.maxDirectMemory in favor of siren.memory.root.limit setting. The use of the deprecated setting will now have no effect and will log a warning message (removal planned in next release).
Removes the compatibility with a boolean join.
Paginating a search request with the scroll API is not supported anymore after the deprecation of the scroll API in Elasticsearch. Instead, the pagination must be done using the search_after feature.
If using SearchGuard as the security system, the node-level setting searchguard.allow_custom_headers: "siren.*" must be added to each node of the cluster.
The minimal JDK required for Federate plugin is set to JDK 11. The recommendation is to use the bundled JDK of Elasticsearch platform.

Fixes issue where no shard was selected for the opening of a point in time, which could have happened if shards were not started.
Job cancel transport action was blocking the management thread pool leading to a deadlock situation. The cancel action now extends the Elasticsearch task cancel API to avoid this situation.
Restore the permissions required by json-ws processor in plugin security policy.
Fixes an issue with Proguard and JDK11.
Enable the use of Elasticsearch preference parameter to limit some join queries to a subset of shards.