|
| 1 | +/// CLI definition for `minikv`. |
| 2 | +/// |
| 3 | +/// # Subcommands |
| 4 | +/// - `server` run the HTTP metadata coordinator |
| 5 | +/// - `rebuild` reconstruct LevelDB from volume server autoindex |
| 6 | +/// - `rebalance` move all keys to their ideal volume set |
| 7 | +/// - `print-nginx-config` emit the nginx volume server config to stdout |
| 8 | +use std::path::PathBuf; |
| 9 | +use std::time::Duration; |
| 10 | + |
| 11 | +use clap::{Args, Parser, Subcommand}; |
| 12 | + |
| 13 | +#[derive(Parser, Debug)] |
| 14 | +#[command( |
| 15 | + name = "minikv", |
| 16 | + version, |
| 17 | + about = "Distributed object storage coordinator", |
| 18 | + long_about = None, |
| 19 | +)] |
| 20 | +pub struct Cli { |
| 21 | + #[command(subcommand)] |
| 22 | + pub command: Command, |
| 23 | +} |
| 24 | + |
| 25 | +#[derive(Subcommand, Debug)] |
| 26 | +pub enum Command { |
| 27 | + /// Run the HTTP metadata coordinator server. |
| 28 | + Server(ServerArgs), |
| 29 | + |
| 30 | + /// Rebuild LevelDB metadata from volume server autoindex listings. |
| 31 | + /// |
| 32 | + /// WARNING: This clears the entire DB before scanning. |
| 33 | + Rebuild(CommonArgs), |
| 34 | + |
| 35 | + /// Rebalance all keys to their ideal volume set. |
| 36 | + Rebalance(CommonArgs), |
| 37 | + |
| 38 | + /// Print the nginx volume server configuration to stdout. |
| 39 | + PrintNginxConfig, |
| 40 | +} |
| 41 | + |
| 42 | +/// Arguments shared by server, rebuild, and rebalance. |
| 43 | +#[derive(Args, Debug, Clone)] |
| 44 | +pub struct CommonArgs { |
| 45 | + /// Path to the LevelDB database directory. |
| 46 | + #[arg(long, env = "MINIKV_DB", required = true)] |
| 47 | + pub db: PathBuf, |
| 48 | + |
| 49 | + /// Comma-separated list of volume server addresses (host:port). |
| 50 | + #[arg(long, env = "MINIKV_VOLUMES", required = true, value_delimiter = ',')] |
| 51 | + pub volumes: Vec<String>, |
| 52 | + |
| 53 | + /// Number of replicas to maintain per object. |
| 54 | + #[arg(long, env = "MINIKV_REPLICAS", default_value = "3")] |
| 55 | + pub replicas: usize, |
| 56 | + |
| 57 | + /// Number of sub-volume shards per volume server. |
| 58 | + /// Use 1 to disable sub-volume path components. |
| 59 | + #[arg(long, env = "MINIKV_SUBVOLUMES", default_value = "10")] |
| 60 | + pub subvolumes: usize, |
| 61 | + |
| 62 | + /// Timeout for HEAD probes to volume servers (e.g. "1s", "500ms"). |
| 63 | + #[arg(long, env = "MINIKV_VOLTIMEOUT", default_value = "1s", value_parser = parse_duration)] |
| 64 | + pub voltimeout: Duration, |
| 65 | +} |
| 66 | + |
| 67 | +/// Additional arguments for the `server` subcommand. |
| 68 | +#[derive(Args, Debug)] |
| 69 | +pub struct ServerArgs { |
| 70 | + #[command(flatten)] |
| 71 | + pub common: CommonArgs, |
| 72 | + |
| 73 | + /// Port to listen on. |
| 74 | + #[arg(long, env = "MINIKV_PORT", default_value = "3000")] |
| 75 | + pub port: u16, |
| 76 | + |
| 77 | + /// Optional fallback server for keys missing from all volumes. |
| 78 | + #[arg(long, env = "MINIKV_FALLBACK")] |
| 79 | + pub fallback: Option<String>, |
| 80 | + |
| 81 | + /// Public-facing addresses for each volume server, used in Location |
| 82 | + /// redirect headers returned to clients. |
| 83 | + /// |
| 84 | + /// Must be the same length as --volumes and in the same order. |
| 85 | + /// Each entry is the externally-reachable host:port for the corresponding |
| 86 | + /// --volumes entry. |
| 87 | + /// |
| 88 | + /// Example: |
| 89 | + /// --volumes=volume1:8080,volume2:8080,volume3:8080 |
| 90 | + /// --public-volumes=localhost:8001,localhost:8002,localhost:8003 |
| 91 | + /// |
| 92 | + /// When omitted, --volumes addresses are used as-is in Location headers |
| 93 | + /// (correct for bare-metal deployments where internal == external). |
| 94 | + #[arg( |
| 95 | + long, |
| 96 | + env = "MINIKV_PUBLIC_VOLUMES", |
| 97 | + value_delimiter = ',', |
| 98 | + requires = "volumes" |
| 99 | + )] |
| 100 | + pub public_volumes: Option<Vec<String>>, |
| 101 | + |
| 102 | + /// Require UNLINK (soft-delete) before a hard DELETE is allowed. |
| 103 | + #[arg(long, env = "MINIKV_PROTECT", default_value = "false")] |
| 104 | + pub protect: bool, |
| 105 | + |
| 106 | + /// Compute and store a BLAKE3 checksum for every uploaded object. |
| 107 | + #[arg(long, env = "MINIKV_CHECKSUM", default_value = "true")] |
| 108 | + pub checksum: bool, |
| 109 | + |
| 110 | + /// Enable X-Accel-Redirect mode for GET/HEAD responses. |
| 111 | + /// |
| 112 | + /// When enabled, the coordinator returns `X-Accel-Redirect: /accel/<host>/<path>` |
| 113 | + /// instead of `302 Location`. A frontend nginx must be configured with: |
| 114 | + /// |
| 115 | + /// proxy_pass http://coordinator:3000; |
| 116 | + /// location ~ ^/accel/([^/]+)/(.+)$ { |
| 117 | + /// internal; |
| 118 | + /// proxy_pass http://$1/$2; |
| 119 | + /// } |
| 120 | + /// |
| 121 | + /// This allows nginx to stream the object body while the coordinator controls |
| 122 | + /// all response headers, including Content-Type from stored object metadata. |
| 123 | + /// |
| 124 | + /// When disabled (default), GET/HEAD returns a standard 302 redirect. |
| 125 | + #[arg(long, env = "MINIKV_ACCEL_REDIRECT", default_value = "false")] |
| 126 | + pub accel_redirect: bool, |
| 127 | + |
| 128 | + /// Enable verbose structured logging. |
| 129 | + #[arg(short, long, env = "MINIKV_VERBOSE", default_value = "false")] |
| 130 | + pub verbose: bool, |
| 131 | +} |
| 132 | + |
| 133 | +/// Parse a human-friendly duration string such as "1s" or "500ms". |
| 134 | +fn parse_duration(s: &str) -> Result<Duration, String> { |
| 135 | + if let Some(ms) = s.strip_suffix("ms") { |
| 136 | + ms.parse::<u64>() |
| 137 | + .map(Duration::from_millis) |
| 138 | + .map_err(|e| e.to_string()) |
| 139 | + } else if let Some(secs) = s.strip_suffix('s') { |
| 140 | + secs.parse::<u64>() |
| 141 | + .map(Duration::from_secs) |
| 142 | + .map_err(|e| e.to_string()) |
| 143 | + } else { |
| 144 | + // Fall back: treat as integer milliseconds. |
| 145 | + s.parse::<u64>() |
| 146 | + .map(Duration::from_millis) |
| 147 | + .map_err(|_| format!("invalid duration '{s}'. Use '1s' or '500ms'")) |
| 148 | + } |
| 149 | +} |
| 150 | + |
| 151 | +/// Validate common arguments and panic with a clear message on bad input. |
| 152 | +pub fn validate_common(args: &CommonArgs) { |
| 153 | + if args.volumes.is_empty() { |
| 154 | + eprintln!("error: --volumes must contain at least one volume server"); |
| 155 | + std::process::exit(1); |
| 156 | + } |
| 157 | + if args.replicas == 0 { |
| 158 | + eprintln!("error: --replicas must be ≥ 1"); |
| 159 | + std::process::exit(1); |
| 160 | + } |
| 161 | + if args.volumes.len() < args.replicas { |
| 162 | + eprintln!( |
| 163 | + "error: need at least as many volumes ({}) as replicas ({})", |
| 164 | + args.volumes.len(), |
| 165 | + args.replicas |
| 166 | + ); |
| 167 | + std::process::exit(1); |
| 168 | + } |
| 169 | + if args.subvolumes == 0 { |
| 170 | + eprintln!("error: --subvolumes must be ≥ 1"); |
| 171 | + std::process::exit(1); |
| 172 | + } |
| 173 | +} |
| 174 | + |
| 175 | +/// Validate server-only arguments. |
| 176 | +#[allow(unused)] |
| 177 | +pub fn validate_server(args: &ServerArgs) { |
| 178 | + validate_common(&args.common); |
| 179 | + if let Some(ref pv) = args.public_volumes |
| 180 | + && pv.len() != args.common.volumes.len() |
| 181 | + { |
| 182 | + eprintln!( |
| 183 | + "error: --public-volumes has {} entries but --volumes has {}; they must match 1:1", |
| 184 | + pv.len(), |
| 185 | + args.common.volumes.len() |
| 186 | + ); |
| 187 | + std::process::exit(1); |
| 188 | + } |
| 189 | +} |
| 190 | + |
| 191 | +#[cfg(test)] |
| 192 | +mod tests { |
| 193 | + use super::*; |
| 194 | + |
| 195 | + #[test] |
| 196 | + fn parse_duration_seconds() { |
| 197 | + assert_eq!(parse_duration("1s").unwrap(), Duration::from_secs(1)); |
| 198 | + assert_eq!(parse_duration("30s").unwrap(), Duration::from_secs(30)); |
| 199 | + } |
| 200 | + |
| 201 | + #[test] |
| 202 | + fn parse_duration_millis() { |
| 203 | + assert_eq!(parse_duration("500ms").unwrap(), Duration::from_millis(500)); |
| 204 | + } |
| 205 | + |
| 206 | + #[test] |
| 207 | + fn parse_duration_invalid() { |
| 208 | + assert!(parse_duration("1h").is_err()); |
| 209 | + } |
| 210 | +} |
0 commit comments