Moving transport from SSE to streamable-http, docker containers restarting

have been running servers in SSE transport mode. Try to switch to http streamable mode but docker containers constantly restart. mcp fetch server is the only one that I could switch.

brave-search-mcp-server does not work
thv run --transport streamable-http --name web-search --proxy-port 2101 --env BRAVE_API_KEY=yourkey --env BRAVE_MCP_TRANSPORT=stdio npx://brave/brave-search-mcp-server

filesystem does not work

thv run --transport streamable-http --name filesystem --proxy-port 2102 --volume /Users/dennisnoto/Documents/NotoData-Dev/MCP/mcp_data/common_data:/projects filesystem -- /projects

mcp-py-repl does not work
thv run --transport streamable-http --name python-interpreter --proxy-port 2103 --volume /Users/dennisnoto/Documents/NotoData-Dev/MCP/mcp_data/common_data:/mnt/data/ ghcr.io/evalstate/mcp-py-repl:latest -- run -i --rm --pull=always

My own mcp server does not work
thv run --transport streamable-http --name etana-trade-mcp --proxy-port 2106 --volume /Users/dennisnoto/Documents/NotoData-Dev/MCP/mcp_data/etana-trade-mcp/data:/app/data/ uvx://etana-trade-mcp

All work for SSE transport mode, but all restart when adding transport param. What am I missing here?

thv logs show

0m MCP server not initialized yet, skipping health check for .....
0m MCP server not initialized yet, skipping health check for filesystem
0m MCP server not initialized yet, skipping health check for python-interpreter
0m MCP server not initialized yet, skipping health check for web-search

Start by reproducing the listed thv run commands with streamable-http and compare their logs with the working SSE mode. Trace the container startup and health-check behavior around “MCP server not initialized yet”; done means the affected containers stay running and complete initialization under streamable-http.