|
| 1 | +use std::path::PathBuf; |
| 2 | + |
| 3 | +use anyhow::Context; |
| 4 | +use aws_config::{retry::RetryConfig, BehaviorVersion, Region}; |
| 5 | +use aws_credential_types::Credentials; |
| 6 | +use aws_sdk_s3::config::SharedCredentialsProvider; |
| 7 | +use clap::{Parser, ValueEnum}; |
| 8 | +use libsql_wal::io::StdIO; |
| 9 | +use libsql_wal::storage::backend::s3::S3Backend; |
| 10 | +use libsql_wal::storage::compaction::strategy::{ |
| 11 | + identity::IdentityStrategy, log_strategy::LogReductionStrategy, PartitionStrategy, |
| 12 | +}; |
| 13 | +use libsql_wal::storage::compaction::Compactor; |
| 14 | + |
| 15 | +#[derive(Debug, clap::Args)] |
| 16 | +struct S3Args { |
| 17 | + #[arg(long, requires = "S3Args")] |
| 18 | + enable_s3: bool, |
| 19 | + #[arg(long, env = "LIBSQL_BOTTOMLESS_DATABASE_ID")] |
| 20 | + cluster_id: Option<String>, |
| 21 | + #[arg(long, env = "LIBSQL_BOTTOMLESS_ENDPOINT")] |
| 22 | + s3_url: Option<String>, |
| 23 | + #[arg(long, env = "LIBSQL_BOTTOMLESS_AWS_SECRET_ACCESS_KEY")] |
| 24 | + s3_access_key: Option<String>, |
| 25 | + #[arg(long, env = "LIBSQL_BOTTOMLESS_AWS_ACCESS_KEY_ID")] |
| 26 | + s3_access_key_id: Option<String>, |
| 27 | + #[arg(long, env = "LIBSQL_BOTTOMLESS_BUCKET")] |
| 28 | + s3_bucket: Option<String>, |
| 29 | + #[arg(long, env = "LIBSQL_BOTTOMLESS_AWS_DEFAULT_REGION")] |
| 30 | + s3_region_id: Option<String>, |
| 31 | +} |
| 32 | + |
| 33 | +#[derive(Clone, Debug, ValueEnum)] |
| 34 | +enum CompactStrategy { |
| 35 | + Logarithmic, |
| 36 | + CompactAll, |
| 37 | +} |
| 38 | + |
| 39 | +#[derive(Debug, clap::Parser)] |
| 40 | +struct Command { |
| 41 | + #[arg(long, short, default_value = "compactor")] |
| 42 | + path: PathBuf, |
| 43 | + #[command(flatten)] |
| 44 | + s3_args: S3Args, |
| 45 | + #[command(subcommand)] |
| 46 | + subcommand: Subcommand, |
| 47 | +} |
| 48 | + |
| 49 | +#[derive(Debug, clap::Subcommand)] |
| 50 | +enum Subcommand { |
| 51 | + /// Register namespaces to monitor |
| 52 | + Monitor { namespace: String }, |
| 53 | + /// Analyze segments for a namespaces |
| 54 | + Analyze { |
| 55 | + /// list all segments |
| 56 | + #[clap(long)] |
| 57 | + list_all: bool, |
| 58 | + namespace: String, |
| 59 | + }, |
| 60 | + /// Compact segments into bigger segments |
| 61 | + Compact { |
| 62 | + // compaction strategy |
| 63 | + #[clap(long, short)] |
| 64 | + strategy: CompactStrategy, |
| 65 | + /// prints the compaction plan, but doesn't perform it. |
| 66 | + #[clap(long)] |
| 67 | + dry_run: bool, |
| 68 | + namespace: String, |
| 69 | + }, |
| 70 | + /// Sync namespace metadata from remote storage |
| 71 | + Sync { |
| 72 | + /// When performing a full sync, all the segment space is scanned again. By default, only |
| 73 | + /// segments with frame_no greated that the last frame_no are retrieved. |
| 74 | + #[clap(long)] |
| 75 | + full: bool, |
| 76 | + /// unless this is specified, all monitored namespaces are synced |
| 77 | + namespace: Option<String>, |
| 78 | + }, |
| 79 | + /// Restore namespace |
| 80 | + Restore { |
| 81 | + #[clap(long)] |
| 82 | + verify: bool, |
| 83 | + namespace: String, |
| 84 | + out: PathBuf, |
| 85 | + }, |
| 86 | +} |
| 87 | + |
| 88 | +#[tokio::main] |
| 89 | +async fn main() -> anyhow::Result<()> { |
| 90 | + let cmd: Command = Command::parse(); |
| 91 | + |
| 92 | + let backend = setup_storage(&cmd.s3_args).await?; |
| 93 | + tokio::fs::create_dir_all(&cmd.path).await?; |
| 94 | + let mut compactor = Compactor::new(backend.into(), &cmd.path)?; |
| 95 | + match cmd.subcommand { |
| 96 | + Subcommand::Monitor { namespace } => { |
| 97 | + let namespace = libsql_sys::name::NamespaceName::from_string(namespace); |
| 98 | + compactor.monitor(&namespace).await?; |
| 99 | + println!("monitoring {namespace}"); |
| 100 | + } |
| 101 | + Subcommand::Analyze { |
| 102 | + namespace, |
| 103 | + list_all, |
| 104 | + } => { |
| 105 | + let namespace = libsql_sys::name::NamespaceName::from_string(namespace); |
| 106 | + let analysis = compactor.analyze(&namespace)?; |
| 107 | + println!("stats for {namespace}:"); |
| 108 | + println!("- segment count: {}", analysis.segment_count()); |
| 109 | + println!("- last frame_no: {}", analysis.last_frame_no()); |
| 110 | + let set = analysis.shortest_restore_path(); |
| 111 | + println!("- shortest restore path len: {}", set.len()); |
| 112 | + if let Some((first, last)) = compactor.get_segment_range(&namespace)? { |
| 113 | + println!( |
| 114 | + "- oldest segment: {}-{} ({})", |
| 115 | + first.key.start_frame_no, first.key.end_frame_no, first.created_at |
| 116 | + ); |
| 117 | + println!( |
| 118 | + "- most recent segment: {}-{} ({})", |
| 119 | + last.key.start_frame_no, last.key.end_frame_no, last.created_at |
| 120 | + ); |
| 121 | + } |
| 122 | + |
| 123 | + if list_all { |
| 124 | + println!("segments:"); |
| 125 | + compactor.list_all(&namespace, |info| { |
| 126 | + println!( |
| 127 | + "- {}-{} ({})", |
| 128 | + info.key.start_frame_no, info.key.end_frame_no, info.created_at |
| 129 | + ); |
| 130 | + })?; |
| 131 | + } |
| 132 | + } |
| 133 | + Subcommand::Compact { |
| 134 | + strategy, |
| 135 | + dry_run, |
| 136 | + namespace, |
| 137 | + } => { |
| 138 | + let namespace = libsql_sys::name::NamespaceName::from_string(namespace); |
| 139 | + let analysis = compactor.analyze(&namespace)?; |
| 140 | + let strat: Box<dyn PartitionStrategy> = match strategy { |
| 141 | + CompactStrategy::Logarithmic => Box::new(LogReductionStrategy), |
| 142 | + CompactStrategy::CompactAll => Box::new(IdentityStrategy), |
| 143 | + }; |
| 144 | + let set = analysis.shortest_restore_path(); |
| 145 | + let partition = strat.partition(&set); |
| 146 | + |
| 147 | + println!("initial shortest restore path len: {}", set.len()); |
| 148 | + println!("compacting into {} segments", partition.len()); |
| 149 | + for set in partition.iter() { |
| 150 | + println!("- {:?}", set.range().unwrap()); |
| 151 | + } |
| 152 | + if dry_run { |
| 153 | + println!("dry run: stopping"); |
| 154 | + } else { |
| 155 | + println!("performing compaction"); |
| 156 | + let part_len = partition.len(); |
| 157 | + for (idx, set) in partition.into_iter().enumerate() { |
| 158 | + let Some((start, end)) = set.range() else { |
| 159 | + continue; |
| 160 | + }; |
| 161 | + println!("compacting {start}-{end} ({}/{})", idx + 1, part_len); |
| 162 | + // TODO: we can compact in conccurently |
| 163 | + compactor.compact(set).await?; |
| 164 | + } |
| 165 | + } |
| 166 | + } |
| 167 | + Subcommand::Sync { full, namespace } => match namespace { |
| 168 | + Some(_ns) => { |
| 169 | + todo!() |
| 170 | + } |
| 171 | + None if full => { |
| 172 | + compactor.sync_full().await?; |
| 173 | + println!("all monitored namespace fully up to date."); |
| 174 | + } |
| 175 | + _ => todo!(), |
| 176 | + }, |
| 177 | + Subcommand::Restore { |
| 178 | + namespace, |
| 179 | + out, |
| 180 | + verify, |
| 181 | + } => { |
| 182 | + let namespace = libsql_sys::name::NamespaceName::from_string(namespace); |
| 183 | + let analysis = compactor.analyze(&namespace)?; |
| 184 | + let set = analysis.shortest_restore_path(); |
| 185 | + compactor.restore(set, &out).await?; |
| 186 | + if verify { |
| 187 | + let conn = libsql_sys::rusqlite::Connection::open(&out)?; |
| 188 | + conn.pragma_query(None, "integrity_check", |r| { |
| 189 | + println!("{r:?}"); |
| 190 | + Ok(()) |
| 191 | + })?; |
| 192 | + } |
| 193 | + } |
| 194 | + } |
| 195 | + |
| 196 | + Ok(()) |
| 197 | +} |
| 198 | + |
| 199 | +async fn setup_storage(opt: &S3Args) -> anyhow::Result<S3Backend<StdIO>> { |
| 200 | + let config = aws_config::load_defaults(BehaviorVersion::latest()).await; |
| 201 | + |
| 202 | + let mut builder = config.into_builder(); |
| 203 | + builder.set_endpoint_url(opt.s3_url.clone()); |
| 204 | + builder.set_retry_config(RetryConfig::standard().with_max_attempts(10).into()); |
| 205 | + builder.set_region(Region::new( |
| 206 | + opt.s3_region_id.clone().expect("expected aws region"), |
| 207 | + )); |
| 208 | + let cred = Credentials::new( |
| 209 | + opt.s3_access_key_id.as_ref().unwrap(), |
| 210 | + opt.s3_access_key.as_ref().unwrap(), |
| 211 | + None, |
| 212 | + None, |
| 213 | + "Static", |
| 214 | + ); |
| 215 | + builder.set_credentials_provider(Some(SharedCredentialsProvider::new(cred))); |
| 216 | + let config = builder.build(); |
| 217 | + let backend = S3Backend::from_sdk_config( |
| 218 | + config, |
| 219 | + opt.s3_bucket.clone().context("missing bucket id")?, |
| 220 | + opt.cluster_id.clone().context("missing cluster id")?, |
| 221 | + ) |
| 222 | + .await?; |
| 223 | + |
| 224 | + Ok(backend) |
| 225 | +} |
0 commit comments