ml-commons
ml-commons copied to clipboard
opensearch-project
[Discussion] Remove unnecessary setting in 2.14
What is the bug? Lots of settings are added for existing features. We should consider removing some of them after they are GA for a while.
The settings that are agreed to be removed:
ML_COMMONS_MEMORY_FEATURE_ENABLED - plugins.ml_commons.memory_feature_enabled
ML_COMMONS_RAG_PIPELINE_FEATURE_ENABLED - plugins.ml_commons.rag_pipeline_feature_enabled
Under discussion:
ML_COMMONS_MODEL_AUTO_REDEPLOY_ENABLE - plugins.ml_commons.model_auto_redeploy.enable
ML_COMMONS_REMOTE_INFERENCE_ENABLED - plugins.ml_commons.remote_inference.enabled
ML_COMMONS_AGENT_FRAMEWORK_ENABLED - plugins.ml_commons.agent_framework_enabled
ML_COMMONS_MODEL_AUTO_REDEPLOY_LIFETIME_RETRY_TIMES - plugins.ml_commons.model_auto_redeploy.lifetime_retry_times
What is the expected behavior? A clear and concise description of what you expected to happen.
What is your host/environment?
- OS: [e.g. iOS]
- Version [e.g. 22]
- Plugins
Do you have any screenshots? If applicable, add screenshots to help explain your problem.
Do you have any additional context? Add any other context about the problem.
Why remove these settings ? If user don't want to use these features, how they disable it?
The conversational search can be considered as a standard search feature like KNN so no feature flag needed? Like we do search in OS, we don't need to open a flag to do that. I think Aryn has agreed for remove these two settings for memories.
For other settings, let's discuss them in the next community meeting. It's not a Bug but Github only provided that sample by default.
I would suggest keep the settings to keep the flexibility for advanced user who may want to customize.
@Zhangxunmt can you open a separate issue to discuss the default for ML_COMMONS_ONLY_RUN_ON_ML_NODE?
@austintlee, Yes I just removed ML_COMMONS_ONLY_RUN_ON_ML_NODE out from this discussion.