| 1 | //! The runner in the app: g1t's self-hosted runner, started for the |
| 2 | //! signed-in person when g1t opens and stopped when it quits, so their |
| 3 | //! computer runs their workspace's `runs-on: self-hosted` jobs with |
| 4 | //! nothing to install, register or start by hand |
| 5 | //! (docs.g1t.sh/guides/desktop/#the-runner). |
| 6 | //! |
| 7 | //! - **When.** On by default; settings turn it off ("Run jobs on this |
| 8 | //! computer while g1t is open"), and Stop in settings or the tray stops |
| 9 | //! it until Start, or until g1t starts again. It starts a few seconds |
| 10 | //! after the window, never before. |
| 11 | //! - **How.** The backend settings choose (backend.rs): Sandboxed, the |
| 12 | //! default, where nothing a job does reaches this computer; Docker; or |
| 13 | //! directly on this computer. The two that are not sandboxes are only |
| 14 | //! ever chosen by the person, and nothing falls back to them. Sandboxed |
| 15 | //! is `g1t-runner sandbox` (launcher.rs) on every OS: `up` a few seconds |
| 16 | //! after the window, `status` every few seconds while it runs, `down` |
| 17 | //! as g1t quits, and `down --drain` when another backend is chosen. |
| 18 | //! Each kind of job it runs (Linux jobs, Windows jobs) starts and stops |
| 19 | //! on its own too: settings keep which are on (`runnerCapabilities`), |
| 20 | //! every `up` names them (`--only`), and stopping one is `down --only` |
| 21 | //! it, which leaves the other running. One that should run and doesn't |
| 22 | //! while the other runs is tried again by itself, every few minutes, and |
| 23 | //! every five while the newest release has no sandbox images. |
| 24 | //! - **Registering.** The first time, and when the workspace or backend |
| 25 | //! changes, the app needs a registration token for the workspace. Only |
| 26 | //! the site can make one, for an owner who is signed in: the app asks |
| 27 | //! the site's page (`g1t-runner-wanted`), the page asks the server |
| 28 | //! (`POST /-/desktop/runner-token`) and hands the token to the app |
| 29 | //! (`runner_register` in bridge.rs), which gives it to `g1t-runner` |
| 30 | //! once: in its environment, or to the sandbox's first `up`. It is never |
| 31 | //! shown or written to the log. The registration is kept |
| 32 | //! (registration.json and the runner's own folder; the sandbox keeps its |
| 33 | //! own, and sandbox-registration.json says whose it is) and used on |
| 34 | //! every later start. |
| 35 | //! - **Up to date.** Soon after it starts and every hour, and when the |
| 36 | //! person asks ("Check for runner updates"), the app looks for a newer |
| 37 | //! release on a thread of its own: its manifest only when g1t's release |
| 38 | //! key signed it (native.rs), its program only when its SHA-256 is the |
| 39 | //! manifest's. What it finds waits, downloaded, until each kind of job |
| 40 | //! is idle, and goes in kind by kind, never in the middle of a job: the |
| 41 | //! sandbox's Linux jobs take its image (`sandbox update`, which drains |
| 42 | //! them itself) once they have no job; its Windows jobs, whose runner is |
| 43 | //! the program itself, are drained (`down --only windows --drain`), the |
| 44 | //! program is replaced, and they start again; a runner in Docker or on |
| 45 | //! this computer is stopped between jobs and started on the new one. |
| 46 | //! "Download updates by themselves" off, it looks only when asked. |
| 47 | //! - **Seen.** What it is doing, and when it is not running why and how to |
| 48 | //! fix it (diagnosis.rs), is in the tray, in settings, and on the site's |
| 49 | //! Runners page (`g1t-runner-status`), read from what its log says |
| 50 | //! (log.rs), or for the sandbox from its status. The whole log is kept |
| 51 | //! in `runner/runner.log`. |
| 52 | //! |
| 53 | //! All of it runs on a thread of its own, which owns the runner's process |
| 54 | //! and is told what happened through one channel. |
| 55 | |
| 56 | mod backend; |
| 57 | pub mod diagnosis; |
| 58 | mod launcher; |
| 59 | mod log; |
| 60 | mod native; |
| 61 | |
| 62 | use std::collections::VecDeque; |
| 63 | use std::fs::{File, OpenOptions}; |
| 64 | use std::io::{BufRead, BufReader, Read, Write}; |
| 65 | use std::path::PathBuf; |
| 66 | use std::process::{Child, Stdio}; |
| 67 | use std::sync::Mutex; |
| 68 | use std::sync::mpsc::{self, Receiver, RecvTimeoutError, Sender}; |
| 69 | use std::time::{Duration, Instant, SystemTime}; |
| 70 | |
| 71 | use serde::{Deserialize, Serialize}; |
| 72 | use tauri::{AppHandle, Emitter, EventTarget, Manager}; |
| 73 | use url::Url; |
| 74 | |
| 75 | pub use backend::Backend; |
| 76 | use backend::{Engine, Missing}; |
| 77 | use diagnosis::{Code, Diagnosis, Facts}; |
| 78 | use launcher::{Down, State, Status}; |
| 79 | use native::Fault; |
| 80 | |
| 81 | use crate::{config, frame, settings, tray}; |
| 82 | |
| 83 | /// What settings, the tray and the site listen for: the runner's status changed. |
| 84 | pub const STATUS_EVENT: &str = "g1t-runner-status"; |
| 85 | /// What the site's page listens for: the app needs a registration token for a workspace. |
| 86 | pub const WANTED_EVENT: &str = "g1t-runner-wanted"; |
| 87 | |
| 88 | /// How long after the window the runner starts. |
| 89 | const AFTER_START: Duration = Duration::from_secs(4); |
| 90 | /// How often the thread looks at the runner when nothing has happened. |
| 91 | const TICK: Duration = Duration::from_secs(5); |
| 92 | /// How long after starting the first look for an update is, then how often. |
| 93 | const FIRST_UPDATE: Duration = Duration::from_secs(90); |
| 94 | const UPDATE_EVERY: Duration = Duration::from_secs(60 * 60); |
| 95 | /// How often a backend that cannot start is looked at again. |
| 96 | const LOOK_AGAIN: Duration = Duration::from_secs(30); |
| 97 | /// How long the app waits for the site to hand over a token before asking again. |
| 98 | const ASK_AGAIN: Duration = Duration::from_secs(60); |
| 99 | /// The longest the app waits before starting a runner that stopped by itself again. |
| 100 | const MOST_BACKOFF: Duration = Duration::from_secs(10 * 60); |
| 101 | /// The runner says when a poll fails, and nothing when one works: this |
| 102 | /// long without another failure, it is back (its longest wait is 60s). |
| 103 | const BACK_ONLINE: Duration = Duration::from_secs(75); |
| 104 | /// The runner's log is started over when it is bigger than this. |
| 105 | const LOG_LIMIT: u64 = 5 * 1024 * 1024; |
| 106 | /// The last lines of the log settings show. |
| 107 | const TAIL: usize = 40; |
| 108 | |
| 109 | /// A workspace the site says the person may add runners to: those they own. |
| 110 | #[derive(Clone, Debug, PartialEq, Eq, Serialize, Deserialize)] |
| 111 | pub struct Offered { |
| 112 | pub slug: String, |
| 113 | #[serde(default)] |
| 114 | pub name: Option<String>, |
| 115 | } |
| 116 | |
| 117 | /// What the runner is doing, as the thread last said. |
| 118 | #[derive(Clone, Debug, Default, PartialEq)] |
| 119 | struct Now { |
| 120 | code: Code, |
| 121 | detail: Option<String>, |
| 122 | exit_code: Option<i32>, |
| 123 | wait_seconds: Option<u64>, |
| 124 | job: Option<String>, |
| 125 | version: Option<String>, |
| 126 | /// What the sandbox says to do, for a cause in its own words. |
| 127 | fix: Option<String>, |
| 128 | /// How the sandbox is, while jobs run sandboxed. |
| 129 | sandbox: Option<Status>, |
| 130 | /// A sandbox run is up, or on its way. |
| 131 | live: bool, |
| 132 | /// The kinds of job a launcher command is starting now. |
| 133 | starting: Vec<String>, |
| 134 | /// How far the sandbox's image has downloaded, while it does. |
| 135 | download: Option<launcher::Download>, |
| 136 | } |
| 137 | |
| 138 | /// How the runner's updates stand, as the thread last said. |
| 139 | #[derive(Clone, Debug, Default, PartialEq)] |
| 140 | struct Updates { |
| 141 | /// When the releases last answered. |
| 142 | checked_at: Option<SystemTime>, |
| 143 | checking: bool, |
| 144 | /// A newer version, downloaded and checked, waiting. |
| 145 | ready: Option<String>, |
| 146 | /// Its program is still to be put in place. |
| 147 | program: bool, |
| 148 | /// The sandbox's image is still to take it. |
| 149 | image: bool, |
| 150 | /// The kinds of job taking it now. |
| 151 | installing: Vec<String>, |
| 152 | } |
| 153 | |
| 154 | /// What a look at the releases found. |
| 155 | struct Found { |
| 156 | version: String, |
| 157 | /// Newer than the program here. |
| 158 | newer: bool, |
| 159 | /// Its program, downloaded and checked: None when it isn't newer, or is |
| 160 | /// already waiting. |
| 161 | program: Option<Vec<u8>>, |
| 162 | /// The release has the sandbox's images. |
| 163 | images: bool, |
| 164 | } |
| 165 | |
| 166 | /// What the site last said about the person. |
| 167 | #[derive(Clone, Debug, Default, PartialEq)] |
| 168 | struct Site { |
| 169 | /// Unknown until the site's first page loads. |
| 170 | signed_in: Option<bool>, |
| 171 | owned: Vec<Offered>, |
| 172 | current: Option<String>, |
| 173 | } |
| 174 | |
| 175 | /// What the rest of the app shares with the thread. |
| 176 | #[derive(Default)] |
| 177 | pub struct Shared { |
| 178 | now: Mutex<Now>, |
| 179 | /// The workspace a token is wanted for. |
| 180 | wanted: Mutex<Option<String>>, |
| 181 | site: Mutex<Site>, |
| 182 | /// Stopped by the person until Start, or until g1t starts again. |
| 183 | paused: Mutex<bool>, |
| 184 | tail: Mutex<VecDeque<String>>, |
| 185 | tx: Mutex<Option<Sender<Ask>>>, |
| 186 | updates: Mutex<Updates>, |
| 187 | } |
| 188 | |
| 189 | /// What settings, the tray and the site are shown. |
| 190 | #[derive(Clone, Debug, PartialEq, Serialize)] |
| 191 | #[serde(rename_all = "camelCase")] |
| 192 | pub struct View { |
| 193 | pub diagnosis: Diagnosis, |
| 194 | /// What the site says of this computer once g1t is closed on it. |
| 195 | pub closed: Diagnosis, |
| 196 | pub running: bool, |
| 197 | pub version: Option<String>, |
| 198 | pub backend: Backend, |
| 199 | pub workspace: Option<String>, |
| 200 | /// The workspaces it may run for, from the site. |
| 201 | pub workspaces: Vec<Offered>, |
| 202 | /// The setting: run jobs while g1t is open. |
| 203 | pub enabled: bool, |
| 204 | pub paused: bool, |
| 205 | /// The workspace's Runners page. |
| 206 | pub link: Option<String>, |
| 207 | /// This computer's name, which the runner has on g1t. |
| 208 | pub machine: String, |
| 209 | pub platform: String, |
| 210 | /// The last lines of the runner's log. |
| 211 | pub log: Vec<String>, |
| 212 | } |
| 213 | |
| 214 | /// What the thread is told. |
| 215 | enum Ask { |
| 216 | /// Something changed: settings, Start or Stop, Retry. |
| 217 | Look, |
| 218 | /// A kind of job was started: start what should run now, stopping nothing. |
| 219 | Soon, |
| 220 | Retry, |
| 221 | /// "Check for runner updates": look now, and install what is found |
| 222 | /// even with "Download updates by themselves" off. |
| 223 | Check, |
| 224 | /// What a look at the releases found. |
| 225 | Checked(Result<Found, Fault>), |
| 226 | Token { workspace: String, token: String }, |
| 227 | Refused { workspace: String, allowed: bool, why: String }, |
| 228 | /// A line of the runner's output, or the launcher's; true from its standard output. |
| 229 | Line(u64, String, bool), |
| 230 | Exited(u64), |
| 231 | /// What `sandbox status` said. |
| 232 | Sandbox(u64, Result<Box<Status>, String>), |
| 233 | Quit(Sender<()>), |
| 234 | } |
| 235 | |
| 236 | /// The registration the app made, besides what the runner keeps itself. |
| 237 | #[derive(Clone, Debug, PartialEq, Serialize, Deserialize)] |
| 238 | struct Registration { |
| 239 | server: String, |
| 240 | workspace: String, |
| 241 | backend: Backend, |
| 242 | } |
| 243 | |
| 244 | /// Whose the sandbox's registration is. The sandbox keeps its own, and |
| 245 | /// keeps it while another backend runs, so choosing Sandboxed again needs |
| 246 | /// no new one; this says which workspace it is for. |
| 247 | #[derive(Clone, Debug, PartialEq, Serialize, Deserialize)] |
| 248 | struct SandboxRegistration { |
| 249 | server: String, |
| 250 | workspace: String, |
| 251 | /// False from the `up` that was given a token until it registered. |
| 252 | registered: bool, |
| 253 | } |
| 254 | |
| 255 | impl SandboxRegistration { |
| 256 | fn path(app: &AppHandle) -> PathBuf { |
| 257 | folder(app).join("sandbox-registration.json") |
| 258 | } |
| 259 | fn load(app: &AppHandle) -> Option<SandboxRegistration> { |
| 260 | std::fs::read(Self::path(app)).ok().and_then(|bytes| serde_json::from_slice(&bytes).ok()) |
| 261 | } |
| 262 | fn save(&self, app: &AppHandle) { |
| 263 | let _ = std::fs::create_dir_all(folder(app)); |
| 264 | let _ = std::fs::write(Self::path(app), serde_json::to_vec_pretty(self).unwrap_or_default()); |
| 265 | } |
| 266 | fn clear(app: &AppHandle) { |
| 267 | let _ = std::fs::remove_file(Self::path(app)); |
| 268 | } |
| 269 | } |
| 270 | |
| 271 | fn folder(app: &AppHandle) -> PathBuf { |
| 272 | app.path().app_data_dir().unwrap_or_else(|_| std::env::temp_dir().join("sh.g1t.desktop")).join("runner") |
| 273 | } |
| 274 | |
| 275 | impl Registration { |
| 276 | fn path(app: &AppHandle) -> PathBuf { |
| 277 | folder(app).join("registration.json") |
| 278 | } |
| 279 | fn load(app: &AppHandle) -> Option<Registration> { |
| 280 | std::fs::read(Self::path(app)).ok().and_then(|bytes| serde_json::from_slice(&bytes).ok()) |
| 281 | } |
| 282 | fn save(&self, app: &AppHandle) { |
| 283 | let _ = std::fs::create_dir_all(folder(app)); |
| 284 | let _ = std::fs::write(Self::path(app), serde_json::to_vec_pretty(self).unwrap_or_default()); |
| 285 | } |
| 286 | fn clear(app: &AppHandle) { |
| 287 | let _ = std::fs::remove_file(Self::path(app)); |
| 288 | } |
| 289 | } |
| 290 | |
| 291 | /// Where a native runner of a backend keeps its configuration. |
| 292 | fn native_dir(app: &AppHandle, backend: Backend) -> PathBuf { |
| 293 | folder(app).join(match backend { |
| 294 | Backend::Docker => "docker", |
| 295 | Backend::Host => "host", |
| 296 | Backend::Sandboxed => launcher::FOLDER, |
| 297 | }) |
| 298 | } |
| 299 | |
| 300 | /// The workspace the runner is for: the one settings name, else the one it |
| 301 | /// is registered with, else the one the person is in on the site, else the |
| 302 | /// first they own. |
| 303 | fn workspace(app: &AppHandle) -> Option<String> { |
| 304 | let settings = config::settings(app); |
| 305 | if let Some(chosen) = settings.runner_workspace.filter(|w| !w.is_empty()) { |
| 306 | return Some(chosen); |
| 307 | } |
| 308 | let server = config::server(app).to_string(); |
| 309 | if let Some(registration) = Registration::load(app).filter(|r| r.server == server) { |
| 310 | return Some(registration.workspace); |
| 311 | } |
| 312 | let site = app.state::<Shared>().site.lock().unwrap().clone(); |
| 313 | site.current.filter(|slug| site.owned.iter().any(|o| &o.slug == slug)).or_else(|| site.owned.first().map(|o| o.slug.clone())) |
| 314 | } |
| 315 | |
| 316 | /// The kinds of job a sandbox can have. |
| 317 | const KINDS: [&str; 3] = ["linux", "windows", "macos"]; |
| 318 | |
| 319 | fn facts(app: &AppHandle, now: &Now) -> Facts { |
| 320 | let server = config::server(app); |
| 321 | let settings = config::settings(app); |
| 322 | let backend = settings.runner_backend; |
| 323 | Facts { |
| 324 | workspace: workspace(app), |
| 325 | backend: backend.label(), |
| 326 | machine: backend::machine_name(), |
| 327 | job: now.job.clone(), |
| 328 | detail: now.detail.clone(), |
| 329 | exit_code: now.exit_code, |
| 330 | wait_seconds: now.wait_seconds, |
| 331 | server: server.host_str().unwrap_or("g1t").to_owned(), |
| 332 | site: server.as_str().trim_end_matches('/').to_owned(), |
| 333 | platform: std::env::consts::OS.into(), |
| 334 | fix: now.fix.clone(), |
| 335 | sandbox: now.sandbox.clone().filter(|_| backend == Backend::Sandboxed), |
| 336 | off: KINDS.iter().filter(|kind| !settings.runner_capabilities.on(kind)).map(|kind| (*kind).to_owned()).collect(), |
| 337 | live: now.live, |
| 338 | starting: now.starting.clone(), |
| 339 | download: now.download.clone(), |
| 340 | held: !settings.runner || *app.state::<Shared>().paused.lock().unwrap(), |
| 341 | update: update_facts(app, now, backend), |
| 342 | } |
| 343 | } |
| 344 | |
| 345 | /// How the runner's updates stand, for the words: what is still to take a |
| 346 | /// waiting version. The sandbox's Windows jobs (any kind but Linux) take |
| 347 | /// its program while they run; a runner with no kinds of job takes it |
| 348 | /// itself. |
| 349 | fn update_facts(app: &AppHandle, now: &Now, backend: Backend) -> diagnosis::Update { |
| 350 | let updates = app.state::<Shared>().updates.lock().unwrap().clone(); |
| 351 | let sandbox = now.sandbox.as_ref().filter(|_| backend == Backend::Sandboxed); |
| 352 | let mut waiting = Vec::new(); |
| 353 | if updates.image { |
| 354 | waiting.push("linux".to_owned()); |
| 355 | } |
| 356 | if updates.program { |
| 357 | match sandbox { |
| 358 | Some(status) => waiting.extend(native_kinds(status).into_iter().map(|c| c.id.clone())), |
| 359 | None => waiting.push("runner".to_owned()), |
| 360 | } |
| 361 | } |
| 362 | diagnosis::Update { |
| 363 | installed: native::installed(&folder(app)), |
| 364 | checked_at: updates.checked_at.map(rfc3339), |
| 365 | checked_ago: updates.checked_at.and_then(|at| at.elapsed().ok()).map(|ago| ago.as_secs()), |
| 366 | checking: updates.checking, |
| 367 | ready: updates.ready, |
| 368 | waiting, |
| 369 | installing: updates.installing, |
| 370 | } |
| 371 | } |
| 372 | |
| 373 | /// The kinds of job running on the program itself, which replacing it |
| 374 | /// stops: the sandbox's Windows jobs (not Linux jobs, which run in its VM). |
| 375 | fn native_kinds(status: &Status) -> Vec<&launcher::Capability> { |
| 376 | status.capabilities.iter().filter(|c| c.id != "linux" && !c.off() && matches!(c.state, State::Running | State::Starting)).collect() |
| 377 | } |
| 378 | |
| 379 | /// A time as RFC 3339, in UTC to the second: `2026-10-12T12:04:05Z`. |
| 380 | fn rfc3339(at: SystemTime) -> String { |
| 381 | let seconds = at.duration_since(SystemTime::UNIX_EPOCH).map_or(0, |since| since.as_secs()); |
| 382 | let (days, rest) = (seconds / 86_400, seconds % 86_400); |
| 383 | // Days since 1970 to a date in the proleptic Gregorian calendar (Howard Hinnant's civil_from_days). |
| 384 | let z = days as i64 + 719_468; |
| 385 | let era = z.div_euclid(146_097); |
| 386 | let day_of_era = z.rem_euclid(146_097); |
| 387 | let year_of_era = (day_of_era - day_of_era / 1460 + day_of_era / 36_524 - day_of_era / 146_096) / 365; |
| 388 | let day_of_year = day_of_era - (365 * year_of_era + year_of_era / 4 - year_of_era / 100); |
| 389 | let month_index = (5 * day_of_year + 2) / 153; |
| 390 | let day = day_of_year - (153 * month_index + 2) / 5 + 1; |
| 391 | let month = if month_index < 10 { month_index + 3 } else { month_index - 9 }; |
| 392 | let year = year_of_era + era * 400 + i64::from(month <= 2); |
| 393 | format!("{year:04}-{month:02}-{day:02}T{:02}:{:02}:{:02}Z", rest / 3600, rest % 3600 / 60, rest % 60) |
| 394 | } |
| 395 | |
| 396 | pub fn view(app: &AppHandle) -> View { |
| 397 | let shared = app.state::<Shared>(); |
| 398 | let now = shared.now.lock().unwrap().clone(); |
| 399 | let settings = config::settings(app); |
| 400 | let facts = facts(app, &now); |
| 401 | let server = config::server(app); |
| 402 | View { |
| 403 | diagnosis: diagnosis::diagnose(now.code, &facts), |
| 404 | closed: diagnosis::closed(&facts), |
| 405 | running: matches!(now.code, Code::Idle | Code::Busy | Code::Offline) || facts.sandbox.as_ref().is_some_and(Status::running), |
| 406 | version: now.version, |
| 407 | backend: settings.runner_backend, |
| 408 | link: facts.workspace.as_ref().and_then(|w| server.join(&format!("{w}/-/runners")).ok()).map(|u| u.to_string()), |
| 409 | workspace: facts.workspace, |
| 410 | workspaces: shared.site.lock().unwrap().owned.clone(), |
| 411 | enabled: settings.runner, |
| 412 | paused: *shared.paused.lock().unwrap(), |
| 413 | machine: facts.machine, |
| 414 | platform: facts.platform, |
| 415 | log: shared.tail.lock().unwrap().iter().cloned().collect(), |
| 416 | } |
| 417 | } |
| 418 | |
| 419 | /// Whether the runner is running a job: what restarts g1t waits for it, |
| 420 | /// since restarting stops the runner and would cut the job short. |
| 421 | pub fn busy(app: &AppHandle) -> bool { |
| 422 | let now = app.state::<Shared>().now.lock().unwrap().clone(); |
| 423 | now.code == Code::Busy || now.sandbox.as_ref().is_some_and(|status| status.job().is_some()) |
| 424 | } |
| 425 | |
| 426 | /// The tray's line: `Runner: Waiting for jobs`, at most 60 characters. |
| 427 | pub fn headline(view: &View) -> String { |
| 428 | let line = format!("Runner: {}", view.diagnosis.title); |
| 429 | if line.chars().count() > 60 { line.chars().take(59).collect::<String>() + "…" } else { line } |
| 430 | } |
| 431 | |
| 432 | /// Tells everyone what the runner is doing now. |
| 433 | fn publish(app: &AppHandle) { |
| 434 | let view = view(app); |
| 435 | tray::runner_changed(app, &view); |
| 436 | let _ = app.emit_to(settings::SETTINGS, STATUS_EVENT, view.clone()); |
| 437 | let _ = app.emit_to(EventTarget::webview(frame::SITE), STATUS_EVENT, view); |
| 438 | } |
| 439 | |
| 440 | fn ask(app: &AppHandle, ask: Ask) { |
| 441 | if let Some(tx) = app.state::<Shared>().tx.lock().unwrap().as_ref() { |
| 442 | let _ = tx.send(ask); |
| 443 | } |
| 444 | } |
| 445 | |
| 446 | /// Starts the thread that keeps the runner, as the app starts. |
| 447 | pub fn install(app: &AppHandle) { |
| 448 | let (tx, rx) = mpsc::channel(); |
| 449 | *app.state::<Shared>().tx.lock().unwrap() = Some(tx.clone()); |
| 450 | let handle = app.clone(); |
| 451 | let spawned = std::thread::Builder::new().name("g1t-runner".into()).spawn(move || Keeper::new(handle, tx).keep(rx)); |
| 452 | if let Err(error) = spawned { |
| 453 | eprintln!("g1t: the runner's thread could not start: {error}"); |
| 454 | } |
| 455 | } |
| 456 | |
| 457 | /// Settings changed: the thread looks again. |
| 458 | pub fn settings_changed(app: &AppHandle) { |
| 459 | ask(app, Ask::Look); |
| 460 | publish(app); |
| 461 | } |
| 462 | |
| 463 | /// Start or Stop, from settings or the tray. |
| 464 | pub fn set_paused(app: &AppHandle, paused: bool) { |
| 465 | *app.state::<Shared>().paused.lock().unwrap() = paused; |
| 466 | // Start with the setting off turns the setting on: that is what was asked. |
| 467 | if !paused && !config::settings(app).runner { |
| 468 | let mut settings = config::settings(app); |
| 469 | settings.runner = true; |
| 470 | let _ = settings::apply(app, settings); |
| 471 | } |
| 472 | settings_changed(app); |
| 473 | } |
| 474 | |
| 475 | /// Start or Stop on one kind of job's card: kept in settings, and the |
| 476 | /// thread starts or stops that kind alone. Starting one with the runner |
| 477 | /// off or stopped starts the runner too: that is what was asked. |
| 478 | pub fn set_kind(app: &AppHandle, kind: &str, on: bool) -> Result<(), String> { |
| 479 | let mut settings = config::settings(app); |
| 480 | if !settings.runner_capabilities.set(kind, on) { |
| 481 | return Err(format!("g1t runs no {kind} jobs")); |
| 482 | } |
| 483 | if on { |
| 484 | settings.runner = true; |
| 485 | *app.state::<Shared>().paused.lock().unwrap() = false; |
| 486 | } |
| 487 | settings::apply(app, settings)?; |
| 488 | if on { |
| 489 | ask(app, Ask::Soon); |
| 490 | } |
| 491 | settings_changed(app); |
| 492 | Ok(()) |
| 493 | } |
| 494 | |
| 495 | /// The tray's Start or Stop row. |
| 496 | pub fn toggle(app: &AppHandle) { |
| 497 | let view = view(app); |
| 498 | set_paused(app, view.enabled && !view.paused); |
| 499 | } |
| 500 | |
| 501 | /// Retry and Restart: now, rather than when the wait is over. |
| 502 | pub fn retry(app: &AppHandle) { |
| 503 | *app.state::<Shared>().paused.lock().unwrap() = false; |
| 504 | ask(app, Ask::Retry); |
| 505 | } |
| 506 | |
| 507 | /// What the site says about the person, on every page: whether they are |
| 508 | /// signed in, the workspaces they own, and the one they are in. |
| 509 | pub fn offer(app: &AppHandle, signed_in: bool, owned: Vec<Offered>, current: Option<String>) { |
| 510 | let next = Site { signed_in: Some(signed_in), owned, current: current.map(|c| c.to_ascii_lowercase()) }; |
| 511 | let changed = { |
| 512 | let shared = app.state::<Shared>(); |
| 513 | let mut site = shared.site.lock().unwrap(); |
| 514 | let changed = *site != next; |
| 515 | *site = next; |
| 516 | changed |
| 517 | }; |
| 518 | if changed { |
| 519 | settings_changed(app); |
| 520 | } |
| 521 | } |
| 522 | |
| 523 | /// The site handed over a registration token, or said why it could not: |
| 524 | /// `allowed` is false when the person may not add runners to the workspace. |
| 525 | pub fn register(app: &AppHandle, workspace: String, token: Option<String>, allowed: bool, error: Option<String>) { |
| 526 | match token { |
| 527 | Some(token) if token.starts_with("g1trt_") => ask(app, Ask::Token { workspace, token }), |
| 528 | _ => ask(app, Ask::Refused { workspace, allowed, why: error.unwrap_or_else(|| "g1t didn't make a registration token.".into()) }), |
| 529 | } |
| 530 | } |
| 531 | |
| 532 | /// A diagnosis's button, or Start and Stop: from the settings window, or |
| 533 | /// from the site's card for this computer. With a kind of job (`linux`, |
| 534 | /// `windows`), Start and Stop are that kind's alone. |
| 535 | pub fn act(app: &AppHandle, action: &str, kind: Option<&str>) -> Result<(), String> { |
| 536 | use tauri_plugin_opener::OpenerExt; |
| 537 | let browse = |url: &str| app.opener().open_url(url, None::<&str>).map_err(|e| e.to_string()); |
| 538 | if let ("start" | "stop", Some(kind)) = (action, kind) { |
| 539 | return set_kind(app, kind, action == "start"); |
| 540 | } |
| 541 | match action { |
| 542 | "start" => { |
| 543 | set_paused(app, false); |
| 544 | retry(app); |
| 545 | } |
| 546 | "stop" => set_paused(app, true), |
| 547 | "retry" | "restart" => retry(app), |
| 548 | "check_updates" => ask(app, Ask::Check), |
| 549 | "sign_in" => crate::links::go(app, "/login"), |
| 550 | "new_workspace" => crate::links::go(app, "/workspaces/new"), |
| 551 | "view_runners" => match workspace(app) { |
| 552 | Some(workspace) => crate::links::go(app, &format!("/{workspace}/-/runners")), |
| 553 | None => crate::links::go(app, "/"), |
| 554 | }, |
| 555 | "choose_workspace" | "choose_backend" => settings::open(app), |
| 556 | "view_log" => { |
| 557 | let path = log_file(app); |
| 558 | if !path.exists() { |
| 559 | return Err("The runner hasn't written a log yet.".into()); |
| 560 | } |
| 561 | app.opener().open_path(path.to_string_lossy(), None::<&str>).map_err(|e| e.to_string())?; |
| 562 | } |
| 563 | "open_docker" => open_docker(app)?, |
| 564 | "copy_wsl_install" => copy(launcher::WSL_INSTALL)?, |
| 565 | "copy_vfkit_install" => copy(launcher::VFKIT_INSTALL)?, |
| 566 | "windows_features" => windows_features()?, |
| 567 | "get_docker" => browse("https://docs.docker.com/get-started/get-docker/")?, |
| 568 | "install_sysbox" => browse("https://github.com/nestybox/sysbox#installation")?, |
| 569 | "learn_more" => browse("https://docs.g1t.sh/guides/desktop/#how-jobs-run")?, |
| 570 | "report" => browse(config::server(app).join("support").map_err(|e| e.to_string())?.as_str())?, |
| 571 | other => return Err(format!("the runner has no action called {other}")), |
| 572 | } |
| 573 | Ok(()) |
| 574 | } |
| 575 | |
| 576 | /// Starts Docker Desktop, where it is installed in its usual place. |
| 577 | fn open_docker(app: &AppHandle) -> Result<(), String> { |
| 578 | use tauri_plugin_opener::OpenerExt; |
| 579 | if cfg!(target_os = "macos") { |
| 580 | return app.opener().open_path("/Applications/Docker.app", None::<&str>).map_err(|e| e.to_string()); |
| 581 | } |
| 582 | if cfg!(windows) { |
| 583 | let program = std::env::var_os("ProgramFiles").map(PathBuf::from).unwrap_or_else(|| PathBuf::from(r"C:\Program Files")); |
| 584 | let desktop = program.join("Docker").join("Docker").join("Docker Desktop.exe"); |
| 585 | if desktop.exists() { |
| 586 | return app.opener().open_path(desktop.to_string_lossy(), None::<&str>).map_err(|e| e.to_string()); |
| 587 | } |
| 588 | return Err("Docker Desktop isn't installed in its usual place. Get Docker to install it.".into()); |
| 589 | } |
| 590 | Err("Start Docker with your system's service manager, such as `sudo systemctl start docker`.".into()) |
| 591 | } |
| 592 | |
| 593 | /// Puts a command on the clipboard, for the person to paste where it runs. |
| 594 | fn copy(text: &str) -> Result<(), String> { |
| 595 | let tools: &[(&str, &[&str])] = if cfg!(windows) { |
| 596 | &[("clip", &[])] |
| 597 | } else if cfg!(target_os = "macos") { |
| 598 | &[("pbcopy", &[])] |
| 599 | } else { |
| 600 | &[("wl-copy", &[]), ("xclip", &["-selection", "clipboard"]), ("xsel", &["--clipboard", "--input"])] |
| 601 | }; |
| 602 | for (program, args) in tools { |
| 603 | let Ok(mut child) = backend::quiet(program).args(*args).stdin(Stdio::piped()).stdout(Stdio::null()).stderr(Stdio::null()).spawn() else { continue }; |
| 604 | if let Some(mut input) = child.stdin.take() { |
| 605 | let _ = input.write_all(text.as_bytes()); |
| 606 | } |
| 607 | if child.wait().is_ok_and(|status| status.success()) { |
| 608 | return Ok(()); |
| 609 | } |
| 610 | } |
| 611 | Err(format!("It couldn't be copied. The command is: {text}")) |
| 612 | } |
| 613 | |
| 614 | /// Windows' "Turn Windows features on or off", where Windows Sandbox and |
| 615 | /// Virtual Machine Platform are. |
| 616 | fn windows_features() -> Result<(), String> { |
| 617 | if !cfg!(windows) { |
| 618 | return Err("Windows features are only on Windows.".into()); |
| 619 | } |
| 620 | backend::quiet("optionalfeatures.exe").spawn().map(|_| ()).map_err(|e| format!("Windows features couldn't be opened: {e}")) |
| 621 | } |
| 622 | |
| 623 | /// The workspace a token is wanted for now, if one is: the site asks as it loads. |
| 624 | pub fn wanted(app: &AppHandle) -> Option<String> { |
| 625 | app.state::<Shared>().wanted.lock().unwrap().clone() |
| 626 | } |
| 627 | |
| 628 | /// The runner's log, for the settings window to open. |
| 629 | pub fn log_file(app: &AppHandle) -> PathBuf { |
| 630 | folder(app).join("runner.log") |
| 631 | } |
| 632 | |
| 633 | /// Stops the runner as g1t quits or restarts, and waits for it. |
| 634 | pub fn shutdown(app: &AppHandle) { |
| 635 | let tx = app.state::<Shared>().tx.lock().unwrap().take(); |
| 636 | if let Some(tx) = tx { |
| 637 | let (done, wait) = mpsc::channel(); |
| 638 | if tx.send(Ask::Quit(done)).is_ok() { |
| 639 | let _ = wait.recv_timeout(Duration::from_secs(30)); |
| 640 | } |
| 641 | } |
| 642 | } |
| 643 | |
| 644 | /// How the runner is started this time. |
| 645 | enum Process { |
| 646 | /// `g1t-runner`, a child of the app, whose output is the runner's log. |
| 647 | Child(Child), |
| 648 | /// The sandbox, which runs on its own once `up` is done; the launcher |
| 649 | /// command the app is waiting for, while there is one. |
| 650 | Sandbox(Option<(Step, Child)>), |
| 651 | } |
| 652 | |
| 653 | /// A launcher command that takes a while. |
| 654 | #[derive(Clone, Debug, PartialEq, Eq)] |
| 655 | enum Step { |
| 656 | Up, |
| 657 | Update, |
| 658 | /// `up --only <kind>`: one kind of job, started while the rest run. |
| 659 | Start(String), |
| 660 | /// `down --only <kind>`: one kind of job, stopped, the rest running on. |
| 661 | Stop(String), |
| 662 | /// `down --only <kind> --drain`: one kind of job, stopped after its |
| 663 | /// job, so the program its runner is can be replaced. |
| 664 | Drain(String), |
| 665 | } |
| 666 | |
| 667 | impl Step { |
| 668 | fn name(&self) -> &'static str { |
| 669 | match self { |
| 670 | Step::Up | Step::Start(_) => "up", |
| 671 | Step::Update => "update", |
| 672 | Step::Stop(_) | Step::Drain(_) => "down", |
| 673 | } |
| 674 | } |
| 675 | } |
| 676 | |
| 677 | struct Run { |
| 678 | generation: u64, |
| 679 | engine: Engine, |
| 680 | backend: Backend, |
| 681 | workspace: String, |
| 682 | process: Process, |
| 683 | /// Registered and polling. |
| 684 | listening: bool, |
| 685 | busy: bool, |
| 686 | /// When a poll last failed to reach g1t. |
| 687 | offline_at: Option<Instant>, |
| 688 | /// Started with a registration token, which a container keeps in its configuration until replaced. |
| 689 | with_token: bool, |
| 690 | forgotten: bool, |
| 691 | token_refused: bool, |
| 692 | error: Option<String>, |
| 693 | /// The status a launcher command is printing, line by line. |
| 694 | json: Option<Vec<String>>, |
| 695 | /// Asking the sandbox how it is, and when it was last asked. |
| 696 | polling: bool, |
| 697 | polled_at: Option<Instant>, |
| 698 | /// A poll asked before a launcher command ended, whose answer is old. |
| 699 | stale: bool, |
| 700 | } |
| 701 | |
| 702 | impl Run { |
| 703 | fn new(generation: u64, engine: Engine, backend: Backend, workspace: String, process: Process, with_token: bool) -> Run { |
| 704 | Run { |
| 705 | generation, |
| 706 | engine, |
| 707 | backend, |
| 708 | workspace, |
| 709 | process, |
| 710 | listening: false, |
| 711 | busy: false, |
| 712 | offline_at: None, |
| 713 | with_token, |
| 714 | forgotten: false, |
| 715 | token_refused: false, |
| 716 | error: None, |
| 717 | json: None, |
| 718 | polling: false, |
| 719 | polled_at: None, |
| 720 | stale: false, |
| 721 | } |
| 722 | } |
| 723 | |
| 724 | fn step(&self) -> Option<Step> { |
| 725 | match &self.process { |
| 726 | Process::Sandbox(Some((step, _))) => Some(step.clone()), |
| 727 | _ => None, |
| 728 | } |
| 729 | } |
| 730 | } |
| 731 | |
| 732 | struct Keeper { |
| 733 | app: AppHandle, |
| 734 | tx: Sender<Ask>, |
| 735 | run: Option<Run>, |
| 736 | generation: u64, |
| 737 | token: Option<(String, String)>, |
| 738 | not_before: Instant, |
| 739 | failures: u32, |
| 740 | next_update: Instant, |
| 741 | log: Option<File>, |
| 742 | /// Why the last update failed (`UpdateFailed`, or `UpdateRefused` for |
| 743 | /// a release g1t didn't sign), said while the runner is idle, until |
| 744 | /// the next one works. |
| 745 | update_failed: Option<(Code, String)>, |
| 746 | /// A newer program, downloaded and checked, waiting for the kinds of |
| 747 | /// job that run on it to be idle: its version and its bytes. |
| 748 | program: Option<(String, Vec<u8>)>, |
| 749 | /// The person asked for this update: it installs with "Download |
| 750 | /// updates by themselves" off. |
| 751 | asked: bool, |
| 752 | /// When g1t last looked, as settings last said it. |
| 753 | said_checked: String, |
| 754 | /// A kind of job that should run and doesn't, while another runs: |
| 755 | /// when it is tried again, and how many times it has been. |
| 756 | retry: Vec<(String, Instant, u32)>, |
| 757 | } |
| 758 | |
| 759 | /// What a set of the state carries besides its code. |
| 760 | #[derive(Default)] |
| 761 | struct Extra { |
| 762 | detail: Option<String>, |
| 763 | exit_code: Option<i32>, |
| 764 | wait_seconds: Option<u64>, |
| 765 | fix: Option<String>, |
| 766 | } |
| 767 | |
| 768 | fn detail(text: impl Into<String>) -> Extra { |
| 769 | Extra { detail: Some(text.into()), ..Extra::default() } |
| 770 | } |
| 771 | |
| 772 | impl Keeper { |
| 773 | fn new(app: AppHandle, tx: Sender<Ask>) -> Keeper { |
| 774 | let now = Instant::now(); |
| 775 | Keeper { |
| 776 | app, |
| 777 | tx, |
| 778 | run: None, |
| 779 | generation: 0, |
| 780 | token: None, |
| 781 | not_before: now + AFTER_START, |
| 782 | failures: 0, |
| 783 | next_update: now + FIRST_UPDATE, |
| 784 | log: None, |
| 785 | update_failed: None, |
| 786 | program: None, |
| 787 | asked: false, |
| 788 | said_checked: String::new(), |
| 789 | retry: Vec::new(), |
| 790 | } |
| 791 | } |
| 792 | |
| 793 | fn keep(mut self, rx: Receiver<Ask>) { |
| 794 | self.reap(); |
| 795 | self.set(Code::Starting, detail("Starting after g1t's window.")); |
| 796 | loop { |
| 797 | match rx.recv_timeout(TICK) { |
| 798 | Ok(Ask::Quit(done)) => { |
| 799 | self.halt(); |
| 800 | let _ = done.send(()); |
| 801 | return; |
| 802 | } |
| 803 | Ok(ask) => self.handle(ask), |
| 804 | Err(RecvTimeoutError::Timeout) => {} |
| 805 | Err(RecvTimeoutError::Disconnected) => { |
| 806 | self.halt(); |
| 807 | return; |
| 808 | } |
| 809 | } |
| 810 | self.look(); |
| 811 | self.said_when(); |
| 812 | } |
| 813 | } |
| 814 | |
| 815 | /// Settings say when g1t last looked for a newer runner, in minutes |
| 816 | /// and hours: said again as that changes. |
| 817 | fn said_when(&mut self) { |
| 818 | let (checked_at, checking) = self.with_updates(|updates| (updates.checked_at, updates.checking)); |
| 819 | let checked_ago = checked_at.and_then(|at| at.elapsed().ok()).map(|ago| ago.as_secs()); |
| 820 | let checked = diagnosis::checked(&diagnosis::Update { checked_ago, checking, ..diagnosis::Update::default() }); |
| 821 | if checked != self.said_checked { |
| 822 | self.said_checked = checked; |
| 823 | publish(&self.app); |
| 824 | } |
| 825 | } |
| 826 | |
| 827 | fn with_updates<R>(&self, change: impl FnOnce(&mut Updates) -> R) -> R { |
| 828 | change(&mut self.app.state::<Shared>().updates.lock().unwrap()) |
| 829 | } |
| 830 | |
| 831 | /// What an idle runner shows: why its last update failed, until one works. |
| 832 | fn idle(&self, code: Code) -> (Code, Extra) { |
| 833 | match (&self.update_failed, code) { |
| 834 | (Some((failed, why)), Code::Idle) => (*failed, detail(why.clone())), |
| 835 | _ => (code, Extra::default()), |
| 836 | } |
| 837 | } |
| 838 | |
| 839 | fn set(&mut self, code: Code, extra: Extra) { |
| 840 | let changed = { |
| 841 | let shared = self.app.state::<Shared>(); |
| 842 | let mut now = shared.now.lock().unwrap(); |
| 843 | let job = if code == Code::Busy { now.job.clone() } else { None }; |
| 844 | let next = Now { |
| 845 | code, |
| 846 | detail: extra.detail, |
| 847 | exit_code: extra.exit_code, |
| 848 | wait_seconds: extra.wait_seconds, |
| 849 | job, |
| 850 | fix: extra.fix, |
| 851 | live: self.run.as_ref().is_some_and(|run| run.engine == Engine::Launcher), |
| 852 | ..now.clone() |
| 853 | }; |
| 854 | let changed = *now != next; |
| 855 | *now = next; |
| 856 | changed |
| 857 | }; |
| 858 | if changed { |
| 859 | publish(&self.app); |
| 860 | } |
| 861 | } |
| 862 | |
| 863 | fn code(&self) -> Code { |
| 864 | self.app.state::<Shared>().now.lock().unwrap().code |
| 865 | } |
| 866 | |
| 867 | fn note(&mut self, job: Option<String>, version: Option<String>) { |
| 868 | let shared = self.app.state::<Shared>(); |
| 869 | let mut now = shared.now.lock().unwrap(); |
| 870 | if job.is_some() { |
| 871 | now.job = job; |
| 872 | } |
| 873 | if version.is_some() { |
| 874 | now.version = version; |
| 875 | } |
| 876 | } |
| 877 | |
| 878 | /// A line of the runner's log, into the log file and the tail settings show. |
| 879 | fn write_log(&mut self, line: &str) { |
| 880 | { |
| 881 | let shared = self.app.state::<Shared>(); |
| 882 | let mut tail = shared.tail.lock().unwrap(); |
| 883 | if tail.len() == TAIL { |
| 884 | tail.pop_front(); |
| 885 | } |
| 886 | tail.push_back(line.to_owned()); |
| 887 | } |
| 888 | let path = log_file(&self.app); |
| 889 | if self.log.is_none() { |
| 890 | let _ = std::fs::create_dir_all(folder(&self.app)); |
| 891 | if std::fs::metadata(&path).is_ok_and(|m| m.len() > LOG_LIMIT) { |
| 892 | let _ = std::fs::rename(&path, path.with_extension("log.old")); |
| 893 | } |
| 894 | self.log = OpenOptions::new().create(true).append(true).open(&path).ok(); |
| 895 | } |
| 896 | if let Some(file) = self.log.as_mut() { |
| 897 | let _ = writeln!(file, "{line}"); |
| 898 | } |
| 899 | } |
| 900 | |
| 901 | fn handle(&mut self, ask: Ask) { |
| 902 | match ask { |
| 903 | Ask::Look | Ask::Quit(_) => {} |
| 904 | Ask::Soon => { |
| 905 | self.not_before = Instant::now(); |
| 906 | for (_, at, _) in &mut self.retry { |
| 907 | *at = Instant::now(); |
| 908 | } |
| 909 | } |
| 910 | Ask::Retry => { |
| 911 | self.not_before = Instant::now(); |
| 912 | self.failures = 0; |
| 913 | // Each kind of job that waits to be tried again is tried now. |
| 914 | for (_, at, _) in &mut self.retry { |
| 915 | *at = Instant::now(); |
| 916 | } |
| 917 | match self.code() { |
| 918 | // The runner is fine: only the update is tried again. |
| 919 | Code::UpdateFailed | Code::UpdateRefused if self.run.is_some() => self.check(true), |
| 920 | Code::Offline => self.halt(), |
| 921 | _ if self.run.as_ref().is_some_and(|run| !run.listening) => self.halt(), |
| 922 | _ => {} |
| 923 | } |
| 924 | } |
| 925 | Ask::Check => self.check(true), |
| 926 | Ask::Checked(found) => self.checked(found), |
| 927 | Ask::Token { workspace, token } => { |
| 928 | let shared = self.app.state::<Shared>(); |
| 929 | let mut wanted = shared.wanted.lock().unwrap(); |
| 930 | if wanted.as_deref() == Some(workspace.as_str()) { |
| 931 | *wanted = None; |
| 932 | self.token = Some((workspace, token)); |
| 933 | self.not_before = Instant::now(); |
| 934 | } |
| 935 | } |
| 936 | Ask::Refused { workspace, allowed, why } => { |
| 937 | let matched = { |
| 938 | let shared = self.app.state::<Shared>(); |
| 939 | let mut wanted = shared.wanted.lock().unwrap(); |
| 940 | let matched = wanted.as_deref() == Some(workspace.as_str()); |
| 941 | if matched { |
| 942 | *wanted = None; |
| 943 | } |
| 944 | matched |
| 945 | }; |
| 946 | if matched { |
| 947 | self.not_before = Instant::now() + MOST_BACKOFF; |
| 948 | self.set(if allowed { Code::RegisterFailed } else { Code::NotAllowed }, detail(why)); |
| 949 | } |
| 950 | } |
| 951 | Ask::Line(generation, line, out) => self.line(generation, &line, out), |
| 952 | Ask::Exited(generation) => self.exited(generation), |
| 953 | Ask::Sandbox(generation, status) => self.polled(generation, status), |
| 954 | } |
| 955 | } |
| 956 | |
| 957 | fn line(&mut self, generation: u64, line: &str, out: bool) { |
| 958 | let Some(run) = self.run.as_mut().filter(|run| run.generation == generation) else { return }; |
| 959 | if run.engine == Engine::Launcher { |
| 960 | // The launcher's log, then its status, which is read when it ends. |
| 961 | if out && (run.json.is_some() || launcher::starts_status(line)) { |
| 962 | run.json.get_or_insert_with(Vec::new).push(line.to_owned()); |
| 963 | } else { |
| 964 | if let Some(download) = launcher::download(line) { |
| 965 | let changed = { |
| 966 | let shared = self.app.state::<Shared>(); |
| 967 | let mut now = shared.now.lock().unwrap(); |
| 968 | let changed = now.download.as_ref() != Some(&download); |
| 969 | now.download = Some(download); |
| 970 | changed |
| 971 | }; |
| 972 | if changed { |
| 973 | publish(&self.app); |
| 974 | } |
| 975 | } |
| 976 | self.write_log(line); |
| 977 | } |
| 978 | return; |
| 979 | } |
| 980 | let mut became = None; |
| 981 | let mut job = None; |
| 982 | let mut version = None; |
| 983 | match log::read(line) { |
| 984 | log::Event::Listening { version: v } => { |
| 985 | run.listening = true; |
| 986 | version = Some(v); |
| 987 | became = Some(Code::Idle); |
| 988 | } |
| 989 | log::Event::Took { name, repo } => { |
| 990 | run.busy = true; |
| 991 | run.offline_at = None; |
| 992 | job = Some(format!("{name} ({repo})")); |
| 993 | became = Some(Code::Busy); |
| 994 | } |
| 995 | log::Event::Ended => { |
| 996 | run.busy = false; |
| 997 | became = Some(Code::Idle); |
| 998 | } |
| 999 | log::Event::Offline => { |
| 1000 | run.offline_at = Some(Instant::now()); |
| 1001 | if !run.busy { |
| 1002 | became = Some(Code::Offline); |
| 1003 | } |
| 1004 | } |
| 1005 | log::Event::Forgotten => run.forgotten = true, |
| 1006 | log::Event::TokenRefused => run.token_refused = true, |
| 1007 | log::Event::Failed(error) => run.error = Some(error), |
| 1008 | log::Event::Other => {} |
| 1009 | } |
| 1010 | let registered = version.is_some().then(|| Registration { |
| 1011 | server: config::server(&self.app).to_string(), |
| 1012 | workspace: run.workspace.clone(), |
| 1013 | backend: run.backend, |
| 1014 | }); |
| 1015 | self.write_log(line); |
| 1016 | if let Some(registration) = registered { |
| 1017 | registration.save(&self.app); |
| 1018 | self.failures = 0; |
| 1019 | } |
| 1020 | self.note(job, version); |
| 1021 | if let Some(code) = became { |
| 1022 | let (code, extra) = self.idle(code); |
| 1023 | self.set(code, extra); |
| 1024 | } |
| 1025 | } |
| 1026 | |
| 1027 | fn exited(&mut self, generation: u64) { |
| 1028 | if self.run.as_ref().is_none_or(|run| run.generation != generation) { |
| 1029 | return; |
| 1030 | } |
| 1031 | if self.run.as_ref().is_some_and(|run| run.engine == Engine::Launcher) { |
| 1032 | return self.launched(generation); |
| 1033 | } |
| 1034 | let mut run = self.run.take().expect("checked above"); |
| 1035 | let exit_code = match &mut run.process { |
| 1036 | Process::Child(child) => child.wait().ok().and_then(|status| status.code()), |
| 1037 | Process::Sandbox(_) => None, |
| 1038 | }; |
| 1039 | self.stop(&mut run, Down::Now); |
| 1040 | if run.forgotten || run.token_refused { |
| 1041 | // The registration is no more, or was never made: the next start asks for a token. |
| 1042 | self.forget(); |
| 1043 | self.not_before = Instant::now(); |
| 1044 | return; |
| 1045 | } |
| 1046 | self.failures += 1; |
| 1047 | let wait = Duration::from_secs(15 * 2u64.pow(self.failures.min(6))).min(MOST_BACKOFF); |
| 1048 | self.not_before = Instant::now() + wait; |
| 1049 | let offline = run.error.as_deref().is_some_and(|e| e.contains("could not reach g1t")); |
| 1050 | if offline { |
| 1051 | self.set(Code::Offline, Extra::default()); |
| 1052 | } else if !run.listening && Registration::load(&self.app).is_none() { |
| 1053 | self.set(Code::RegisterFailed, Extra { detail: run.error, ..Extra::default() }); |
| 1054 | } else { |
| 1055 | self.set(Code::Crashed, Extra { detail: run.error, exit_code, wait_seconds: Some(wait.as_secs()), ..Extra::default() }); |
| 1056 | } |
| 1057 | } |
| 1058 | |
| 1059 | /// Stops a run's process, the way it was started: the sandbox `how` says. |
| 1060 | fn stop(&mut self, run: &mut Run, how: Down) { |
| 1061 | match (&mut run.process, run.engine) { |
| 1062 | (Process::Sandbox(step), _) => { |
| 1063 | // A launcher command still going is ended first, then the sandbox stopped. |
| 1064 | if let Some((_, child)) = step.as_mut() { |
| 1065 | native::stop(child); |
| 1066 | } |
| 1067 | let binary = native::binary(&folder(&self.app)); |
| 1068 | // How it is once stopped, so each kind of job's card says so. |
| 1069 | let stopped = match launcher::down(&binary, &native_dir(&self.app, Backend::Sandboxed), how) { |
| 1070 | Ok(status) => Some(status), |
| 1071 | Err(why) => { |
| 1072 | self.write_log(&format!("g1t: the sandbox didn't stop cleanly: {why}")); |
| 1073 | None |
| 1074 | } |
| 1075 | }; |
| 1076 | let shared = self.app.state::<Shared>(); |
| 1077 | let mut now = shared.now.lock().unwrap(); |
| 1078 | now.sandbox = stopped.filter(|_| how != Down::Delete); |
| 1079 | now.starting.clear(); |
| 1080 | now.download = None; |
| 1081 | } |
| 1082 | (Process::Child(child), Engine::Launcher) => native::stop(child), |
| 1083 | (Process::Child(child), Engine::Native { docker }) => { |
| 1084 | native::stop(child); |
| 1085 | // A job's container outlives the runner killed under it: removed now, not at its next start. |
| 1086 | if docker { |
| 1087 | native::sweep(&native_dir(&self.app, run.backend)); |
| 1088 | } |
| 1089 | } |
| 1090 | } |
| 1091 | let _ = std::fs::remove_file(folder(&self.app).join("pid")); |
| 1092 | } |
| 1093 | |
| 1094 | /// Stops whatever runs now. |
| 1095 | fn halt(&mut self) { |
| 1096 | self.halt_as(Down::Now); |
| 1097 | } |
| 1098 | |
| 1099 | /// Stops whatever runs: the sandbox now, after its job, or for good. |
| 1100 | fn halt_as(&mut self, how: Down) { |
| 1101 | self.generation += 1; |
| 1102 | if let Some(mut run) = self.run.take() { |
| 1103 | self.stop(&mut run, how); |
| 1104 | } |
| 1105 | } |
| 1106 | |
| 1107 | /// A runner an earlier g1t started and did not stop: it ended without quitting. |
| 1108 | fn reap(&self) { |
| 1109 | if let Some(pid) = std::fs::read_to_string(folder(&self.app).join("pid")).ok().and_then(|text| text.trim().parse().ok()) { |
| 1110 | native::reap(pid); |
| 1111 | let _ = std::fs::remove_file(folder(&self.app).join("pid")); |
| 1112 | } |
| 1113 | } |
| 1114 | |
| 1115 | /// Unregisters the runner the app registered, whatever backend it was. |
| 1116 | fn forget(&mut self) { |
| 1117 | let app = self.app.clone(); |
| 1118 | let Some(registration) = Registration::load(&app) else { return }; |
| 1119 | let binary = native::binary(&folder(&app)); |
| 1120 | match registration.backend { |
| 1121 | // The sandbox keeps its registration for when it is chosen |
| 1122 | // again (sandbox-registration.json says whose): it only stops, |
| 1123 | // after its job. |
| 1124 | Backend::Sandboxed => { |
| 1125 | let _ = launcher::down(&binary, &native_dir(&app, Backend::Sandboxed), Down::Drain); |
| 1126 | } |
| 1127 | backend => native::forget(&binary, &native_dir(&app, backend)), |
| 1128 | } |
| 1129 | Registration::clear(&app); |
| 1130 | } |
| 1131 | |
| 1132 | /// Removes the sandbox and its registration: it was for another |
| 1133 | /// workspace, or g1t no longer knows its runners. |
| 1134 | fn forget_sandbox(&mut self) { |
| 1135 | let binary = native::binary(&folder(&self.app)); |
| 1136 | if let Err(why) = launcher::down(&binary, &native_dir(&self.app, Backend::Sandboxed), Down::Delete) { |
| 1137 | self.write_log(&format!("g1t: the sandbox couldn't be removed: {why}")); |
| 1138 | } |
| 1139 | SandboxRegistration::clear(&self.app); |
| 1140 | if Registration::load(&self.app).is_some_and(|r| r.backend == Backend::Sandboxed) { |
| 1141 | Registration::clear(&self.app); |
| 1142 | } |
| 1143 | } |
| 1144 | |
| 1145 | /// Whether what a native runner keeps of a registration is still here. |
| 1146 | fn kept(&self, backend: Backend) -> bool { |
| 1147 | native_dir(&self.app, backend).join("config.json").exists() |
| 1148 | } |
| 1149 | |
| 1150 | /// Asks the site for a registration token for `workspace`. |
| 1151 | fn want(&mut self, workspace: &str) { |
| 1152 | *self.app.state::<Shared>().wanted.lock().unwrap() = Some(workspace.to_owned()); |
| 1153 | let _ = self.app.emit_to(EventTarget::webview(frame::SITE), WANTED_EVENT, workspace.to_owned()); |
| 1154 | self.not_before = Instant::now() + ASK_AGAIN; |
| 1155 | let signed_out = self.app.state::<Shared>().site.lock().unwrap().signed_in == Some(false); |
| 1156 | self.set(if signed_out { Code::SignedOut } else { Code::Registering }, Extra::default()); |
| 1157 | } |
| 1158 | |
| 1159 | /// The thread's look round: what should run, and whether it does. |
| 1160 | fn look(&mut self) { |
| 1161 | let settings = config::settings(&self.app); |
| 1162 | let paused = *self.app.state::<Shared>().paused.lock().unwrap(); |
| 1163 | if !settings.runner || paused { |
| 1164 | if self.run.is_some() { |
| 1165 | self.halt(); |
| 1166 | } |
| 1167 | *self.app.state::<Shared>().wanted.lock().unwrap() = None; |
| 1168 | self.set(if paused { Code::Stopped } else { Code::Off }, Extra::default()); |
| 1169 | return; |
| 1170 | } |
| 1171 | // Looking needs a runner here to compare with; the first comes with the first start. |
| 1172 | if Instant::now() >= self.next_update && native::installed(&folder(&self.app)).is_some() { |
| 1173 | self.check(false); |
| 1174 | } |
| 1175 | let workspace = workspace(&self.app); |
| 1176 | if let Some(run) = &self.run { |
| 1177 | let moved = run.backend != settings.runner_backend || Some(&run.workspace) != workspace.as_ref(); |
| 1178 | let back = run.offline_at.is_some_and(|at| at.elapsed() >= BACK_ONLINE); |
| 1179 | let (busy, stepping) = (run.busy, run.step().is_some()); |
| 1180 | if back { |
| 1181 | if let Some(run) = self.run.as_mut() { |
| 1182 | run.offline_at = None; |
| 1183 | } |
| 1184 | self.set(Code::Idle, Extra::default()); |
| 1185 | } |
| 1186 | // A change of backend or workspace waits for the job to end, as an update does. |
| 1187 | if moved && !busy && !stepping { |
| 1188 | self.halt_as(Down::Drain); |
| 1189 | self.not_before = Instant::now(); |
| 1190 | } else if !moved && !stepping && !self.install() { |
| 1191 | self.kinds(workspace.as_deref()); |
| 1192 | } |
| 1193 | self.poll(); |
| 1194 | return; |
| 1195 | } |
| 1196 | if Instant::now() < self.not_before { |
| 1197 | return; |
| 1198 | } |
| 1199 | // Nothing runs: a waiting program goes in before anything starts. |
| 1200 | if self.program.is_some() |
| 1201 | && self.allowed() |
| 1202 | && let Some((version, bytes)) = self.program.take() |
| 1203 | { |
| 1204 | self.put(version, &bytes); |
| 1205 | } |
| 1206 | self.start(settings.runner_backend, workspace); |
| 1207 | } |
| 1208 | |
| 1209 | fn start(&mut self, backend: Backend, workspace: Option<String>) { |
| 1210 | let app = self.app.clone(); |
| 1211 | let engine = match backend::engine(backend) { |
| 1212 | Ok(engine) => engine, |
| 1213 | Err(missing) => { |
| 1214 | self.not_before = Instant::now() + LOOK_AGAIN; |
| 1215 | let code = match missing { |
| 1216 | Missing::Docker => Code::NoDocker, |
| 1217 | }; |
| 1218 | self.set(code, Extra::default()); |
| 1219 | return; |
| 1220 | } |
| 1221 | }; |
| 1222 | let Some(workspace) = workspace else { |
| 1223 | self.not_before = Instant::now() + LOOK_AGAIN; |
| 1224 | let code = match app.state::<Shared>().site.lock().unwrap().signed_in { |
| 1225 | Some(false) => Code::SignedOut, |
| 1226 | Some(true) => Code::NoWorkspace, |
| 1227 | None => Code::Starting, |
| 1228 | }; |
| 1229 | self.set(code, detail("Waiting for g1t to open.")); |
| 1230 | return; |
| 1231 | }; |
| 1232 | if engine == Engine::Launcher { |
| 1233 | return self.start_sandbox(workspace); |
| 1234 | } |
| 1235 | let server = config::server(&app); |
| 1236 | let registration = Registration::load(&app); |
| 1237 | let ours = registration.as_ref().is_some_and(|r| r.server == server.as_str() && r.workspace == workspace && r.backend == backend) && self.kept(backend); |
| 1238 | let mut token = None; |
| 1239 | if !ours { |
| 1240 | if registration.is_some() { |
| 1241 | self.set(Code::Starting, detail("Removing this computer's earlier registration.")); |
| 1242 | self.forget(); |
| 1243 | } |
| 1244 | match self.token.take() { |
| 1245 | Some((for_workspace, given)) if for_workspace == workspace => token = Some(given), |
| 1246 | _ => return self.want(&workspace), |
| 1247 | } |
| 1248 | } |
| 1249 | self.hold_workspace(&workspace); |
| 1250 | self.set(Code::Starting, Extra::default()); |
| 1251 | let envs = environment(&server, backend, token.as_deref()); |
| 1252 | let generation = self.generation + 1; |
| 1253 | self.generation = generation; |
| 1254 | let binary = match self.ensure_binary(&server) { |
| 1255 | Ok(binary) => binary, |
| 1256 | Err(fault) => return self.not_installed(workspace, token, fault), |
| 1257 | }; |
| 1258 | let started = native::spawn(&binary, &native_dir(&app, backend), &envs).map_err(|e| format!("g1t-runner could not start: {e}")).inspect(|child| { |
| 1259 | let _ = std::fs::write(folder(&app).join("pid"), child.id().to_string()); |
| 1260 | }); |
| 1261 | match started { |
| 1262 | Ok(mut child) => { |
| 1263 | self.follow(generation, &mut child); |
| 1264 | self.run = Some(Run::new(generation, engine, backend, workspace, Process::Child(child), token.is_some())); |
| 1265 | self.set(Code::Starting, Extra::default()); |
| 1266 | } |
| 1267 | Err(why) => self.not_started(workspace, token, why), |
| 1268 | } |
| 1269 | } |
| 1270 | |
| 1271 | /// The runner couldn't be started: the token is kept for the next try, |
| 1272 | /// while it lasts, and it is tried again in a while. |
| 1273 | fn not_started(&mut self, workspace: String, token: Option<String>, why: String) { |
| 1274 | if let Some(given) = token { |
| 1275 | self.token = Some((workspace, given)); |
| 1276 | } |
| 1277 | self.failures += 1; |
| 1278 | self.not_before = Instant::now() + LOOK_AGAIN; |
| 1279 | let offline = why.contains("error sending request") || why.contains("dns") || why.contains("connect"); |
| 1280 | self.write_log(&format!("g1t: the runner could not start: {why}")); |
| 1281 | if offline { |
| 1282 | self.set(Code::Offline, Extra::default()); |
| 1283 | } else { |
| 1284 | self.set(Code::Crashed, Extra { detail: Some(why), wait_seconds: Some(LOOK_AGAIN.as_secs()), ..Extra::default() }); |
| 1285 | } |
| 1286 | } |
| 1287 | |
| 1288 | /// The setting names the workspace from now on, so browsing another one never moves the runner. |
| 1289 | fn hold_workspace(&self, workspace: &str) { |
| 1290 | if config::settings(&self.app).runner_workspace.as_deref() != Some(workspace) { |
| 1291 | let mut settings = config::settings(&self.app); |
| 1292 | settings.runner_workspace = Some(workspace.to_owned()); |
| 1293 | let _ = config::set_settings(&self.app, settings); |
| 1294 | } |
| 1295 | } |
| 1296 | |
| 1297 | /// Reads the child's output, a line at a time, onto the thread's channel. |
| 1298 | fn follow(&self, generation: u64, child: &mut Child) { |
| 1299 | fn pipe(tx: Sender<Ask>, generation: u64, from: impl Read + Send + 'static, last: bool) { |
| 1300 | std::thread::spawn(move || { |
| 1301 | for line in BufReader::new(from).lines().map_while(Result::ok) { |
| 1302 | if tx.send(Ask::Line(generation, line, last)).is_err() { |
| 1303 | return; |
| 1304 | } |
| 1305 | } |
| 1306 | if last { |
| 1307 | let _ = tx.send(Ask::Exited(generation)); |
| 1308 | } |
| 1309 | }); |
| 1310 | } |
| 1311 | if let Some(err) = child.stderr.take() { |
| 1312 | pipe(self.tx.clone(), generation, err, false); |
| 1313 | } |
| 1314 | if let Some(out) = child.stdout.take() { |
| 1315 | pipe(self.tx.clone(), generation, out, true); |
| 1316 | } |
| 1317 | } |
| 1318 | |
| 1319 | /// `g1t-runner` installed, from the newest release when it is not: |
| 1320 | /// one g1t signed, as every update is. |
| 1321 | fn ensure_binary(&mut self, server: &Url) -> Result<PathBuf, Fault> { |
| 1322 | let folder = folder(&self.app); |
| 1323 | let binary = native::binary(&folder); |
| 1324 | if native::installed(&folder).is_some() { |
| 1325 | return Ok(binary); |
| 1326 | } |
| 1327 | self.set(Code::Updating, detail("Downloading g1t-runner.")); |
| 1328 | let manifest = native::latest(server)?; |
| 1329 | let bytes = native::download(server, &manifest).map_err(Fault::Failed)?; |
| 1330 | native::put(&folder, &bytes, &manifest.version).map_err(Fault::Failed)?; |
| 1331 | self.note(None, Some(manifest.version)); |
| 1332 | self.with_updates(|updates| updates.checked_at = Some(SystemTime::now())); |
| 1333 | Ok(binary) |
| 1334 | } |
| 1335 | |
| 1336 | /// There is no `g1t-runner` here, and none could be had: a release g1t |
| 1337 | /// didn't sign is refused, and asked for again in a while. |
| 1338 | fn not_installed(&mut self, workspace: String, token: Option<String>, fault: Fault) { |
| 1339 | match fault { |
| 1340 | Fault::Failed(why) => self.not_started(workspace, token, why), |
| 1341 | Fault::Unsigned(why) => { |
| 1342 | if let Some(given) = token { |
| 1343 | self.token = Some((workspace, given)); |
| 1344 | } |
| 1345 | self.write_log(&format!("g1t: g1t-runner wasn't installed: {why}")); |
| 1346 | self.not_before = Instant::now() + MOST_BACKOFF; |
| 1347 | self.set(Code::UpdateRefused, detail(why)); |
| 1348 | } |
| 1349 | } |
| 1350 | } |
| 1351 | |
| 1352 | // --- Updates ------------------------------------------------------------------ |
| 1353 | |
| 1354 | /// Looks for a newer release on a thread of its own, so the runner is |
| 1355 | /// looked after meanwhile: its signed manifest, and its program when it |
| 1356 | /// is newer than the one here. `asked` is the person asking, which |
| 1357 | /// works with "Download updates by themselves" off. |
| 1358 | fn check(&mut self, asked: bool) { |
| 1359 | self.next_update = Instant::now() + UPDATE_EVERY; |
| 1360 | if !asked && !config::settings(&self.app).auto_update { |
| 1361 | return; |
| 1362 | } |
| 1363 | if self.with_updates(|updates| std::mem::replace(&mut updates.checking, true)) { |
| 1364 | return; |
| 1365 | } |
| 1366 | self.asked |= asked; |
| 1367 | publish(&self.app); |
| 1368 | let (server, folder) = (config::server(&self.app), folder(&self.app)); |
| 1369 | let waiting = self.program.as_ref().map(|(version, _)| version.clone()); |
| 1370 | let tx = self.tx.clone(); |
| 1371 | let spawned = std::thread::Builder::new().name("g1t-runner-releases".into()).spawn(move || { |
| 1372 | let _ = tx.send(Ask::Checked(look_for_release(&server, &folder, waiting.as_deref()))); |
| 1373 | }); |
| 1374 | if spawned.is_err() { |
| 1375 | self.with_updates(|updates| updates.checking = false); |
| 1376 | } |
| 1377 | } |
| 1378 | |
| 1379 | /// What the look found: a newer program waits for its kinds of job to |
| 1380 | /// be idle, and so does the sandbox's image when it is older than the |
| 1381 | /// release's. A release g1t didn't sign, or one that didn't download, |
| 1382 | /// leaves the runner as it is, and is said while it is idle. |
| 1383 | fn checked(&mut self, found: Result<Found, Fault>) { |
| 1384 | self.with_updates(|updates| updates.checking = false); |
| 1385 | let found = match found { |
| 1386 | Ok(found) => found, |
| 1387 | Err(fault) => { |
| 1388 | let failed = match fault { |
| 1389 | Fault::Unsigned(why) => (Code::UpdateRefused, why), |
| 1390 | Fault::Failed(why) => (Code::UpdateFailed, format!("g1t-runner couldn't be updated: {why}")), |
| 1391 | }; |
| 1392 | self.write_log(&format!("g1t: {}", failed.1)); |
| 1393 | self.update_failed = Some(failed); |
| 1394 | self.asked = false; |
| 1395 | if matches!(self.code(), Code::Idle | Code::UpdateFailed | Code::UpdateRefused) { |
| 1396 | let (code, extra) = self.idle(Code::Idle); |
| 1397 | self.set(code, extra); |
| 1398 | } |
| 1399 | return publish(&self.app); |
| 1400 | } |
| 1401 | }; |
| 1402 | if let Some(bytes) = found.program { |
| 1403 | self.program = Some((found.version.clone(), bytes)); |
| 1404 | } |
| 1405 | if !found.newer { |
| 1406 | self.program = None; |
| 1407 | } |
| 1408 | // The sandbox's image is the release's version; one older takes the newest. |
| 1409 | let sandbox = self.app.state::<Shared>().now.lock().unwrap().sandbox.clone(); |
| 1410 | let launcher = self.run.as_ref().is_some_and(|run| run.engine == Engine::Launcher); |
| 1411 | let image = launcher |
| 1412 | && found.images |
| 1413 | && sandbox.is_some_and(|status| status.linux().state != State::Absent && status.image_version.as_deref().is_some_and(|have| native::newer(&found.version, have))); |
| 1414 | let program = self.program.is_some(); |
| 1415 | self.with_updates(|updates| { |
| 1416 | updates.checked_at = Some(SystemTime::now()); |
| 1417 | updates.program = program; |
| 1418 | updates.image = image; |
| 1419 | updates.ready = (program || image).then(|| found.version.clone()); |
| 1420 | }); |
| 1421 | if program || image { |
| 1422 | self.write_log(&format!("g1t: g1t-runner {} is ready, and installs as each kind of job is idle.", found.version)); |
| 1423 | } else { |
| 1424 | self.asked = false; |
| 1425 | } |
| 1426 | // It worked: what the last failure said goes. |
| 1427 | if self.update_failed.take().is_some() && matches!(self.code(), Code::UpdateFailed | Code::UpdateRefused) { |
| 1428 | self.set(Code::Idle, Extra::default()); |
| 1429 | } |
| 1430 | publish(&self.app); |
| 1431 | } |
| 1432 | |
| 1433 | /// Whether a waiting update may go in: "Download updates by |
| 1434 | /// themselves" is on, or the person asked for it. Turned off since it |
| 1435 | /// downloaded, it no longer waits. |
| 1436 | fn allowed(&mut self) -> bool { |
| 1437 | if self.asked || config::settings(&self.app).auto_update { |
| 1438 | return true; |
| 1439 | } |
| 1440 | self.program = None; |
| 1441 | self.with_updates(|updates| { |
| 1442 | updates.program = false; |
| 1443 | updates.image = false; |
| 1444 | updates.ready = None; |
| 1445 | }); |
| 1446 | false |
| 1447 | } |
| 1448 | |
| 1449 | /// Puts a waiting update in, kind by kind, as each kind of job is idle: |
| 1450 | /// true when it started something, which is waited for. Never in the |
| 1451 | /// middle of a job. |
| 1452 | fn install(&mut self) -> bool { |
| 1453 | let image = self.with_updates(|updates| updates.image); |
| 1454 | if (self.program.is_none() && !image) || !self.allowed() { |
| 1455 | return false; |
| 1456 | } |
| 1457 | let Some((engine, busy, listening)) = self.run.as_ref().map(|run| (run.engine, run.busy, run.listening)) else { return false }; |
| 1458 | if engine != Engine::Launcher { |
| 1459 | // A runner in Docker or on this computer is the program: stopped between jobs, started on the new one. |
| 1460 | if busy || !listening { |
| 1461 | return false; |
| 1462 | } |
| 1463 | let Some((version, bytes)) = self.program.take() else { return false }; |
| 1464 | self.set(Code::Updating, detail(format!("Updating g1t-runner to {version}."))); |
| 1465 | self.halt(); |
| 1466 | self.put(version, &bytes); |
| 1467 | self.not_before = Instant::now(); |
| 1468 | return true; |
| 1469 | } |
| 1470 | let Some(status) = self.app.state::<Shared>().now.lock().unwrap().sandbox.clone() else { return false }; |
| 1471 | if self.program.is_some() { |
| 1472 | let running = native_kinds(&status); |
| 1473 | if running.is_empty() { |
| 1474 | // Nothing runs on the program: it is replaced now, and |
| 1475 | // kinds() starts what was drained on the new one first, |
| 1476 | // before the image, which takes minutes, is updated. |
| 1477 | let (version, bytes) = self.program.take().expect("checked above"); |
| 1478 | self.put(version, &bytes); |
| 1479 | return false; |
| 1480 | } else if !status.per_kind() { |
| 1481 | // A launcher from before kinds of job stopped on their own: the whole sandbox stops between jobs. |
| 1482 | if busy { |
| 1483 | return false; |
| 1484 | } |
| 1485 | let (version, bytes) = self.program.take().expect("checked above"); |
| 1486 | self.set(Code::Updating, detail(format!("Updating g1t-runner to {version}."))); |
| 1487 | self.halt_as(Down::Drain); |
| 1488 | self.put(version, &bytes); |
| 1489 | self.with_updates(|updates| updates.image = true); |
| 1490 | self.not_before = Instant::now(); |
| 1491 | return true; |
| 1492 | } else if let Some(idle) = running.iter().find(|kind| kind.job.is_none()) { |
| 1493 | // Stopped after any job it takes meanwhile; replaced once it has. |
| 1494 | let kind = idle.id.clone(); |
| 1495 | self.with_updates(|updates| updates.installing = vec![kind.clone()]); |
| 1496 | self.kind_step(Step::Drain(kind), None); |
| 1497 | return true; |
| 1498 | } |
| 1499 | } |
| 1500 | // Linux jobs take the image once they have no job (`update` drains them itself too). |
| 1501 | let image = self.with_updates(|updates| updates.image); |
| 1502 | if image && status.linux().job.is_none() { |
| 1503 | self.with_updates(|updates| updates.installing.push("linux".into())); |
| 1504 | self.sandbox_step(Step::Update); |
| 1505 | return true; |
| 1506 | } |
| 1507 | false |
| 1508 | } |
| 1509 | |
| 1510 | /// Puts a downloaded program in place, with nothing running on it. |
| 1511 | fn put(&mut self, version: String, bytes: &[u8]) { |
| 1512 | match native::put(&folder(&self.app), bytes, &version) { |
| 1513 | Ok(()) => { |
| 1514 | self.write_log(&format!("g1t: g1t-runner {version} is in place.")); |
| 1515 | if self.run.as_ref().is_none_or(|run| run.engine != Engine::Launcher) { |
| 1516 | self.note(None, Some(version)); |
| 1517 | } |
| 1518 | } |
| 1519 | Err(why) => { |
| 1520 | let why = format!("g1t-runner {version} couldn't be put in place: {why}"); |
| 1521 | self.write_log(&format!("g1t: {why}")); |
| 1522 | self.update_failed = Some((Code::UpdateFailed, why)); |
| 1523 | } |
| 1524 | } |
| 1525 | self.done_with(|updates| { |
| 1526 | updates.program = false; |
| 1527 | updates.installing.retain(|kind| kind == "linux"); |
| 1528 | }); |
| 1529 | } |
| 1530 | |
| 1531 | /// Part of an update is done: what waits is said again. |
| 1532 | fn done_with(&mut self, change: impl FnOnce(&mut Updates)) { |
| 1533 | let left = self.with_updates(|updates| { |
| 1534 | change(updates); |
| 1535 | if !updates.program && !updates.image { |
| 1536 | updates.ready = None; |
| 1537 | } |
| 1538 | updates.program || updates.image |
| 1539 | }); |
| 1540 | if !left { |
| 1541 | self.asked = false; |
| 1542 | } |
| 1543 | publish(&self.app); |
| 1544 | } |
| 1545 | |
| 1546 | // --- The sandbox (launcher.rs) ------------------------------------------------ |
| 1547 | |
| 1548 | /// Starts the sandbox for `workspace`: registered with it the first |
| 1549 | /// time, with a token from the site; from its own registration after. |
| 1550 | fn start_sandbox(&mut self, workspace: String) { |
| 1551 | let app = self.app.clone(); |
| 1552 | let server = config::server(&app); |
| 1553 | let binary = match self.ensure_binary(&server) { |
| 1554 | Ok(binary) => binary, |
| 1555 | Err(fault) => return self.not_installed(workspace, None, fault), |
| 1556 | }; |
| 1557 | let dir = native_dir(&app, Backend::Sandboxed); |
| 1558 | let mut status = match launcher::status(&binary, &dir) { |
| 1559 | Ok(status) => status, |
| 1560 | Err(why) => return self.no_status(why), |
| 1561 | }; |
| 1562 | self.seen(&status); |
| 1563 | // Nothing here can be sandboxed: why, and how to fix it; looked at again in a while. |
| 1564 | if !status.anything_available() { |
| 1565 | self.not_before = Instant::now() + LOOK_AGAIN; |
| 1566 | let (code, extra) = shown(&status); |
| 1567 | return self.set(code, extra); |
| 1568 | } |
| 1569 | // Another backend's registration goes; the sandbox's own is kept for its workspace. |
| 1570 | if Registration::load(&app).is_some_and(|r| r.backend != Backend::Sandboxed) { |
| 1571 | self.set(Code::Starting, detail("Removing this computer's earlier registration.")); |
| 1572 | self.forget(); |
| 1573 | } |
| 1574 | let kept = SandboxRegistration::load(&app); |
| 1575 | let ours = kept.as_ref().is_some_and(|k| k.server == server.as_str() && k.workspace == workspace); |
| 1576 | let made = status.capabilities.iter().any(|c| c.state != State::Absent) || status.state != State::Absent; |
| 1577 | if !ours && made && kept.is_none() { |
| 1578 | // Made by hand (`g1t-runner sandbox up`), and there is one per computer: it is the person's to remove. |
| 1579 | self.not_before = Instant::now() + LOOK_AGAIN; |
| 1580 | return self.set(Code::SandboxTaken, Extra::default()); |
| 1581 | } |
| 1582 | if !ours && made { |
| 1583 | // The app's sandbox, registered with another workspace or server. |
| 1584 | self.set(Code::Starting, detail("Removing this computer's sandbox for another workspace.")); |
| 1585 | self.forget_sandbox(); |
| 1586 | status = match launcher::status(&binary, &dir) { |
| 1587 | Ok(status) => status, |
| 1588 | Err(why) => return self.no_status(why), |
| 1589 | }; |
| 1590 | } |
| 1591 | // The kinds of job the person stopped are stopped here too, before |
| 1592 | // anything starts (the launcher keeps them off); with every kind |
| 1593 | // stopped, nothing starts. |
| 1594 | let kinds = config::settings(&app).runner_capabilities; |
| 1595 | if status.per_kind() { |
| 1596 | let stopping: Vec<String> = status.capabilities.iter().filter(|c| !kinds.on(&c.id) && !c.off()).map(|c| c.id.clone()).collect(); |
| 1597 | for kind in &stopping { |
| 1598 | if let Err(why) = launcher::down_only(&binary, &dir, kind) { |
| 1599 | self.write_log(&format!("g1t: {} couldn't be stopped: {why}", diagnosis::kind_label(kind))); |
| 1600 | } |
| 1601 | } |
| 1602 | if !stopping.is_empty() { |
| 1603 | status = match launcher::status(&binary, &dir) { |
| 1604 | Ok(status) => status, |
| 1605 | Err(why) => return self.no_status(why), |
| 1606 | }; |
| 1607 | self.seen(&status); |
| 1608 | } |
| 1609 | } |
| 1610 | let wanted: Vec<String> = status.capabilities.iter().filter(|c| c.available && kinds.on(&c.id)).map(|c| c.id.clone()).collect(); |
| 1611 | if status.capabilities.iter().any(|c| c.available) && wanted.is_empty() { |
| 1612 | self.not_before = Instant::now() + LOOK_AGAIN; |
| 1613 | return self.set(Code::KindsStopped, Extra::default()); |
| 1614 | } |
| 1615 | let mut token = None; |
| 1616 | if !ours || !kept.as_ref().is_some_and(|k| k.registered) || status.unregistered() { |
| 1617 | match self.token.take() { |
| 1618 | Some((for_workspace, given)) if for_workspace == workspace => token = Some(given), |
| 1619 | _ => return self.want(&workspace), |
| 1620 | } |
| 1621 | SandboxRegistration { server: server.to_string(), workspace: workspace.clone(), registered: false }.save(&app); |
| 1622 | } |
| 1623 | self.hold_workspace(&workspace); |
| 1624 | let generation = self.generation + 1; |
| 1625 | self.generation = generation; |
| 1626 | self.run = Some(Run::new(generation, Engine::Launcher, Backend::Sandboxed, workspace.clone(), Process::Sandbox(None), token.is_some())); |
| 1627 | // A new release's image first, then the sandbox starts again. |
| 1628 | let image = self.with_updates(|updates| std::mem::take(&mut updates.image)); |
| 1629 | let step = if image && status.linux().state != State::Absent { Step::Update } else { Step::Up }; |
| 1630 | if step == Step::Update { |
| 1631 | self.with_updates(|updates| updates.installing.push("linux".into())); |
| 1632 | } else if image { |
| 1633 | self.done_with(|_| {}); |
| 1634 | } |
| 1635 | let name = backend::machine_name(); |
| 1636 | let url = server.as_str().trim_end_matches('/').to_owned(); |
| 1637 | let up = launcher::Up { url: &url, name: &name, labels: "desktop", token: token.as_deref(), only: wanted.iter().map(String::as_str).collect() }; |
| 1638 | let (args, starting) = match step { |
| 1639 | Step::Update => (launcher::update_args(&dir), vec![]), |
| 1640 | _ => (launcher::up_args(&dir, &up), wanted.clone()), |
| 1641 | }; |
| 1642 | if let Err(why) = self.spawn_step(step, args, up.token) { |
| 1643 | self.run = None; |
| 1644 | return self.not_started(workspace, token, why); |
| 1645 | } |
| 1646 | self.now_starting(starting); |
| 1647 | self.set(Code::Starting, detail(STARTING_SANDBOX)); |
| 1648 | } |
| 1649 | |
| 1650 | /// Says which kinds of job a launcher command is starting now. |
| 1651 | fn now_starting(&self, kinds: Vec<String>) { |
| 1652 | let changed = { |
| 1653 | let shared = self.app.state::<Shared>(); |
| 1654 | let mut now = shared.now.lock().unwrap(); |
| 1655 | let changed = now.starting != kinds; |
| 1656 | now.starting = kinds; |
| 1657 | changed |
| 1658 | }; |
| 1659 | if changed { |
| 1660 | publish(&self.app); |
| 1661 | } |
| 1662 | } |
| 1663 | |
| 1664 | /// Each kind of job as the person has it: one they stopped is stopped |
| 1665 | /// (`down --only`), one they started is started (`up --only`), and one |
| 1666 | /// that should run and doesn't while another runs is tried again in a |
| 1667 | /// while: every five minutes while the release has no sandbox images. |
| 1668 | /// One launcher command at a time, each leaving the others as they are. |
| 1669 | fn kinds(&mut self, workspace: Option<&str>) { |
| 1670 | if !self.run.as_ref().is_some_and(|run| run.engine == Engine::Launcher) { |
| 1671 | return; |
| 1672 | } |
| 1673 | let Some(status) = self.app.state::<Shared>().now.lock().unwrap().sandbox.clone() else { return }; |
| 1674 | if !status.per_kind() { |
| 1675 | return; |
| 1676 | } |
| 1677 | let on = config::settings(&self.app).runner_capabilities; |
| 1678 | // A kind of job drained for a new program starts on the new one, once it is in. |
| 1679 | let replacing = self.program.is_some(); |
| 1680 | for capability in &status.capabilities { |
| 1681 | if !on.on(&capability.id) && !capability.off() { |
| 1682 | return self.kind_step(Step::Stop(capability.id.clone()), None); |
| 1683 | } |
| 1684 | if on.on(&capability.id) && capability.off() && capability.available && !(replacing && capability.id != "linux") { |
| 1685 | return self.start_kind(capability, workspace); |
| 1686 | } |
| 1687 | } |
| 1688 | // Nothing runs: the sandbox as a whole starts again (sandbox_stopped). |
| 1689 | if !status.running() { |
| 1690 | self.retry.clear(); |
| 1691 | return; |
| 1692 | } |
| 1693 | let now = Instant::now(); |
| 1694 | for capability in &status.capabilities { |
| 1695 | let id = capability.id.as_str(); |
| 1696 | let wanting = on.on(id) && !capability.off() && capability.available && matches!(capability.state, State::Error | State::Stopped | State::Absent); |
| 1697 | if !wanting { |
| 1698 | self.retry.retain(|(kind, _, _)| kind != id); |
| 1699 | continue; |
| 1700 | } |
| 1701 | let images = id == "linux" && status.trouble().is_some_and(|(cause, _)| launcher::why(&cause) == launcher::Why::NoImages); |
| 1702 | match self.retry.iter().position(|(kind, _, _)| kind == id) { |
| 1703 | // Just seen not running: tried again in a while. |
| 1704 | None => self.retry.push((id.to_owned(), now + again(images, 1), 1)), |
| 1705 | Some(at) if now >= self.retry[at].1 => { |
| 1706 | let tries = self.retry[at].2 + 1; |
| 1707 | self.retry[at] = (id.to_owned(), now + again(images, tries), tries); |
| 1708 | self.write_log(&format!("g1t: trying {} again.", diagnosis::kind_label(id))); |
| 1709 | return self.start_kind(capability, workspace); |
| 1710 | } |
| 1711 | Some(_) => {} |
| 1712 | } |
| 1713 | } |
| 1714 | } |
| 1715 | |
| 1716 | /// Starts one kind of job; one never registered needs a token from the |
| 1717 | /// site first, which is asked for, and it starts once it is here. |
| 1718 | fn start_kind(&mut self, capability: &launcher::Capability, workspace: Option<&str>) { |
| 1719 | let mut token = None; |
| 1720 | if capability.state == State::Absent { |
| 1721 | let Some(workspace) = workspace else { return }; |
| 1722 | match self.token.take() { |
| 1723 | Some((for_workspace, given)) if for_workspace == workspace => token = Some(given), |
| 1724 | other => { |
| 1725 | self.token = other; |
| 1726 | if wanted(&self.app).is_none() { |
| 1727 | *self.app.state::<Shared>().wanted.lock().unwrap() = Some(workspace.to_owned()); |
| 1728 | let _ = self.app.emit_to(EventTarget::webview(frame::SITE), WANTED_EVENT, workspace.to_owned()); |
| 1729 | } |
| 1730 | return; |
| 1731 | } |
| 1732 | } |
| 1733 | } |
| 1734 | self.kind_step(Step::Start(capability.id.clone()), token); |
| 1735 | } |
| 1736 | |
| 1737 | /// `up --only` or `down --only` one kind of job, while the rest run on. |
| 1738 | fn kind_step(&mut self, step: Step, token: Option<String>) { |
| 1739 | let dir = native_dir(&self.app, Backend::Sandboxed); |
| 1740 | let (args, starting, said) = match &step { |
| 1741 | Step::Start(kind) => { |
| 1742 | let url = config::server(&self.app).as_str().trim_end_matches('/').to_owned(); |
| 1743 | let name = backend::machine_name(); |
| 1744 | let up = launcher::Up { url: &url, name: &name, labels: "desktop", token: token.as_deref(), only: vec![kind.as_str()] }; |
| 1745 | (launcher::up_args(&dir, &up), vec![kind.clone()], format!("starting {}", diagnosis::kind_label(kind))) |
| 1746 | } |
| 1747 | Step::Stop(kind) => (launcher::down_only_args(&dir, kind), vec![], format!("stopping {}", diagnosis::kind_label(kind))), |
| 1748 | Step::Drain(kind) => (launcher::drain_only_args(&dir, kind), vec![], format!("stopping {} after their job, to update g1t-runner", diagnosis::kind_label(kind))), |
| 1749 | Step::Up | Step::Update => return, |
| 1750 | }; |
| 1751 | self.write_log(&format!("g1t: {said}.")); |
| 1752 | match self.spawn_step(step, args, token.as_deref()) { |
| 1753 | Ok(()) => self.now_starting(starting), |
| 1754 | Err(why) => self.write_log(&format!("g1t: {why}")), |
| 1755 | } |
| 1756 | } |
| 1757 | |
| 1758 | /// `sandbox status` didn't answer: a `g1t-runner` from before its |
| 1759 | /// launcher, which a newer release replaces, or one that failed. |
| 1760 | fn no_status(&mut self, why: String) { |
| 1761 | self.write_log(&format!("g1t: the sandbox's status: {why}")); |
| 1762 | self.failures += 1; |
| 1763 | if why.contains("older than its sandbox launcher") { |
| 1764 | let (server, folder) = (config::server(&self.app), folder(&self.app)); |
| 1765 | if let Ok(Found { version, program: Some(bytes), .. }) = look_for_release(&server, &folder, None) |
| 1766 | && native::put(&folder, &bytes, &version).is_ok() |
| 1767 | { |
| 1768 | self.note(None, Some(version)); |
| 1769 | self.not_before = Instant::now(); |
| 1770 | return; |
| 1771 | } |
| 1772 | self.not_before = Instant::now() + MOST_BACKOFF; |
| 1773 | return self.set(Code::NoLauncher, detail(why)); |
| 1774 | } |
| 1775 | self.not_before = Instant::now() + LOOK_AGAIN; |
| 1776 | self.set(Code::SandboxFailed, Extra { detail: Some(why), wait_seconds: Some(LOOK_AGAIN.as_secs()), ..Extra::default() }); |
| 1777 | } |
| 1778 | |
| 1779 | /// Runs a launcher command for the sandbox's run, reading its output. |
| 1780 | fn spawn_step(&mut self, step: Step, args: Vec<std::ffi::OsString>, token: Option<&str>) -> Result<(), String> { |
| 1781 | let binary = native::binary(&folder(&self.app)); |
| 1782 | let mut command = launcher::command(&binary, args); |
| 1783 | launcher::with_token(&mut command, token); |
| 1784 | let mut child = command.spawn().map_err(|e| format!("g1t-runner sandbox {} could not start: {e}", step.name()))?; |
| 1785 | let _ = std::fs::write(folder(&self.app).join("pid"), child.id().to_string()); |
| 1786 | let Some(generation) = self.run.as_ref().map(|run| run.generation) else { |
| 1787 | native::stop(&mut child); |
| 1788 | return Ok(()); |
| 1789 | }; |
| 1790 | self.follow(generation, &mut child); |
| 1791 | let run = self.run.as_mut().expect("checked above"); |
| 1792 | run.json = None; |
| 1793 | run.process = Process::Sandbox(Some((step, child))); |
| 1794 | Ok(()) |
| 1795 | } |
| 1796 | |
| 1797 | /// `update` on the running sandbox, while it has no job. |
| 1798 | fn sandbox_step(&mut self, step: Step) { |
| 1799 | if self.run.as_ref().is_none_or(|run| run.step().is_some()) { |
| 1800 | return; |
| 1801 | } |
| 1802 | let dir = native_dir(&self.app, Backend::Sandboxed); |
| 1803 | if let Err(why) = self.spawn_step(step, launcher::update_args(&dir), None) { |
| 1804 | self.write_log(&format!("g1t: {why}")); |
| 1805 | } |
| 1806 | } |
| 1807 | |
| 1808 | /// Asks the sandbox how it is, every few seconds, on a thread of its own. |
| 1809 | fn poll(&mut self) { |
| 1810 | let Some(run) = self.run.as_mut().filter(|run| run.engine == Engine::Launcher) else { return }; |
| 1811 | if run.polling || run.polled_at.is_some_and(|at| at.elapsed() < launcher::STATUS_EVERY) { |
| 1812 | return; |
| 1813 | } |
| 1814 | run.polling = true; |
| 1815 | run.polled_at = Some(Instant::now()); |
| 1816 | let (tx, generation) = (self.tx.clone(), run.generation); |
| 1817 | let binary = native::binary(&folder(&self.app)); |
| 1818 | let dir = native_dir(&self.app, Backend::Sandboxed); |
| 1819 | let spawned = std::thread::Builder::new().name("g1t-sandbox-status".into()).spawn(move || { |
| 1820 | let _ = tx.send(Ask::Sandbox(generation, launcher::status(&binary, &dir).map(Box::new))); |
| 1821 | }); |
| 1822 | if spawned.is_err() { |
| 1823 | run.polling = false; |
| 1824 | } |
| 1825 | } |
| 1826 | |
| 1827 | /// What a poll said. |
| 1828 | fn polled(&mut self, generation: u64, status: Result<Box<Status>, String>) { |
| 1829 | let Some(run) = self.run.as_mut().filter(|run| run.generation == generation) else { return }; |
| 1830 | run.polling = false; |
| 1831 | if std::mem::take(&mut run.stale) { |
| 1832 | return; |
| 1833 | } |
| 1834 | match status { |
| 1835 | Ok(status) => self.sandbox_now(*status), |
| 1836 | // Once is nothing: the next poll asks again. |
| 1837 | Err(why) => self.write_log(&format!("g1t: the sandbox's status: {why}")), |
| 1838 | } |
| 1839 | } |
| 1840 | |
| 1841 | /// Keeps what a status says for the tray, settings and the site. |
| 1842 | fn seen(&mut self, status: &Status) { |
| 1843 | let shared = self.app.state::<Shared>(); |
| 1844 | let mut now = shared.now.lock().unwrap(); |
| 1845 | now.sandbox = Some(status.clone()); |
| 1846 | if let Some(version) = status.version() { |
| 1847 | now.version = Some(version); |
| 1848 | } |
| 1849 | now.job = status.job().map(launcher::Job::said); |
| 1850 | } |
| 1851 | |
| 1852 | /// The sandbox as it is now, while it is the run. |
| 1853 | fn sandbox_now(&mut self, status: Status) { |
| 1854 | let had = self.app.state::<Shared>().now.lock().unwrap().job.clone(); |
| 1855 | let Some(run) = self.run.as_mut() else { return }; |
| 1856 | let step = run.step(); |
| 1857 | let job = status.job().map(launcher::Job::said); |
| 1858 | let was = had.filter(|_| run.busy); |
| 1859 | run.busy = job.is_some(); |
| 1860 | run.listening = status.running(); |
| 1861 | let update_failed = self.update_failed.clone(); |
| 1862 | if was != job { |
| 1863 | if let Some(ended) = &was { |
| 1864 | self.write_log(&format!("g1t: {ended} ended.")); |
| 1865 | } |
| 1866 | if let Some(took) = &job { |
| 1867 | self.write_log(&format!("g1t: Took {took}.")); |
| 1868 | } |
| 1869 | } |
| 1870 | self.seen(&status); |
| 1871 | match &step { |
| 1872 | Some(Step::Up) => return self.set(Code::Starting, detail(STARTING_SANDBOX)), |
| 1873 | Some(Step::Update) if status.state == State::Updating => return self.set(Code::Updating, detail("Updating the sandbox to the newest image.")), |
| 1874 | _ => {} |
| 1875 | } |
| 1876 | let (code, extra) = shown(&status); |
| 1877 | // Every kind of job stopped by the person: nothing to start again. |
| 1878 | if step.is_none() && status.all_off() && !status.running() { |
| 1879 | self.run = None; |
| 1880 | self.generation += 1; |
| 1881 | self.retry.clear(); |
| 1882 | self.not_before = Instant::now() + LOOK_AGAIN; |
| 1883 | return self.set(Code::KindsStopped, Extra::default()); |
| 1884 | } |
| 1885 | if step.is_none() && !status.running() && !matches!(status.state, State::Starting | State::Updating) { |
| 1886 | return self.sandbox_stopped(&status, code, extra); |
| 1887 | } |
| 1888 | match (code, update_failed) { |
| 1889 | (Code::Idle, Some((failed, why))) => self.set(failed, detail(why)), |
| 1890 | (code, _) => self.set(code, extra), |
| 1891 | } |
| 1892 | } |
| 1893 | |
| 1894 | /// The sandbox stopped by itself, or never got going: started again, |
| 1895 | /// waiting longer each time, or registered again when that is why. |
| 1896 | fn sandbox_stopped(&mut self, status: &Status, code: Code, extra: Extra) { |
| 1897 | let error = status.error.clone().or_else(|| extra.detail.clone()).unwrap_or_default(); |
| 1898 | if self.registration_trouble(&error) { |
| 1899 | return; |
| 1900 | } |
| 1901 | self.run = None; |
| 1902 | self.generation += 1; |
| 1903 | self.failures += 1; |
| 1904 | // A release without the sandbox's images is looked at again every few minutes. |
| 1905 | let wait = if code == Code::WaitingForImage { launcher::IMAGES_AGAIN } else { again(false, self.failures) }; |
| 1906 | self.not_before = Instant::now() + wait; |
| 1907 | let (code, extra) = match code { |
| 1908 | Code::Idle | Code::Busy | Code::Starting | Code::Updating => (Code::SandboxFailed, Extra { detail: status.error.clone().or_else(|| some_text("The sandbox stopped by itself.")), ..Extra::default() }), |
| 1909 | code => (code, extra), |
| 1910 | }; |
| 1911 | self.set(code, Extra { wait_seconds: Some(wait.as_secs()), ..extra }); |
| 1912 | } |
| 1913 | |
| 1914 | /// When the sandbox's trouble is its registration: true when the app |
| 1915 | /// has seen to it, and the next start registers again. |
| 1916 | fn registration_trouble(&mut self, error: &str) -> bool { |
| 1917 | match launcher::why(error) { |
| 1918 | launcher::Why::TokenRefused | launcher::Why::Unregistered => { |
| 1919 | if let Some(mut kept) = SandboxRegistration::load(&self.app) { |
| 1920 | kept.registered = false; |
| 1921 | kept.save(&self.app); |
| 1922 | } |
| 1923 | self.halt(); |
| 1924 | self.not_before = Instant::now(); |
| 1925 | true |
| 1926 | } |
| 1927 | launcher::Why::Forgotten => { |
| 1928 | self.halt(); |
| 1929 | self.forget_sandbox(); |
| 1930 | self.not_before = Instant::now(); |
| 1931 | true |
| 1932 | } |
| 1933 | _ => false, |
| 1934 | } |
| 1935 | } |
| 1936 | |
| 1937 | /// A launcher command ended: what it printed last is the sandbox's status. |
| 1938 | fn launched(&mut self, generation: u64) { |
| 1939 | let Some(run) = self.run.as_mut().filter(|run| run.generation == generation) else { return }; |
| 1940 | let Process::Sandbox(slot) = &mut run.process else { return }; |
| 1941 | let Some((step, mut child)) = slot.take() else { return }; |
| 1942 | let worked = child.wait().is_ok_and(|status| status.success()); |
| 1943 | let _ = std::fs::remove_file(folder(&self.app).join("pid")); |
| 1944 | let printed = run.json.take().unwrap_or_default().join("\n"); |
| 1945 | run.stale = run.polling; |
| 1946 | let with_token = run.with_token; |
| 1947 | let status = launcher::parse(&printed); |
| 1948 | { |
| 1949 | let shared = self.app.state::<Shared>(); |
| 1950 | let mut now = shared.now.lock().unwrap(); |
| 1951 | now.starting.clear(); |
| 1952 | now.download = None; |
| 1953 | } |
| 1954 | let error = match (&status, worked) { |
| 1955 | (Ok(status), true) => status.error.clone().filter(|_| !status.running()), |
| 1956 | (Ok(status), false) => Some(status.error.clone().unwrap_or_else(|| format!("g1t-runner sandbox {} failed.", step.name()))), |
| 1957 | (Err(why), _) => Some(why.clone()), |
| 1958 | }; |
| 1959 | let status = status.ok(); |
| 1960 | let running = status.as_ref().is_some_and(Status::running); |
| 1961 | if let Some(error) = &error { |
| 1962 | self.write_log(&format!("g1t: sandbox {}: {error}", step.name())); |
| 1963 | } |
| 1964 | // Drained for a new program, which goes in on the next look; one |
| 1965 | // that didn't stop leaves the update for the next one, rather than |
| 1966 | // being asked again and again. |
| 1967 | if let (Step::Drain(_), Some(error)) = (&step, &error) { |
| 1968 | self.program = None; |
| 1969 | self.update_failed = Some((Code::UpdateFailed, format!("g1t-runner couldn't be updated: {error}"))); |
| 1970 | self.done_with(|updates| { |
| 1971 | updates.program = false; |
| 1972 | updates.installing.retain(|kind| kind == "linux"); |
| 1973 | }); |
| 1974 | } |
| 1975 | // One kind of job started or stopped: how the sandbox is now says |
| 1976 | // the rest, and one that didn't start is tried again in a while. |
| 1977 | if matches!(step, Step::Start(_) | Step::Stop(_) | Step::Drain(_)) { |
| 1978 | if let Some(status) = status { |
| 1979 | self.sandbox_now(status); |
| 1980 | } |
| 1981 | return; |
| 1982 | } |
| 1983 | // Registered: from now on it starts with no token. |
| 1984 | if step == Step::Up && with_token && (error.is_none() || running) { |
| 1985 | let app = self.app.clone(); |
| 1986 | if let Some(run) = self.run.as_mut() { |
| 1987 | run.with_token = false; |
| 1988 | let registration = Registration { server: config::server(&app).to_string(), workspace: run.workspace.clone(), backend: Backend::Sandboxed }; |
| 1989 | SandboxRegistration { server: registration.server.clone(), workspace: registration.workspace.clone(), registered: true }.save(&app); |
| 1990 | registration.save(&app); |
| 1991 | } |
| 1992 | } |
| 1993 | // The image is taken, or failed to be: the next look says whether another waits. |
| 1994 | if step == Step::Update { |
| 1995 | self.done_with(|updates| { |
| 1996 | updates.image = false; |
| 1997 | updates.installing.retain(|kind| kind != "linux"); |
| 1998 | }); |
| 1999 | } |
| 2000 | if let Some(error) = error.clone() { |
| 2001 | if self.registration_trouble(&error) { |
| 2002 | return; |
| 2003 | } |
| 2004 | if step == Step::Update { |
| 2005 | self.update_failed = Some((Code::UpdateFailed, format!("The sandbox's image couldn't be updated: {error}"))); |
| 2006 | } else if !running { |
| 2007 | // Nothing runs: tidied away, and tried again in a while; |
| 2008 | // a release without the sandbox's images every few minutes. |
| 2009 | self.halt(); |
| 2010 | self.failures += 1; |
| 2011 | let (code, cause, fix) = status |
| 2012 | .as_ref() |
| 2013 | .and_then(diagnosis::sandbox_trouble) |
| 2014 | .unwrap_or_else(|| (diagnosis::sandbox_code(launcher::why(&error)), error.clone(), None)); |
| 2015 | let wait = if code == Code::WaitingForImage { launcher::IMAGES_AGAIN } else { again(false, self.failures) }; |
| 2016 | self.not_before = Instant::now() + wait; |
| 2017 | // The cards say why from what `up` said, not from the stopped sandbox. |
| 2018 | if let Some(failed) = status { |
| 2019 | self.app.state::<Shared>().now.lock().unwrap().sandbox = Some(failed); |
| 2020 | } |
| 2021 | return self.set(code, Extra { detail: Some(cause), fix, wait_seconds: Some(wait.as_secs()), ..Extra::default() }); |
| 2022 | } |
| 2023 | } else if step == Step::Up { |
| 2024 | self.failures = 0; |
| 2025 | } else { |
| 2026 | self.update_failed = None; |
| 2027 | } |
| 2028 | if step == Step::Update && !running { |
| 2029 | // Updated while stopped: it starts again with the new image. |
| 2030 | self.run = None; |
| 2031 | self.generation += 1; |
| 2032 | self.not_before = Instant::now(); |
| 2033 | return; |
| 2034 | } |
| 2035 | if let Some(status) = status { |
| 2036 | self.sandbox_now(status); |
| 2037 | } |
| 2038 | } |
| 2039 | } |
| 2040 | |
| 2041 | /// What the tray says while the sandbox starts. |
| 2042 | const STARTING_SANDBOX: &str = "Starting the sandbox. The first time, it downloads and checks the sandbox's image, which takes a few minutes."; |
| 2043 | |
| 2044 | /// The newest release, signed, and its program when it is newer than the |
| 2045 | /// one in `folder` and isn't already `waiting`: downloaded and checked |
| 2046 | /// against its SHA-256. On the update thread, never the runner's. |
| 2047 | fn look_for_release(server: &Url, folder: &std::path::Path, waiting: Option<&str>) -> Result<Found, Fault> { |
| 2048 | let manifest = native::latest(server)?; |
| 2049 | let newer = native::installed(folder).is_none_or(|have| native::newer(&manifest.version, &have)); |
| 2050 | let program = if newer && waiting != Some(manifest.version.as_str()) { |
| 2051 | Some(native::download(server, &manifest).map_err(|why| Fault::Failed(format!("{} didn't download: {why}", manifest.version)))?) |
| 2052 | } else { |
| 2053 | None |
| 2054 | }; |
| 2055 | Ok(Found { images: manifest.has_images(), version: manifest.version, newer, program }) |
| 2056 | } |
| 2057 | |
| 2058 | fn some_text(text: &str) -> Option<String> { |
| 2059 | Some(text.to_owned()) |
| 2060 | } |
| 2061 | |
| 2062 | /// How long before trying again, the `tries`th time: every few minutes |
| 2063 | /// while the release has no sandbox images, else longer each time. |
| 2064 | fn again(images: bool, tries: u32) -> Duration { |
| 2065 | if images { launcher::IMAGES_AGAIN } else { Duration::from_secs(15 * 2u64.pow(tries.min(6))).min(MOST_BACKOFF) } |
| 2066 | } |
| 2067 | |
| 2068 | /// What the app shows for how the sandbox is, with no launcher command |
| 2069 | /// running: what its runners do while one runs (each kind of job's card |
| 2070 | /// says the rest), else what is wrong with Linux jobs. |
| 2071 | fn shown(status: &Status) -> (Code, Extra) { |
| 2072 | if status.job().is_some() { |
| 2073 | return (Code::Busy, Extra::default()); |
| 2074 | } |
| 2075 | if status.running() { |
| 2076 | return (Code::Idle, Extra::default()); |
| 2077 | } |
| 2078 | if let Some((code, cause, fix)) = diagnosis::sandbox_trouble(status) { |
| 2079 | return (code, Extra { detail: Some(cause), fix, ..Extra::default() }); |
| 2080 | } |
| 2081 | match status.linux().state { |
| 2082 | State::Starting => (Code::Starting, detail(STARTING_SANDBOX)), |
| 2083 | State::Updating => (Code::Updating, detail("Updating the sandbox to the newest image.")), |
| 2084 | _ if status.all_off() => (Code::KindsStopped, Extra::default()), |
| 2085 | _ => (Code::SandboxFailed, detail("The sandbox isn't running.")), |
| 2086 | } |
| 2087 | } |
| 2088 | |
| 2089 | /// What `register-and-run` reads: where g1t is, the runner's name and |
| 2090 | /// labels, and the token, the first time only. The app updates the runner, |
| 2091 | /// so the runner does not update itself; registering again takes the |
| 2092 | /// place of this computer's earlier runner of the same name. |
| 2093 | fn environment(server: &Url, backend: Backend, token: Option<&str>) -> Vec<(&'static str, String)> { |
| 2094 | let mut envs = vec![ |
| 2095 | ("G1T_RUNNER_URL", server.as_str().trim_end_matches('/').to_owned()), |
| 2096 | ("G1T_RUNNER_NAME", backend::machine_name()), |
| 2097 | ("G1T_RUNNER_LABELS", "desktop".to_owned()), |
| 2098 | ("G1T_RUNNER_REPLACE", "1".to_owned()), |
| 2099 | ("G1T_RUNNER_NO_AUTO_UPDATE", "1".to_owned()), |
| 2100 | ]; |
| 2101 | if backend == Backend::Host { |
| 2102 | envs.push(("G1T_RUNNER_NO_DOCKER", "1".to_owned())); |
| 2103 | } |
| 2104 | if let Some(token) = token { |
| 2105 | envs.push(("G1T_RUNNER_TOKEN", token.to_owned())); |
| 2106 | } |
| 2107 | envs |
| 2108 | } |
| 2109 | |
| 2110 | #[cfg(test)] |
| 2111 | mod tests { |
| 2112 | use super::*; |
| 2113 | |
| 2114 | fn view(code: Code, job: Option<&str>) -> View { |
| 2115 | let facts = Facts { workspace: Some("acme".into()), backend: "sandboxed", machine: "box".into(), job: job.map(str::to_owned), ..Facts::default() }; |
| 2116 | View { |
| 2117 | diagnosis: diagnosis::diagnose(code, &facts), |
| 2118 | closed: diagnosis::closed(&facts), |
| 2119 | running: code == Code::Idle, |
| 2120 | version: None, |
| 2121 | backend: Backend::Sandboxed, |
| 2122 | workspace: Some("acme".into()), |
| 2123 | workspaces: vec![], |
| 2124 | enabled: true, |
| 2125 | paused: false, |
| 2126 | link: None, |
| 2127 | machine: "box".into(), |
| 2128 | platform: "windows".into(), |
| 2129 | log: vec![], |
| 2130 | } |
| 2131 | } |
| 2132 | |
| 2133 | #[test] |
| 2134 | fn the_tray_says_what_the_runner_is_doing() { |
| 2135 | assert_eq!(headline(&view(Code::Idle, None)), "Runner: Waiting for jobs"); |
| 2136 | assert_eq!(headline(&view(Code::Busy, Some("Build (acme/web)"))), "Runner: Running Build (acme/web)"); |
| 2137 | assert_eq!(headline(&view(Code::NoDocker, None)), "Runner: Not running: Docker isn't running"); |
| 2138 | assert_eq!(headline(&view(Code::Busy, Some(&"x".repeat(200)))).chars().count(), 60); |
| 2139 | } |
| 2140 | |
| 2141 | #[test] |
| 2142 | fn the_runner_is_told_where_and_who_and_the_token_only_once() { |
| 2143 | let server = Url::parse("https://g1t.sh/").unwrap(); |
| 2144 | let envs = environment(&server, Backend::Host, Some("g1trt_abc")); |
| 2145 | let get = |name: &str| envs.iter().find(|(n, _)| *n == name).map(|(_, v)| v.as_str()); |
| 2146 | assert_eq!(get("G1T_RUNNER_URL"), Some("https://g1t.sh")); |
| 2147 | assert_eq!(get("G1T_RUNNER_LABELS"), Some("desktop")); |
| 2148 | assert_eq!(get("G1T_RUNNER_NO_DOCKER"), Some("1")); |
| 2149 | assert_eq!(get("G1T_RUNNER_NO_AUTO_UPDATE"), Some("1")); |
| 2150 | assert_eq!(get("G1T_RUNNER_TOKEN"), Some("g1trt_abc")); |
| 2151 | let again = environment(&server, Backend::Docker, None); |
| 2152 | assert!(!again.iter().any(|(name, _)| *name == "G1T_RUNNER_TOKEN" || *name == "G1T_RUNNER_NO_DOCKER")); |
| 2153 | } |
| 2154 | |
| 2155 | fn status(json: &str) -> Status { |
| 2156 | launcher::parse(json).unwrap() |
| 2157 | } |
| 2158 | |
| 2159 | #[test] |
| 2160 | fn the_sandbox_shows_what_its_runners_do_and_linux_trouble_first() { |
| 2161 | let running = status(r#"{"schema":1,"state":"running","capabilities":[{"id":"linux","available":true,"state":"running"},{"id":"windows","available":false,"state":"absent","cause":"Windows Sandbox is not turned on, or this edition of Windows has none (Home does not)."}]}"#); |
| 2162 | assert_eq!(shown(&running).0, Code::Idle); |
| 2163 | let busy = status(r#"{"schema":1,"state":"running","job":{"kind":"workflow","id":"j","name":"build","repo":"acme/web"}}"#); |
| 2164 | assert_eq!(shown(&busy).0, Code::Busy); |
| 2165 | let starting = status(r#"{"schema":1,"state":"starting","capabilities":[{"id":"linux","available":true,"state":"starting"}]}"#); |
| 2166 | assert_eq!(shown(&starting).0, Code::Starting); |
| 2167 | let updating = status(r#"{"schema":1,"state":"updating","capabilities":[{"id":"linux","available":true,"state":"updating"}]}"#); |
| 2168 | assert_eq!(shown(&updating).0, Code::Updating); |
| 2169 | // Windows jobs run while WSL is missing: the runner waits for jobs, |
| 2170 | // and the Linux card says what is wrong (diagnosis.rs). |
| 2171 | let no_wsl = status(r#"{"schema":1,"state":"error","error":"WSL is not installed, and the sandbox for Linux jobs is a WSL distribution of its own.","capabilities":[ |
| 2172 | {"id":"linux","available":false,"state":"error","cause":"WSL is not installed, and the sandbox for Linux jobs is a WSL distribution of its own.","fix":"Open a terminal as administrator and run `wsl --install --no-distribution`."}, |
| 2173 | {"id":"windows","available":true,"state":"running"}]}"#); |
| 2174 | assert_eq!(shown(&no_wsl).0, Code::Idle); |
| 2175 | let (code, extra) = shown(&status(r#"{"schema":1,"state":"error","error":"WSL is not installed, and the sandbox for Linux jobs is a WSL distribution of its own.","capabilities":[ |
| 2176 | {"id":"linux","available":false,"state":"error","cause":"WSL is not installed, and the sandbox for Linux jobs is a WSL distribution of its own."}, |
| 2177 | {"id":"windows","available":false,"state":"absent"}]}"#)); |
| 2178 | assert_eq!(code, Code::NoWsl); |
| 2179 | assert!(extra.detail.unwrap().starts_with("WSL is not installed")); |
| 2180 | // 0.3.0 without its images, on Home: waiting for them, not failing. |
| 2181 | let images = status(r#"{"schema":1,"state":"error","error":"release 0.3.0 has no sandbox images","capabilities":[{"id":"linux","available":true,"state":"error","turned_off":false},{"id":"windows","available":false,"state":"absent","turned_off":false}]}"#); |
| 2182 | assert_eq!(shown(&images).0, Code::WaitingForImage); |
| 2183 | assert_eq!(again(true, 5), launcher::IMAGES_AGAIN); |
| 2184 | assert_eq!(again(false, 1), Duration::from_secs(30)); |
| 2185 | assert_eq!(again(false, 9), MOST_BACKOFF); |
| 2186 | // Every kind of job stopped by the person. |
| 2187 | let off = status(r#"{"schema":1,"state":"stopped","capabilities":[{"id":"linux","available":true,"state":"stopped","turned_off":true},{"id":"windows","available":true,"state":"stopped","turned_off":true}]}"#); |
| 2188 | assert_eq!(shown(&off).0, Code::KindsStopped); |
| 2189 | let failed = status(r#"{"schema":1,"state":"error","error":"the sandbox did not start within 180 seconds","capabilities":[{"id":"linux","available":true,"state":"error"}]}"#); |
| 2190 | let (code, extra) = shown(&failed); |
| 2191 | assert_eq!((code, extra.detail.as_deref()), (Code::SandboxFailed, Some("the sandbox did not start within 180 seconds"))); |
| 2192 | let stopped = status(r#"{"schema":1,"state":"stopped","capabilities":[{"id":"linux","available":true,"state":"stopped"},{"id":"windows","available":true,"state":"stopped"}]}"#); |
| 2193 | assert_eq!(shown(&stopped).0, Code::SandboxFailed); |
| 2194 | } |
| 2195 | |
| 2196 | #[test] |
| 2197 | fn when_is_said_as_the_site_reads_it() { |
| 2198 | let at = |seconds: u64| rfc3339(SystemTime::UNIX_EPOCH + Duration::from_secs(seconds)); |
| 2199 | assert_eq!(at(0), "1970-01-01T00:00:00Z"); |
| 2200 | assert_eq!(at(1_791_806_645), "2026-10-12T12:04:05Z"); |
| 2201 | // The day after February 29th of a leap year. |
| 2202 | assert_eq!(at(1_709_251_199 + 1), "2024-03-01T00:00:00Z"); |
| 2203 | assert_eq!(at(951_782_400), "2000-02-29T00:00:00Z"); |
| 2204 | } |
| 2205 | |
| 2206 | /// What is running on the program itself, which an update drains first: |
| 2207 | /// Windows jobs running or starting, not Linux jobs, nor ones stopped. |
| 2208 | #[test] |
| 2209 | fn an_update_drains_only_what_runs_on_the_program() { |
| 2210 | let kinds = |json: &str| native_kinds(&status(json)).iter().map(|c| c.id.clone()).collect::<Vec<_>>(); |
| 2211 | assert_eq!( |
| 2212 | kinds(r#"{"schema":1,"state":"running","capabilities":[{"id":"linux","available":true,"state":"running","turned_off":false},{"id":"windows","available":true,"state":"running","turned_off":false}]}"#), |
| 2213 | ["windows"] |
| 2214 | ); |
| 2215 | assert!(kinds(r#"{"schema":1,"state":"running","capabilities":[{"id":"linux","available":true,"state":"running","turned_off":false},{"id":"windows","available":true,"state":"stopped","turned_off":true}]}"#).is_empty()); |
| 2216 | assert!(kinds(r#"{"schema":1,"state":"running","capabilities":[{"id":"linux","available":true,"state":"running"},{"id":"windows","available":false,"state":"absent"}]}"#).is_empty()); |
| 2217 | } |
| 2218 | |
| 2219 | #[test] |
| 2220 | fn the_site_names_workspaces_in_its_own_words() { |
| 2221 | let offered: Vec<Offered> = serde_json::from_str(r#"[{"slug":"acme","name":"Acme"},{"slug":"solo"}]"#).unwrap(); |
| 2222 | assert_eq!(offered[1], Offered { slug: "solo".into(), name: None }); |
| 2223 | } |
| 2224 | } |