people · May 15, 2026
Tiny Milvus Deployment Faces CrashLoopBackOff in Watsonx Orchestrate
Share the canonical public link.
IBM documented a Tiny Milvus pod crash on May 12, 2026, in Watsonx Orchestrate On-Prem deployments due to etcd backend capacity limits. The standalone Milvus instance restarts repeatedly when vector data exceeds 100 megabytes, contrasting with stable operations under lighter loads in previous setups. Administrators must increase etcd resources or switch to distributed mode. Watsonx Orchestrate version 1.2.0 integrates Tiny Milvus for local AI orchestration as of May 12, 2026.
Failed after 3 attempts. Last error: Service unavailable