LindormSearch release notes - Lindorm - Alibaba Cloud Documentation Center

Alibaba Cloud releases LindormSearch versions for Lindorm to provide new features, fix known bugs, and improve user experience. This topic describes the release notes of LindormSearch versions. This topic also provides usage notes on how to upgrade the LindormSearch version of your Lindorm instance during off-peak hours.

View or upgrade the LindormSearch version of a Lindorm instance

You can check the current version of the search engine in the console. For detailed instructions and important notes about upgrading, see Upgrade the version. In the Configuration Information section of your instance, locate the Search Engine row and find the current version number in the Version column.

Warning

During a version upgrade, a rolling restart is performed on all LindormSearch nodes. During the restart process, index shards are automatically moved to available nodes. Therefore, slight jitters may occur when applications access LindormSearch. We recommend that you upgrade the version during off-peak hours.
If only one LindormSearch node is available, LindormSearch be unavailable for applications to access during the upgrade. To ensures service availability, we recommend that you add LindormSearch nodes before you upgrade LindormSearch.
The system automatically checks the search engine version of your instance. If the Upgrade Minor Version button is not displayed, your instance is already running the latest version.
Version updates may differ from region to region. The LindormSearch version displayed in the Lindorm console prevails.

Release notes of LindormSearch that is compatible with ElasticSearch

Minor version	Category	Description
3.9.31	New features and enhancements	Enumerated partitions support numeric types as enumeration values. The IK analyzer supports customization for all files. The `efficient` and `postFilter` searches support pagination.
3.9.30	New features and enhancements	Optimized the tiered storage for hot and cold data to improve performance for cold indexes. Optimized the `search_embedding` logic. You can restore a snapshot to a specified storage type.
3.9.27.3	New features and enhancements	Added support for volatile files. Added support for configuring real-time searchability at the index level. Optimized performance by using `ThreadLocal`.
3.9.24	Enhancements	Improved the performance of retrieving the `_source` field. Improved prefilter performance.
3.9.22	New features	Introduced a new routing strategy to improve the performance of OR queries.
3.9.19	New features and enhancements	Added support for custom user plug-ins. Upgrades now have less impact on query services. Optimized update operations and raw data retrieval for vector retrieval.
3.9.14	New features	The rolling update of synonyms is supported. The number of storage replicas can be dynamically modified. Slow query isolation is supported.
3.9.13	New features	The Index State Management (ISM) plug-in is supported. SQL plug-ins are supported. Data switching between versions is supported. The elasticsearch-repository-oss plug-in is supported.
3.9.10.6	Optimized features	The vector retrieval performance is optimized.
3.9.10	New features	Automatic embedding is supported in vector retrieval. Scalar filters can be pushed down in vector retrieval. The list of top queries can be retrieved and multiple queries can be killed at a time. Wildcard queries are supported.
3.9.6	New features	Weight configuration and service degradation are supported by Reciprocal Rank Fusion (RRF) in vector retrieval. The thread pool size can be dynamically modified.
3.8.13	New features	The built-in hanlp analyzer is supported. The function_score method is supported in vector retrieval.
3.8.9	New features	Table-level monitoring is supported. Multi-query RRF is supported in vector retrieval. Automatic embedding during write operations is supported in vector retrieval. Sparse vectors are supported in vector retrieval. Directed routing is supported in batch writes. Enumerated partitions are supported for time-partitioned indexing. Nested queries are supported by flat objects. The rolling update of the IK dictionary is supported.
3.8.3	New features	Multi-table queries and the DataStream feature are supported in vector retrieval. DDL operations related to index building are supported in vector retrieval. Data export is supported in vector retrieval. The time-to-live (TTL) of an index can be configured. Tiered storage of cold and hot data is supported.
3.8.2	New features	The web-based search page is added. Full-text search and vector RRF are supported in vector retrieval. Vector read and write operations can be monitored.
3.8.1	New features	The vector retrieval feature is supported. Custom routing is supported in time-partitioned indexing.
3.7.10	New features	Search indexes based on the time range partitioning policy are supported. Partitions can be automatically pruned based on time. The built-in pinyin analyzer is supported.
3.7.7	New features	The ElasticSearch 7.10 protocol is supported. The architecture in which storage and computing resources are separated is supported. The storage of LindormSearch can be shared by LindormDFS. The Zstandard compression algorithm is supported. The built-in IK analyzer is supported.

Release notes of LindormSearch that is compatible with Solr

Minor version	Category	Description
8.10.6	New features	The vector retrieval of unstructured data is supported. Dictionary files can be dynamically updated.
8.10.1	New features	Operators in more SQL statements can be pushed down. Cold storage is supported. For more information, see Configure cold storage.
7.7.13	Security update	Security patches are added to improve system security.
7.7.12	New features	Open source Apache Solr Java Database Connectivity (JDBC) can be used to access LindormSearch. For more information, see Use a JDBC driver to connect to and send SQL requests to LindormSearch.
7.7.11	New features	The real-time search feature is supported. You can search for data immediately after the data is written. The TTL feature for data can be dynamically configured. Indexes can be created for nested documents.
7.7.10	New features	When pre-sorting is used, queries can be terminated before results are returned or queries can be accelerated. The commit() method that is used to obtain the size of an index table is added.