377 issues opened 2 personal-project PRs 67 other threads joined
Dates show when each thread opened. “Closed” is a thread status, not a claim that a request shipped. Checked Sep 20, 2026.
Back to search- ADK Web hangs and logs “ASGI callable returned without completing response” when MCP Tool (Streamable HTTP) endpoint is unreachable
- ImportError: cannot import name validate_core_schema from pydantic_core with pydantic==2.11.7 and pydantic_core==2.39.0 (Python 3.13, Windows)
- Google search tool is not supported for Gemini 2.0 and 2.5 models
- Add support for additional Memory Bank services: DatabaseMemoryService/RedisMemoryService
- SDK for the REST API
- Update API Server Docs for all available endpoints
- MCP Streamable HTTP Support
- Trino schema introspection fail causes unexpected error
- AI-powered SQL assistance to the OSS version
- Upgrade DuckDB JDBC driver to support the DuckLake extension
- Error when expanding to Show full text
- View Resource Manager for Anonymous/Not logged in Users
- What is configuration for setting defaults for "Group by Project" and "Filter" in Settings?
- bug upgradgin to v23
- Trino driver issue with CB 22.3.1
- Some images Links do not respect custom rootURI property properly
- Connection hamburger icon hidden on resize
- Hide the GlobalConfiguration and Other folders
- Reconnection options for Trino connection closed due to manual cancels
- Error when saving script - Variable 'projectId' has coerced Null value for NonNull type 'String!'
- Login link still present after allowing anonymous log in (How to remove it?)
- Trino - cancelling running query gives connection closed
- Session Expired - Grammar/Typo
- Connector for StarRocks
- Support Iceberg v3 spec
- [FEATURE] Support LDAP auth
- 0.9.1 and 10.0.0 images
- [FEATURE] Next release timeline
- [Bug report] Iceberg REST, HMS specific configs
- [FEATURE] Update Iceberg version to latest
- [FEATURE] Way to get the Docker image of latest dev builds
- [info-request] change the default example oauth client and secret
- Iceberg Rest Unable to list tables for Hive Metastore with mixed Hive and Iceberg
- Example config iceberg with kerberized HMS backend
- [Bug report] Purge meta use LIMIT in the delete statement for POSTGRESQL
- [Subtask] upgrade Iceberg version to 1.7 in Iceberg REST server version
- [Agent Builder] Make no-activity timeout configurable and improve timeout UX
- [Agent Builder] Opt-in Agent Builder metadata passthrough to LLM connectors
- [Agent Builder] Enriched Context for Agent Builder Pre-Execution Workflows
- Agent Builder MCP tools surface generic action error instead of detailed connector service message
- MCP tool responses are serialized as a single very long JSON line, causing line too long failures in some MCP clients
- Discover in Kibana 8.7.0 shows All mapping/template fields incl. Empty fields
- [UnifiedFieldList] Fields are not marked as Empty in Discover sidebar
- LiteLLM Stability Sprint Roadmap
- fix(spend-logs): rehydrate metadata JSONB text on ui_view_spend_logs
- Spend Logs UI regression: upstream provider failures are hidden because /spend/logs/ui returns metadata as JSON text
- Admin UI router settings crashes on dict-based fallback entries that runtime supports
- [BUG]: llm_requests_hanging Slack alert fired at 600s threshold but no >=600s request in spend logs
- [Feature Request]: Support for NVIDIA NeMo Guardrails (Python Library & Proxy API Server)
- Bug: `get_generic_data()` no longer retries/reconnects on Prisma transport errors in `1.83.1` / `1.83.2`
- Focus: Stabilize core features and fix accumulated bugs/UX issues
- Fix Usage AI chat for proxy model aliases and model groups
- Bug: OpenTelemetry callback crashes on Usage AI chat due to non-dict `response_obj`
- ## Bug: Usage AI Chat fails when selected LiteLLM model name is a proxy alias / model group
- [Bug]: OTEL invalid attribute types for `gen_ai.prompt`
- [Bug] UI very slow in version 1.82.x
- [Bug]: Request Logs / Spend logs no longer showing in the UI
- [Feature]: Onboard Agents via Agent.json
- [Feature]: Support custom and dynamic headers for A2A Agents
- [Performance] LiteLLM Overhead and Performance in Production
- [Bug]: Significant performance regression after upgrading from 1.80.5 to 1.81.x (UI + API slowness)
- [Bug]: Custom SSO Handler does not get called anymore
- [Bug]: Can't connect to my Elastic Agent via a2a
- [Feature]: Allow dynamic prompts for Arize Phoenix through the the gateway/proxy
- [Bug]: LiteLLM Content Filter just works on the first content item
- [Bug]: Regression:- log_pre_api_call stopped getting called for passthrough endpoints
- [Bug]: async_log_pre_api_call not working
- [Bug]: router settings for redis not allowed
- [Bug]: Team level "Allowed Pass Through Routes" not working
- [Feature]: Prompt Support Integration Arize Phoenix
- [Bug]: passthrough endpoints not respecting SERVER_ROOT_PATH
- [Feature]: Support github for prompt store
- [Bug]: Semantic Redis Cache "No module named 'redisvl'
- [Bug]: Caching does not work with a custom api_base
- [Feature]: Allow uploading agent.json files for the new A2A Agent in AI HUB
- [Feature]: Add Elasticsearch as vector store
- [Bug]: Setting RPM / TPM per member in Team no effect
- [Bug]: SSO free for 5 users but getting licence error
- [Bug]: Galileo integration gives errors
- [Feature]: Support GALILEO_API_KEY for auth
- [Feature]: Allow Fallbacks in Router to use keys from litellm_params
- [Bug]: Litellm FastAPI doc "Try it out" does not respect SERVER_ROOT_PATH
- [Bug]: passthrough endpoints not added to config.yml
- [Bug]: multiple otel collectors (otel and arize_phoenix) does not mork
- [Bug]: Litellm proxy Custom auth using UserAPIKeyAuth does not honor the allowed models
- [Bug]:
- [Bug]: Custom pass-through endpoint can't work
- [Bug]: only the first callback works from the callback list
- [Feature]: support custom auth for model providers
- [Bug]: The model level guardrail only works on the first guardrail in the list
- [Bug]: Toolcall results with role as "tool" are not guardrailed
- fix: share a single _OPTIONAL_PresidioPIIMasking instance to ensure unmasking works when output_parse_pii=true
- [Bug]: Setting "output_parse_pii" has no effect
- [Bug]: Guardrail config on completion request fails
- [Bug]: MCP servers names for a team show as null in the UI
- [Feature]: Support AI Studio Imagen4 models
- [Bug]: Unable to use imagen4 preview (Ai Studio Version)
- [Bug]: Gemini tool call "default" errors
- [Bug]: Models UI: Potential invalid entry
- [Bug]: Logging and Alert UI : No Delete option
- [Bug]: Budget UI - Should show Budget Created
- [Bug]: Add Model freeform text field from UI errors out
- [Bug]: Cant Add Ollama Model in the UI (DB Model)
- Context Center search does not include Documents
- LDAP configuration changes from Kubernetes Secrets ignored in 2.0.1 until persisted auth config is removed
- Labeling BigQuery Jobs for Cost Attribution
- Openlineage REST Endpoint Support:
- fix issue 14802: disable self-sigup interface when using ldap authent…
- SELF_SIGNUP setting doesn't work for LDAP authentication provider
- Ingestion: Allow users to input SSL config instead of providing paths
- Ldap feature: Allow user login using username and reassign roles based on users' ldap groups when login
- bedrock claude models seem to raise "InvalidLLMOutputType("Response validation failed!")"
- Docker compose platform errors at startup in the browser
- Metrics and Metadata for each request
- docker-compose up fails
- Support Open WebUI API
- Field descriptions not seemed by used by agent and connector
- The generated code has a shifted alignment on the top line which leads to NameError on exec
- Support of Hive Metastore (HMS) for Iceberg catalogs
- Documentation for MCP Use
- Support Ducklake and Native iceberg
- Native Iceberg Support
- Onprem K8s: Sling Platform UI
- trino limit offset errors
- Support for Trino
- Support for Elasticsearch
- Starrocks: random timeouts when using stream load feature
- Starrocks: How do we set the fe_url in ENV variable or Python wrapper?
- Add Built-in Support for TPC-H/TPC-DS Data Generation via External Catalog
- Add support for Apache Iceberg V3 Table Specification
- Add support for the PostgreSQL Wire Protocol
- Introduce Native AI/ML Functions for In-Database Analytics
- Support ducklake with starrocks
- LDAPS and Group support
- OpenLineage Support
- Django support
- Starrocks MySQL protocol version
- Support FE metadata on S3
- Trino - Starrocks connector - revisit
- Issue with Python Django Support
- In trino's sql_dialect mode, date_add returns different precision as trino.
- Allow multiple hive-site.xml for kerberized and non kerberized catalogs
- Support LDAPS authentication
- Ability to hide/get creds from external sources for FILES() queries.
- Show external catalog with their schemas in IDEs like Dbeaver, Datagrip etc
- Issue with LDAPS on port 636 for Authentication
- Issue connecting to a Kerberized hive metastore
- The JAVA_OPTS conf does not seem to be taking effect in the CN nodes
- Support LDAP Group Memberships for AuthZ and RBAC
- support connect differtent hive cluster by create different catalog
- Add an official MCP (Model Context Protocol) Server
- Error with build
- [OpenTelemetry] Add option for Headers / Auth
- OpenMetrics Endpoint showing duplicate metrics
- Trino 473 - Glue V1 Hive Metastore Config Errors
- S3 Tables Iceberg warehouse
- Ai functions LLM additional parameters
- Bigquery : Support access of other project ids / dataset ids from a catalog(project id)
- value for "iceberg.rest-catalog.oauth2.credential"
- Add read support for S3 Tables in Iceberg
- Trino attempts to list views from an iceberg REST catalog that does not support views
- Use Ubi 9 micro image for docker
- Feature: Aliasing catalogs
- fs.cache seems to be enabled by default
- Allow exporting java.lang:type=Memory metrics as separate fields via openmetrics
- Add more options to S3FileSystemConfig
- Switch runtime Java version to Temurin 21.0.1
- Adding proxy support for AWS Glue metastore
- Elasticsearch "ignore_throttled" is deprecated
- Adding an http proxy option for the bigquery connector
- Add StarRocks connector
- Kafka confluent schema registry error on startup on v415
- Malformed ORC file. Cannot read SQL type 'bigint' error
- Handling ranges for Kafka timestamp queries millisecond precision
- Error reading ORC file with short zone id in the footer
- SingleStore Connector: Inserts Not Working
- iceberg : using table properties in DDL
- Allow multiple hosts for elasticsearch connector (elasticsearch.host property)
- Adding an http proxy option for the kinesis connector
No threads match. Try another project or topic, or clear the search.