Review: JuiceFS block storage support

Summary: Adds JuiceFS as a new block storage backend by reusing the HDFS service implementation. Detects JuiceFS providers via jfs:// prefix pattern.

What's good:

• Clever approach — leverages existing HDFS service since JuiceFS provides HDFS-compatible interface
• Provider detection logic (is_juicefs_provider) validates format correctly (prefix, @, non-empty host)
• Clean separation of concerns for args handling between JuiceFS and standard providers

Issues / Suggestions:

1. Hardcoded jar version in pack_server.sh — juicefs-hadoop-1.3.1.jar is downloaded with wget -N at build time. This will fail in air-gapped build environments. Consider making the version configurable or bundling it as a dependency.

2. jfs:// prefix is a magic string — Extract to a constant (e.g., JUICEFS_PROVIDER_PREFIX) alongside BLOCK_SERVICE_JUICEFS.

3. No unit tests for is_juicefs_provider — The validation logic is simple but worth testing edge cases like jfs:// (no user), jfs://user@ (empty host), jfs://@host (empty user), dfs://user@host (wrong prefix).

4. Parser throws away parsed components — The function validates user@host format but then reconstructs args from the raw provider string. The parsed user could be validated further (e.g., non-empty after prefix).

5. Test script uses example placeholder — The package_dir="example: ..." in run.sh looks like documentation rather than a working config. Should this be a parameter or env var default?

Verdict: ⚠️ Request Changes — The hardcoded jar download and lack of unit tests for the detection logic should be addressed before merge.