Skip to content

2.12.0.0
Compare
Choose a tag to compare
@github-actions github-actions released this 20 Feb 23:57
· 703 commits to main since this release
2.12.0.0
7722020
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Learn about vigilant mode.

Version 2.12.0.0 Release Notes

Compatible with OpenSearch 2.12.0

Maintenance

  • Add Austin and Henry as maintainer (#1579)
  • Updated version of awssdk (#1607)
  • Update Gradle to 8.4 (#1697)
  • Address CVE-2023-42503 (#1727)
  • Fix CVE-2023-2976 and upgrade guava to be consistent (#2013)
  • Fix CVE-2023-42503 due to djl models (#2016)
  • Add maintainer (#1952)
  • Updating maintainers list (#1938)
  • Bump Mockito dependencies (#1868)
  • Updated OpenSearch Lucene snapshot location (#1834)
  • Add kotlin stblib dependency for SearchAlertTool (#1861)
  • Update dependency com.jayway.jsonpath:json-path to v2.9.0 (#1956)
  • Update http package import for 2.x (#1957)
  • Update the lucene snapshot url (#2082)
  • removing skills first (#2089)

Infrastructure

  • Onboarding Jenkins prod docker images to github ci check (#1566)
  • Add code coverage report for commons and memory modules (#1585)
  • Adding UT coverage for in-cache update and fine-tuning throttling feature (#1913)
  • Add tests for MLAgent Get and Delete (#1794)
  • Fix race confition in index initialization and RestUpdateConnector UT (#1857)
  • Fix failing flaky tests due to pytorch library not available for dependency plugins (#1886)
  • Adding tests for all the agent runners (#1792)
  • Update deprecated openAI mode in integration test and fix flaky tests (#1858)

Documentation

  • Updated developer guide to include links for getting started with git (#1409)
  • Add triaging doc (#1250)
  • Add bedrock blueprint doc (#1501)
  • Add bedrock titan embedding model blueprint (#1527)
  • Fix bedrock embedding model blueprint (#1563)
  • Add openai embedding model blueprint (#1602)
  • Added Connector Blueprint for AI21 Labs Jurassic-2 Mid (#1617)
  • Update default model_access_mode for model group (#1677)
  • Add cohere version 3 embedding model (#1721)
  • Blueprint for multi-model titan model (#1729)
  • AI connector blueprint for the Aleph Alpha Luminous-Base Embedding Model (#2003)
  • Add tutorial doc for semantic search on amazon opensearch (#1928)
  • Add tutorial doc for semantic search with OpenAI embedding model (#1936)
  • Add connector blueprint for Azure OpenAI Embedding and Chat model (#2062)
  • Add updated Cohere Embedding blueprint (#2063)
  • add tutorial for conversational search (#2075)
  • Add Cohere Chat blueprint with RAG (#1991)

Bug Fixes

  • Fix missing lombok version compilation failure issue (#1278)
  • Fix update connector API (#1484)
  • Fixes bugs in the Cohere Blueprint (#1505)
  • Fix register client API (#1561)
  • Fixing MachineLearningNodeClient create connector, deploy model, register model group actions (#1584)
  • Fix model/connector update API to address security concern (#1595)
  • Fixing class casting exception for MachineLearningNodeClient GetMLTask API (#1618)
  • Fix for controller error stack trace and tokenbucket (#1985)
  • Fix internal connector (#1992)
  • Fix argument pass (#1993)
  • Fix error code when failed to delete model (#2037)
  • Fix edge case for validate json method (#2045)
  • Fix long pending issue when deleting model (#2046)
  • Add escape method for process function (#2055)
  • Fix bug in delete empty memory (#1966)
  • Fix dup last trace (#1975)
  • Add a version filter to enable bwc in 2.12 (#1944)
  • Fix error message (#1976)
  • Fix string.format wrong parameter position (#1960)
  • Fix null taskId causing model undeploy issue (#1945)
  • Fix updating plugins.ml_commons.jvm_heap_memory_threshold takes no effect (#1946)
  • Fix compilation when backport (#1798)
  • Fix duplicate node if node has both data and ml role (#1830)
  • Fix partially response issue in profile API result (#1775)
  • Fix: RestStatus 500 returned ml validation (#1811)
  • Fix the hardcode password in IT (#1856)
  • Fix model not deploy issue under intensive prediction tasks (#1930)
  • Fix bug - not found agent index (#1867)
  • Issue #1787: Fixing connector endpoint returns index not found (#1885)
  • Issue #1878/#1879/#1880: Fixing index not found for model group/model/tasks (#1895)
  • Change searchResponse method to fix breaking change in managed service (#1917)
  • Adjust ListTool response format (#1912)
  • Issue #844: Add accessUserInformation to the plugin security policy (#1959)
  • Escape input data (#1974)
  • flow agent suggestions missing (#2064)
  • bug fix - tool parameters missing (#2065)
  • fix empty tool parameter issue (#2067)
  • tool uses original input (#2068)
  • update Unthrotized error code to 401 (#2076)
  • handle null value exceptions when arguments are missing or Null in caling RAG pipeline (#2079)
  • changing error message and error code (#2073) (#2086)
  • Fix bwc issue in remote prediction (#2085)

Experimental Feature

  • Update Model API (#1350)
  • Hidden model implementation (#1755)
  • Model & user level throttling (#1814)
  • Search agent api (#1826)
  • Add GetTool API and ListTools API (#1850)
  • Enable in-place update model (#1796)
  • Memory Manager and Update Memory Actions/APIs (#1776)
  • Add CatIndexTool (#1770)
  • Add search and singular APIs to conversation memory (#1720)
  • Memory interface in spi (#1771)
  • Tool interface (#1772)
  • Add get config api to retrieve root agent id (#1995)
  • Register agent rest and transport actions (#1801)
  • IndicesHandler and conversationIndexMemory (#1777)
  • Adding mlmodeltool and agent tool with tests (#1778)
  • Get and delete agent APIs (#1779)
  • Add register action request/response (#1780)
  • Add execute agent api; add load extension (#1810)
  • Add IndexMapping Tool (#1934)
  • Agent meta classes in common (#1759)
  • Agent framework disable/enable flag (#1994)
  • Do not allow non super admin users to undeploy hidden models (#1981)
  • System error handling (#2051)
  • Handling tool errors (#1881)
  • Changes to hidden model code to use OPENDISTRO_SECURITY_USER instad of ssl principal (#1897) (#1900)
  • Support regenerate for chatbot (#1823)
  • Update IndexMappingTool Description (#1998)
  • Add more user based permission check in Memory (#1935)
  • Update memory index name and add updated_time (#1793)
  • fine tune prompt;refactor conversational agent code (#2094)

Feature

  • Add cross encoder support (#1739)
  • Enable conversation memory feature flags (#2095)

Enhancements

  • Add register and deploy api in client (#1502)
  • Added create connector API for MLClient (#1506)
  • Added register model group API for MLClient (#1519)
  • Expose execute api for MLClient (#1541)
  • Add prefix to show the error is from remote service (#1515)
  • Fine tune predict API: read model from index directly (#1559)
  • Removed empty host check, inherently within httpHost object generation (#1599)
  • Cluster restart model auto redeploy (#1627)
  • Add new data fields in the memory layer and update tests (#1753)
  • Check if model id is null when undeploy (#2015)
  • Adds inputs validation for create memory (#2040)
  • Make response_field customizable in MLModelTool (#2043)
  • Add memory id and interation id for non-verbose (#2005)
  • Issue #1965: Remove logging of sensitive chat history (#2012)
  • Update memory if tool output needs to be included in response (#2018)
  • Disable dynamic mapping for config index (#2027)
  • Move allow model setting from rest to transport (#1977)
  • Fine tune connector process function (#1963)
  • Change model auto redeploy enabled to true (#1809)
  • Add auto expand replica settings to memories (#1824)
  • Add interaction id into execute response (#1825)
  • Add more methods to client (#1782)
  • Support charset input params and change default charset as utf8 (#1828)
  • Fine tune log message based on error type (#1842)
  • Stash thread context before running forward action (#1906)
  • Refine the error message on cluster status not ready scenario (#1931)
  • Add Request-Source header (#1892)
  • Add process function for bedrock (#1554)
  • send agent execution response after saving memory (#2066)
  • add conversational flow agent (#2069)

REFACTOR

  • Rename memory field names in responses (#2020)
  • Refactor memory layer APIs (#1890)
Assets 2
Loading