flagon-io/g1t

public

1	//! Self-hosted runners: registering them, handing them work, hearing back,
2	//! and the people's side (listing, removing, groups, settings). See
3	//! `g1t_contracts::runners` for the model and the wire.
4	//!
5	//! Work reaches a runner only when it asks for it (`runner_poll`), so a
6	//! runner never needs an open port. A workflow job is taken by flipping its
7	//! row from `queued` to `in_progress` in one statement, exactly as g1t's
8	//! own sandboxes take one, and the harness on the machine then fetches it
9	//! and reports with the job's own token through the same API. Agent work
10	//! from the runner service is kept as a task with its environment sealed,
11	//! and handed over once.
12
13	use g1t_contracts::access::Capability;
14	use g1t_contracts::audit::{AuditActor, AuditOutcome, AuditTarget, NewAuditEntry, RecordAuditArgs, Surface};
15	use g1t_contracts::billing::{ComputeKind, RecordSandboxArgs};
16	use g1t_contracts::repos::{Repo, RepoPath};
17	use g1t_contracts::runners::{
18	self as model, ACTIVITY_LIMIT, CloudJob, DesktopDiagnosis, DesktopRunner, DesktopRunnersArgs, ReportDesktopArgs, HandedOverTask, RunnerActivity, RunnerActivityArgs, StuckJob, StuckJobsArgs, Assignment, CancelTaskArgs, CreateRegistrationTokenArgs, DeleteRunnerGroupArgs, EnqueueTaskArgs, FinishedArgs,
19	ListRunnersArgs, Poll, PollArgs, RegisterArgs, Registered, RegistrationToken, RemoveRunnerArgs, RemoveSelfArgs, RouteArgs, Runner,
20	RunnerAuth, RunnerGroup, RunnerGroupsArgs, RunnerSettings, RunnerSettingsArgs, RunnerWork, RunnersOwner, SetRunnerGroupArgs,
21	SetRunnerSettingsArgs, Wanted,
22	};
23	use g1t_contracts::time::rfc3339;
24	use g1t_contracts::{FailureCode, Outcome, PrincipalKind, Role, User, new_id};
25	use g1t_kit::now_ms;
26	use g1t_secrets::{random_hex, sha256_hex};
27	use serde::Deserialize;
28	use serde_json::{Map, Value, json};
29	use worker::Result;
30	use worker::wasm_bindgen::JsValue;
31
32	use crate::plan::JobRow;
33	use crate::{Actions, Count, SILENT_MS, check, fail, optional};
34
35	/// What one poll has learned while it looks for work, so its looks again
36	/// do not ask again: each repository's visibility, and the jobs it already
37	/// said wait for a sandboxed runner.
38	#[derive(Default)]
39	struct Looked {
40	visibilities: std::collections::HashMap<String, Option<String>>,
41	noted: std::collections::HashSet<String>,
42	}
43
44	/// How often a poll looks again for work while it waits.
45	const POLL_EVERY_MS: u64 = 2_500;
46	/// A runner that has been offline this long is removed, as on GitHub.
47	const FORGET_OFFLINE_MS: u64 = 14 * 24 * 60 * 60 * 1000;
48	/// How often a poll writes down that a runner, or the work it holds, is
49	/// still there. Polls come every twenty seconds or so; a write each time
50	/// would be most of the service's writes for nothing. Well inside
51	/// [`model::ONLINE_WITHIN_MS`] and [`crate::SILENT_MS`].
52	const SEEN_RESOLUTION_MS: u64 = 60 * 1000;
53
54	/// Whether `seen` is old enough, at `at`, to be written down again.
55	fn seen_is_stale(seen: Option<&str>, at: u64) -> bool {
56	seen.is_none_or(\|seen\| seen < rfc3339(at.saturating_sub(SEEN_RESOLUTION_MS)).as_str())
57	}
58
59	#[derive(Clone, Debug, Deserialize)]
60	pub struct RunnerRow {
61	pub id: String,
62	pub workspace: String,
63	pub repo_id: Option<String>,
64	pub repo: Option<String>,
65	pub group_id: Option<String>,
66	pub name: String,
67	pub labels: String,
68	pub os: String,
69	pub arch: String,
70	pub version: String,
71	pub ephemeral: u32,
72	pub credential_hash: String,
73	pub previous_hash: Option<String>,
74	pub rotated_at: String,
75	pub work_id: Option<String>,
76	pub work_kind: Option<String>,
77	pub spent: u32,
78	pub last_seen_at: Option<String>,
79	pub created_at: String,
80	pub created_by: Option<String>,
81	/// `nested-container`, `container` or `host` (migration 0015); none
82	/// for a runner that never said.
83	#[serde(default)]
84	pub isolation: Option<String>,
85	}
86
87	impl RunnerRow {
88	fn labels(&self) -> Vec<String> {
89	serde_json::from_str(&self.labels).unwrap_or_default()
90	}
91
92	fn online(&self, now: u64) -> bool {
93	self.last_seen_at.as_deref().is_some_and(\|seen\| seen >= rfc3339(now.saturating_sub(model::ONLINE_WITHIN_MS)).as_str())
94	}
95	}
96
97	#[derive(Clone, Debug, Deserialize)]
98	struct GroupRow {
99	id: String,
100	name: String,
101	is_default: u32,
102	repositories: String,
103	updated_at: String,
104	}
105
106	impl GroupRow {
107	fn repositories(&self) -> Vec<String> {
108	serde_json::from_str(&self.repositories).unwrap_or_default()
109	}
110
111	/// Whether `repo` (`owner/name`, or a bare name) may use the group.
112	fn allows(&self, repo: &str) -> bool {
113	let name = repo.rsplit('/').next().unwrap_or(repo);
114	let listed = self.repositories();
115	listed.is_empty() \|\| listed.iter().any(\|r\| r.eq_ignore_ascii_case(name))
116	}
117	}
118
119	#[derive(Clone, Debug, Deserialize)]
120	struct RegistrationRow {
121	workspace: String,
122	repo_id: Option<String>,
123	repo: Option<String>,
124	group_id: Option<String>,
125	expires_at: String,
126	/// Who made the token: the runners it registers are theirs.
127	#[serde(default)]
128	created_by: Option<String>,
129	}
130
131	#[derive(Clone, Debug, Deserialize)]
132	struct SettingsRow {
133	agents: u32,
134	agent_labels: String,
135	fork_pulls: u32,
136	}
137
138	#[derive(Clone, Debug, Deserialize)]
139	pub struct TaskRow {
140	pub id: String,
141	pub sandbox: String,
142	pub repo_id: Option<String>,
143	pub repo: String,
144	pub kind: String,
145	pub title: String,
146	pub labels: String,
147	pub env: Option<String>,
148	pub timeout_minutes: u32,
149	pub status: String,
150	pub runner_id: Option<String>,
151	pub runner_name: Option<String>,
152	pub started_at: Option<String>,
153	#[serde(default)]
154	pub seen_at: Option<String>,
155	/// The agent run it belongs to (migration 0012); none for older tasks.
156	#[serde(default)]
157	pub run_id: Option<String>,
158	}
159
160	/// Who may do what with a set of runners: the workspace, and the
161	/// repository when they are a repository's own.
162	struct Place {
163	workspace: String,
164	repo: Option<Repo>,
165	}
166
167	fn now() -> String {
168	rfc3339(now_ms())
169	}
170
171	fn valid_name(name: &str) -> std::result::Result<String, String> {
172	let name = name.trim();
173	let ok = !name.is_empty()
174	&& name.len() <= 64
175	&& name.chars().all(\|c\| c.is_ascii_alphanumeric() \|\| matches!(c, '-' \| '_' \| '.'));
176	if ok {
177	Ok(name.to_owned())
178	} else {
179	Err("A runner's name is 1 to 64 letters, digits, `-`, `_` and `.`.".to_owned())
180	}
181	}
182
183	fn valid_group_name(name: &str) -> std::result::Result<String, String> {
184	let name = name.trim();
185	if name.is_empty() \|\| name.len() > 64 \|\| name.chars().any(\|c\| c.is_control()) {
186	return Err("A group's name is 1 to 64 characters.".to_owned());
187	}
188	Ok(name.to_owned())
189	}
190
191	/// What a runner's credential is checked against, and what a registration
192	/// token is found by.
193	fn hash(secret: &str) -> String {
194	sha256_hex(secret)
195	}
196
197	/// The actor of a runner's own entries in the audit log.
198	fn runner_actor(runner: &RunnerRow) -> AuditActor {
199	AuditActor {
200	actor_kind: Some(g1t_contracts::audit::ActorKind::Runner),
201	actor: runner.name.clone(),
202	actor_id: runner.id.clone(),
203	..AuditActor::default()
204	}
205	}
206
207	fn entry(actor: AuditActor, action: &str, workspace: &str, repo: Option<String>, path: Option<String>, message: String) -> NewAuditEntry {
208	NewAuditEntry {
209	actor,
210	action: action.to_owned(),
211	surface: Surface::Rest,
212	target: AuditTarget {
213	workspace: workspace.to_owned(),
214	repo,
215	path,
216	..AuditTarget::default()
217	},
218	outcome: AuditOutcome::Allowed,
219	rule: "runner".to_owned(),
220	result: Some("ok".to_owned()),
221	message: Some(message),
222	request_id: new_id("req", now_ms()),
223	}
224	}
225
226	impl Actions {
227	// --- Shared ------------------------------------------------------------
228
229	pub(crate) async fn audit(&self, entries: Vec<NewAuditEntry>) {
230	let recorded: Result<u32> = g1t_kit::call(&self.events, "audit_record", &RecordAuditArgs { entries }).await;
231	if let Err(error) = recorded {
232	worker::console_error!("actions: audit entries not recorded: {error}");
233	}
234	}
235
236	/// Where `owner` points, if `actor` may see it (`manage` false) or
237	/// change it (`manage` true): a workspace's runners are seen by its
238	/// members and changed by its owners; a repository's own, by its
239	/// admins. Agents and workspace tokens never change them, so a
240	/// workflow's `G1T_TOKEN` cannot add a machine to run its own jobs.
241	async fn runner_place(&self, actor: &User, owner: &RunnersOwner, manage: bool) -> Result<Outcome<Place>> {
242	if actor.kind == PrincipalKind::Agent {
243	return Ok(fail(FailureCode::Forbidden, "Agents cannot see or change self-hosted runners."));
244	}
245	if manage && actor.kind == PrincipalKind::Workspace {
246	return Ok(fail(
247	FailureCode::Forbidden,
248	"A workspace's tokens, G1T_TOKEN included, cannot change self-hosted runners. Use a person's token with the runners:admin scope.",
249	));
250	}
251	match (&owner.repo, &owner.workspace) {
252	(Some(path), _) => {
253	let repo = check!(self.may(actor, path, Capability::ManageIntegrations).await?);
254	Ok(Outcome::Ok(Place { workspace: repo.namespace.to_lowercase(), repo: Some(repo) }))
255	}
256	(None, Some(slug)) => {
257	let slug = slug.to_lowercase();
258	let role = actor.workspaces.iter().find(\|m\| m.slug.eq_ignore_ascii_case(&slug)).map(\|m\| m.role);
259	match role {
260	None => Ok(fail(FailureCode::NotFound, "There is no such workspace, or you are not a member of it.")),
261	// A workspace's machines and their tokens are its owners' alone, to see as well as change.
262	Some(Role::Member) => Ok(fail(
263	FailureCode::Forbidden,
264	format!("Only owners of {slug} can {} its self-hosted runners.", if manage { "change" } else { "see" }),
265	)),
266	Some(_) => Ok(Outcome::Ok(Place { workspace: slug, repo: None })),
267	}
268	}
269	(None, None) => Ok(fail(FailureCode::Invalid, "Give `repo` or `workspace`.")),
270	}
271	}
272
273	async fn group_rows(&self, workspace: &str) -> Result<Vec<GroupRow>> {
274	self.db
275	.prepare("SELECT * FROM runner_groups WHERE workspace = ? ORDER BY is_default DESC, name")
276	.bind(&[workspace.into()])?
277	.all()
278	.await?
279	.results::<GroupRow>()
280	}
281
282	/// The workspace's default group, made the first time it is needed.
283	async fn default_group(&self, workspace: &str) -> Result<GroupRow> {
284	let at = now();
285	self.db
286	.prepare(
287	"INSERT OR IGNORE INTO runner_groups (id, workspace, name, is_default, repositories, created_at, updated_at)
288	VALUES (?, ?, 'Default', 1, '[]', ?, ?)",
289	)
290	.bind(&[new_id("rng", now_ms()).into(), workspace.into(), at.as_str().into(), at.as_str().into()])?
291	.run()
292	.await?;
293	let row = self
294	.db
295	.prepare("SELECT * FROM runner_groups WHERE workspace = ? AND is_default = 1 LIMIT 1")
296	.bind(&[workspace.into()])?
297	.first::<GroupRow>(None)
298	.await?;
299	row.ok_or_else(\|\| worker::Error::RustError("the default runner group could not be made".into()))
300	}
301
302	/// A group of the workspace by id or name.
303	async fn find_group(&self, workspace: &str, wanted: &str) -> Result<Option<GroupRow>> {
304	self.default_group(workspace).await?;
305	Ok(self
306	.group_rows(workspace)
307	.await?
308	.into_iter()
309	.find(\|g\| g.id == wanted \|\| g.name.eq_ignore_ascii_case(wanted.trim())))
310	}
311
312	async fn runner_rows(&self, workspace: &str) -> Result<Vec<RunnerRow>> {
313	self.db
314	.prepare("SELECT * FROM runners WHERE workspace = ? ORDER BY name")
315	.bind(&[workspace.into()])?
316	.all()
317	.await?
318	.results::<RunnerRow>()
319	}
320
321	async fn runner_by_id(&self, id: &str) -> Result<Option<RunnerRow>> {
322	self.db.prepare("SELECT * FROM runners WHERE id = ?").bind(&[id.into()])?.first::<RunnerRow>(None).await
323	}
324
325	/// What a runner is doing, for people.
326	async fn work_of(&self, row: &RunnerRow) -> Result<Option<RunnerWork>> {
327	let Some(id) = &row.work_id else { return Ok(None) };
328	if row.work_kind.as_deref() == Some("agent") {
329	let task = self.db.prepare("SELECT * FROM runner_tasks WHERE id = ?").bind(&[id.as_str().into()])?.first::<TaskRow>(None).await?;
330	return Ok(task.filter(\|t\| t.status == "in_progress").map(\|t\| RunnerWork {
331	kind: "agent".into(),
332	id: t.id,
333	name: t.title,
334	repo: Some(t.repo),
335	run_id: t.run_id,
336	started_at: t.started_at,
337	}));
338	}
339	#[derive(Deserialize)]
340	struct Found {
341	id: String,
342	name: String,
343	run_id: String,
344	status: String,
345	started_at: Option<String>,
346	repo: String,
347	}
348	let job = self
349	.db
350	.prepare("SELECT jobs.id, jobs.name, jobs.run_id, jobs.status, jobs.started_at, runs.repo FROM jobs JOIN runs ON runs.id = jobs.run_id WHERE jobs.id = ?")
351	.bind(&[id.as_str().into()])?
352	.first::<Found>(None)
353	.await?;
354	Ok(job.filter(\|j\| j.status == "in_progress").map(\|j\| RunnerWork {
355	kind: "workflow".into(),
356	id: j.id,
357	name: j.name,
358	repo: Some(j.repo),
359	run_id: Some(j.run_id),
360	started_at: j.started_at,
361	}))
362	}
363
364	async fn runner_view(&self, row: &RunnerRow, groups: &[GroupRow], at: u64) -> Result<Runner> {
365	let online = row.online(at);
366	let work = if online { self.work_of(row).await? } else { None };
367	Ok(Runner {
368	id: row.id.clone(),
369	name: row.name.clone(),
370	workspace: row.workspace.clone(),
371	repo: row.repo.clone(),
372	group: row
373	.group_id
374	.as_ref()
375	.and_then(\|id\| groups.iter().find(\|g\| &g.id == id))
376	.map(\|g\| g.name.clone()),
377	labels: row.labels(),
378	os: row.os.clone(),
379	arch: row.arch.clone(),
380	version: row.version.clone(),
381	ephemeral: row.ephemeral != 0,
382	status: if !online {
383	"offline"
384	} else if work.is_some() {
385	"busy"
386	} else {
387	"online"
388	}
389	.to_owned(),
390	work,
391	last_seen_at: row.last_seen_at.clone(),
392	created_at: row.created_at.clone(),
393	created_by: row.created_by.clone(),
394	})
395	}
396
397	// --- People's side -----------------------------------------------------
398
399	/// `runners`: a workspace's runners, or a repository's: its own and
400	/// the workspace's that its group lets it use.
401	pub async fn runners(&self, a: ListRunnersArgs) -> Result<Outcome<Vec<Runner>>> {
402	let place = check!(self.runner_place(&a.actor, &a.owner, false).await?);
403	self.default_group(&place.workspace).await?;
404	let groups = self.group_rows(&place.workspace).await?;
405	let at = now_ms();
406	let mut out = Vec::new();
407	for row in self.runner_rows(&place.workspace).await? {
408	let shown = match &place.repo {
409	None => row.repo_id.is_none(),
410	Some(repo) => match &row.repo_id {
411	Some(id) => id == &repo.id,
412	None => groups.iter().find(\|g\| Some(&g.id) == row.group_id.as_ref()).is_none_or(\|g\| g.allows(&repo.name)),
413	},
414	};
415	if shown {
416	out.push(self.runner_view(&row, &groups, at).await?);
417	}
418	}
419	Ok(Outcome::Ok(out))
420	}
421
422	/// `create_registration_token`.
423	pub async fn create_registration_token(&self, a: CreateRegistrationTokenArgs) -> Result<Outcome<RegistrationToken>> {
424	let place = check!(self.runner_place(&a.actor, &a.owner, true).await?);
425	let made = self
426	.db
427	.prepare("SELECT COUNT(*) AS n FROM runner_registrations WHERE workspace = ? AND created_at > ?")
428	.bind(&[place.workspace.as_str().into(), rfc3339(now_ms().saturating_sub(60 * 60 * 1000)).into()])?
429	.first::<Count>(None)
430	.await?
431	.map_or(0, \|c\| c.n);
432	if made >= model::MAX_TOKENS_PER_HOUR {
433	return Ok(fail(FailureCode::Conflict, "Too many registration tokens in the last hour. Each lasts an hour and registers any number of runners."));
434	}
435	let group = match (&place.repo, a.group.as_deref().map(str::trim).filter(\|g\| !g.is_empty())) {
436	(Some(_), Some(_)) => return Ok(fail(FailureCode::Invalid, "A repository's own runners are in no group.")),
437	(Some(_), None) => None,
438	(None, Some(wanted)) => match self.find_group(&place.workspace, wanted).await? {
439	Some(group) => Some(group),
440	None => return Ok(fail(FailureCode::NotFound, format!("There is no runner group called {wanted}."))),
441	},
442	(None, None) => Some(self.default_group(&place.workspace).await?),
443	};
444	let token = format!("{}{}", model::REGISTRATION_PREFIX, random_hex(24));
445	let at = now_ms();
446	let expires_at = rfc3339(at + model::REGISTRATION_TTL_SECONDS * 1000);
447	let repo_name = place.repo.as_ref().map(\|r\| format!("{}/{}", r.namespace, r.name));
448	self.db
449	.prepare(
450	"INSERT INTO runner_registrations (token_hash, workspace, repo_id, repo, group_id, created_by, created_at, expires_at)
451	VALUES (?, ?, ?, ?, ?, ?, ?, ?)",
452	)
453	.bind(&[
454	hash(&token).into(),
455	place.workspace.as_str().into(),
456	optional(place.repo.as_ref().map(\|r\| r.id.as_str())),
457	optional(repo_name.as_deref()),
458	optional(group.as_ref().map(\|g\| g.id.as_str())),
459	a.actor.username.as_str().into(),
460	rfc3339(at).into(),
461	expires_at.as_str().into(),
462	])?
463	.run()
464	.await?;
465	self.audit(vec![entry(
466	AuditActor::of(&a.actor),
467	"create_runner_registration_token",
468	&place.workspace,
469	repo_name.clone(),
470	None,
471	format!("Made a runner registration token{}.", group.as_ref().map(\|g\| format!(" for group {}", g.name)).unwrap_or_default()),
472	)])
473	.await;
474	Ok(Outcome::Ok(RegistrationToken {
475	token,
476	expires_at,
477	workspace: place.workspace,
478	repo: repo_name,
479	group: group.map(\|g\| g.name),
480	url: crate::SITE.to_owned(),
481	}))
482	}
483
484	/// `remove_runner`: a person removing one.
485	pub async fn remove_runner(&self, a: RemoveRunnerArgs) -> Result<Outcome<bool>> {
486	let place = check!(self.runner_place(&a.actor, &a.owner, true).await?);
487	let Some(row) = self.runner_by_id(&a.id).await? else {
488	return Ok(fail(FailureCode::NotFound, "There is no such runner."));
489	};
490	let ours = row.workspace == place.workspace
491	&& match &place.repo {
492	Some(repo) => row.repo_id.as_deref() == Some(repo.id.as_str()),
493	None => row.repo_id.is_none(),
494	};
495	if !ours {
496	return Ok(fail(FailureCode::NotFound, "There is no such runner."));
497	}
498	self.forget_runner(&row, &format!("{} removed the runner {}.", a.actor.username, row.name)).await?;
499	self.audit(vec![entry(
500	AuditActor::of(&a.actor),
501	"remove_runner",
502	&row.workspace,
503	row.repo.clone(),
504	Some(row.name.clone()),
505	format!("Removed the self-hosted runner {}.", row.name),
506	)])
507	.await;
508	Ok(Outcome::Ok(true))
509	}
510
511	/// Removes a runner: what it was running fails, and its credential
512	/// stops working at once.
513	async fn forget_runner(&self, row: &RunnerRow, why: &str) -> Result<()> {
514	self.db.prepare("DELETE FROM runners WHERE id = ?").bind(&[row.id.as_str().into()])?.run().await?;
515	self.abandon_work(&row.id, why).await
516	}
517
518	/// Fails whatever a runner was running.
519	async fn abandon_work(&self, runner_id: &str, why: &str) -> Result<()> {
520	let jobs = self
521	.db
522	.prepare("SELECT id FROM jobs WHERE runner_id = ? AND status = 'in_progress'")
523	.bind(&[runner_id.into()])?
524	.all()
525	.await?
526	.results::<IdRow>()?;
527	for job in jobs {
528	self.finish_job(&job.id, "failure", Some(why), None).await?;
529	}
530	let tasks = self
531	.db
532	.prepare("SELECT * FROM runner_tasks WHERE runner_id = ? AND status = 'in_progress'")
533	.bind(&[runner_id.into()])?
534	.all()
535	.await?
536	.results::<TaskRow>()?;
537	for task in tasks {
538	self.end_task(&task, 1, why).await?;
539	}
540	Ok(())
541	}
542
543	/// `runner_groups`.
544	pub async fn runner_groups(&self, a: RunnerGroupsArgs) -> Result<Outcome<Vec<RunnerGroup>>> {
545	let owner = RunnersOwner { repo: None, workspace: Some(a.workspace.clone()) };
546	let place = check!(self.runner_place(&a.actor, &owner, false).await?);
547	self.default_group(&place.workspace).await?;
548	let groups = self.group_rows(&place.workspace).await?;
549	let runners = self.runner_rows(&place.workspace).await?;
550	Ok(Outcome::Ok(groups.iter().map(\|g\| group_view(g, &runners)).collect()))
551	}
552
553	/// `set_runner_group`.
554	pub async fn set_runner_group(&self, a: SetRunnerGroupArgs) -> Result<Outcome<RunnerGroup>> {
555	let owner = RunnersOwner { repo: None, workspace: Some(a.workspace.clone()) };
556	let place = check!(self.runner_place(&a.actor, &owner, true).await?);
557	self.default_group(&place.workspace).await?;
558	let groups = self.group_rows(&place.workspace).await?;
559	let name = match a.name.as_deref().map(valid_group_name).transpose() {
560	Ok(name) => name,
561	Err(problem) => return Ok(fail(FailureCode::Invalid, problem)),
562	};
563	let repositories: Option<Vec<String>> = a.repositories.map(\|list\| {
564	let mut out: Vec<String> = Vec::new();
565	for repo in list {
566	let name = repo.trim().rsplit('/').next().unwrap_or_default().to_owned();
567	if !name.is_empty() && !out.iter().any(\|r: &String\| r.eq_ignore_ascii_case(&name)) {
568	out.push(name);
569	}
570	}
571	out
572	});
573	if let Some(name) = &name
574	&& groups.iter().any(\|g\| g.name.eq_ignore_ascii_case(name) && Some(&g.id) != a.id.as_ref())
575	{
576	return Ok(fail(FailureCode::Conflict, format!("There is already a group called {name}.")));
577	}
578	let at = now();
579	let id = match &a.id {
580	Some(id) => {
581	let Some(group) = groups.iter().find(\|g\| &g.id == id) else {
582	return Ok(fail(FailureCode::NotFound, "There is no such runner group."));
583	};
584	self.db
585	.prepare("UPDATE runner_groups SET name = ?, repositories = ?, updated_at = ? WHERE id = ?")
586	.bind(&[
587	name.clone().unwrap_or(group.name.clone()).into(),
588	serde_json::to_string(&repositories.clone().unwrap_or(group.repositories()))?.into(),
589	at.as_str().into(),
590	id.as_str().into(),
591	])?
592	.run()
593	.await?;
594	id.clone()
595	}
596	None => {
597	let Some(name) = name.clone() else {
598	return Ok(fail(FailureCode::Invalid, "A new group needs a name."));
599	};
600	if groups.len() >= 50 {
601	return Ok(fail(FailureCode::Conflict, "A workspace has at most 50 runner groups."));
602	}
603	let id = new_id("rng", now_ms());
604	self.db
605	.prepare(
606	"INSERT INTO runner_groups (id, workspace, name, is_default, repositories, created_at, updated_at) VALUES (?, ?, ?, 0, ?, ?, ?)",
607	)
608	.bind(&[
609	id.as_str().into(),
610	place.workspace.as_str().into(),
611	name.into(),
612	serde_json::to_string(&repositories.clone().unwrap_or_default())?.into(),
613	at.as_str().into(),
614	at.as_str().into(),
615	])?
616	.run()
617	.await?;
618	id
619	}
620	};
621	let groups = self.group_rows(&place.workspace).await?;
622	let runners = self.runner_rows(&place.workspace).await?;
623	let Some(group) = groups.iter().find(\|g\| g.id == id) else {
624	return Ok(fail(FailureCode::NotFound, "There is no such runner group."));
625	};
626	self.audit(vec![entry(
627	AuditActor::of(&a.actor),
628	if a.id.is_some() { "update_runner_group" } else { "create_runner_group" },
629	&place.workspace,
630	None,
631	Some(group.name.clone()),
632	format!(
633	"Runner group {}: {}.",
634	group.name,
635	if group.repositories().is_empty() { "every repository".to_owned() } else { group.repositories().join(", ") }
636	),
637	)])
638	.await;
639	Ok(Outcome::Ok(group_view(group, &runners)))
640	}
641
642	/// `delete_runner_group`.
643	pub async fn delete_runner_group(&self, a: DeleteRunnerGroupArgs) -> Result<Outcome<bool>> {
644	let owner = RunnersOwner { repo: None, workspace: Some(a.workspace.clone()) };
645	let place = check!(self.runner_place(&a.actor, &owner, true).await?);
646	let default = self.default_group(&place.workspace).await?;
647	let Some(group) = self.group_rows(&place.workspace).await?.into_iter().find(\|g\| g.id == a.id) else {
648	return Ok(fail(FailureCode::NotFound, "There is no such runner group."));
649	};
650	if group.is_default != 0 {
651	return Ok(fail(FailureCode::Conflict, "The default group cannot be deleted."));
652	}
653	self.db
654	.batch(vec![
655	self.db.prepare("UPDATE runners SET group_id = ? WHERE group_id = ?").bind(&[default.id.as_str().into(), group.id.as_str().into()])?,
656	self.db.prepare("DELETE FROM runner_groups WHERE id = ?").bind(&[group.id.as_str().into()])?,
657	])
658	.await?;
659	self.audit(vec![entry(
660	AuditActor::of(&a.actor),
661	"delete_runner_group",
662	&place.workspace,
663	None,
664	Some(group.name.clone()),
665	format!("Deleted the runner group {}; its runners joined {}.", group.name, default.name),
666	)])
667	.await;
668	Ok(Outcome::Ok(true))
669	}
670
671	async fn settings_row(&self, owner: &str) -> Result<Option<SettingsRow>> {
672	self.db.prepare("SELECT * FROM runner_settings WHERE owner = ?").bind(&[owner.into()])?.first::<SettingsRow>(None).await
673	}
674
675	fn settings_of(row: Option<SettingsRow>, inherited: bool) -> RunnerSettings {
676	match row {
677	Some(row) => RunnerSettings {
678	agents_on_self_hosted: row.agents != 0,
679	agent_labels: serde_json::from_str(&row.agent_labels).unwrap_or_else(\|_\| vec![model::SELF_HOSTED.to_owned()]),
680	fork_pull_requests: row.fork_pulls != 0,
681	inherited,
682	},
683	None => RunnerSettings {
684	agent_labels: vec![model::SELF_HOSTED.to_owned()],
685	inherited,
686	..RunnerSettings::default()
687	},
688	}
689	}
690
691	/// The settings that apply in a workspace, or in one of its
692	/// repositories: the repository's own if it has them.
693	pub async fn effective_runner_settings(&self, workspace: &str, repo_id: Option<&str>) -> Result<RunnerSettings> {
694	if let Some(repo_id) = repo_id
695	&& let Some(own) = self.settings_row(repo_id).await?
696	{
697	return Ok(Self::settings_of(Some(own), false));
698	}
699	let inherited = repo_id.is_some();
700	Ok(Self::settings_of(self.settings_row(&workspace.to_lowercase()).await?, inherited))
701	}
702
703	/// `runner_settings`.
704	pub async fn runner_settings(&self, a: RunnerSettingsArgs) -> Result<Outcome<RunnerSettings>> {
705	let place = check!(self.runner_place(&a.actor, &a.owner, false).await?);
706	Ok(Outcome::Ok(self.effective_runner_settings(&place.workspace, place.repo.as_ref().map(\|r\| r.id.as_str())).await?))
707	}
708
709	/// `report_desktop`: what a person's desktop app says of its runner,
710	/// in a workspace they own. Kept per computer, replaced each time.
711	pub async fn report_desktop(&self, a: ReportDesktopArgs) -> Result<Outcome<bool>> {
712	let owner = RunnersOwner { repo: None, workspace: Some(a.workspace.clone()) };
713	let place = check!(self.runner_place(&a.actor, &owner, true).await?);
714	let machine = match valid_name(&a.machine) {
715	Ok(name) => name,
716	Err(problem) => return Ok(fail(FailureCode::Invalid, problem)),
717	};
718	if !matches!(a.backend.as_str(), "sandboxed" \| "docker" \| "host") {
719	return Ok(fail(FailureCode::Invalid, "A desktop runs jobs sandboxed, in Docker or on the host."));
720	}
721	let clip = \|text: &str, most: usize\| text.chars().take(most).collect::<String>();
722	let tidy = \|d: &DesktopDiagnosis\| DesktopDiagnosis {
723	code: clip(&d.code, 40),
724	tone: clip(&d.tone, 20),
725	title: clip(&d.title, 200),
726	cause: d.cause.as_deref().map(\|t\| clip(t, 1000)),
727	fix: d.fix.as_deref().map(\|t\| clip(t, 1000)),
728	};
729	let at = now_ms();
730	let known = self
731	.db
732	.prepare("SELECT COUNT(*) AS n FROM desktop_runners WHERE workspace = ? AND lower(machine) <> lower(?)")
733	.bind(&[place.workspace.as_str().into(), machine.as_str().into()])?
734	.first::<Count>(None)
735	.await?
736	.map_or(0, \|c\| c.n);
737	if known >= model::MAX_DESKTOPS {
738	// The oldest report makes room: a computer long gone.
739	self.db
740	.prepare("DELETE FROM desktop_runners WHERE rowid IN (SELECT rowid FROM desktop_runners WHERE workspace = ? ORDER BY reported_at LIMIT 1)")
741	.bind(&[place.workspace.as_str().into()])?
742	.run()
743	.await?;
744	}
745	self.db
746	.prepare(
747	"INSERT INTO desktop_runners (workspace, machine, person, backend, version, diagnosis, closed, reported_at)
748	VALUES (?, ?, ?, ?, ?, ?, ?, ?)
749	ON CONFLICT (workspace, machine) DO UPDATE SET person = ?3, backend = ?4, version = ?5, diagnosis = ?6, closed = ?7, reported_at = ?8",
750	)
751	.bind(&[
752	place.workspace.as_str().into(),
753	machine.as_str().into(),
754	a.actor.username.as_str().into(),
755	a.backend.as_str().into(),
756	optional(a.version.as_deref().map(\|v\| clip(v, 40)).as_deref()),
757	serde_json::to_string(&tidy(&a.diagnosis))?.into(),
758	serde_json::to_string(&tidy(&a.closed))?.into(),
759	rfc3339(at).into(),
760	])?
761	.run()
762	.await?;
763	Ok(Outcome::Ok(true))
764	}
765
766	/// `desktop_runners`: the workspace's desktops, as each app last said.
767	pub async fn desktop_runners(&self, a: DesktopRunnersArgs) -> Result<Outcome<Vec<DesktopRunner>>> {
768	let owner = RunnersOwner { repo: None, workspace: Some(a.workspace.clone()) };
769	let place = check!(self.runner_place(&a.actor, &owner, false).await?);
770	#[derive(Deserialize)]
771	struct Row {
772	workspace: String,
773	machine: String,
774	person: String,
775	backend: String,
776	version: Option<String>,
777	diagnosis: String,
778	closed: String,
779	reported_at: String,
780	}
781	let rows = self
782	.db
783	.prepare("SELECT * FROM desktop_runners WHERE workspace = ? ORDER BY lower(machine)")
784	.bind(&[place.workspace.as_str().into()])?
785	.all()
786	.await?
787	.results::<Row>()?;
788	let open_since = rfc3339(now_ms().saturating_sub(model::DESKTOP_OPEN_WITHIN_MS));
789	Ok(Outcome::Ok(
790	rows.into_iter()
791	.map(\|row\| DesktopRunner {
792	open: row.reported_at.as_str() > open_since.as_str(),
793	diagnosis: serde_json::from_str(&row.diagnosis).unwrap_or_default(),
794	closed: serde_json::from_str(&row.closed).unwrap_or_default(),
795	workspace: row.workspace,
796	machine: row.machine,
797	person: row.person,
798	backend: row.backend,
799	version: row.version,
800	reported_at: row.reported_at,
801	})
802	.collect(),
803	))
804	}
805
806	/// `set_runner_settings`.
807	pub async fn set_runner_settings(&self, a: SetRunnerSettingsArgs) -> Result<Outcome<RunnerSettings>> {
808	let place = check!(self.runner_place(&a.actor, &a.owner, true).await?);
809	let (owner, scope) = match &place.repo {
810	Some(repo) => (repo.id.clone(), "repository"),
811	None => (place.workspace.clone(), "workspace"),
812	};
813	if a.inherit && place.repo.is_some() {
814	self.db.prepare("DELETE FROM runner_settings WHERE owner = ?").bind(&[owner.as_str().into()])?.run().await?;
815	} else {
816	let current = self.effective_runner_settings(&place.workspace, place.repo.as_ref().map(\|r\| r.id.as_str())).await?;
817	let labels = match &a.agent_labels {
818	Some(given) => match model::runner_labels(given, model::SELF_HOSTED, model::SELF_HOSTED) {
819	Ok(labels) => labels,
820	Err(problem) => return Ok(fail(FailureCode::Invalid, problem)),
821	},
822	None => current.agent_labels.clone(),
823	};
824	let agents = a.agents_on_self_hosted.unwrap_or(current.agents_on_self_hosted);
825	let forks = a.fork_pull_requests.unwrap_or(current.fork_pull_requests);
826	self.db
827	.prepare(
828	"INSERT INTO runner_settings (owner, scope, agents, agent_labels, fork_pulls, updated_at, updated_by)
829	VALUES (?1, ?2, ?3, ?4, ?5, ?6, ?7)
830	ON CONFLICT (owner) DO UPDATE SET agents = ?3, agent_labels = ?4, fork_pulls = ?5, updated_at = ?6, updated_by = ?7",
831	)
832	.bind(&[
833	owner.as_str().into(),
834	scope.into(),
835	u32::from(agents).into(),
836	serde_json::to_string(&labels)?.into(),
837	u32::from(forks).into(),
838	now().into(),
839	a.actor.username.as_str().into(),
840	])?
841	.run()
842	.await?;
843	}
844	let settings = self.effective_runner_settings(&place.workspace, place.repo.as_ref().map(\|r\| r.id.as_str())).await?;
845	self.audit(vec![entry(
846	AuditActor::of(&a.actor),
847	"update_runner_settings",
848	&place.workspace,
849	place.repo.as_ref().map(\|r\| format!("{}/{}", r.namespace, r.name)),
850	None,
851	format!(
852	"Agents on self-hosted runners: {}; labels {}; pull requests from forks: {}.",
853	if settings.agents_on_self_hosted { "on" } else { "off" },
854	settings.agent_labels.join(", "),
855	if settings.fork_pull_requests { "allowed" } else { "not allowed" }
856	),
857	)])
858	.await;
859	Ok(Outcome::Ok(settings))
860	}
861
862	// --- The runner's side ---------------------------------------------------
863
864	/// `runner_register`.
865	pub async fn runner_register(&self, a: RegisterArgs) -> Result<Outcome<Registered>> {
866	let refused = \|\| fail(FailureCode::Unauthenticated, "That registration token is not valid, or it expired. Make a new one under Settings, Runners.");
867	if !a.token.starts_with(model::REGISTRATION_PREFIX) {
868	return Ok(refused());
869	}
870	let registration = self
871	.db
872	.prepare("SELECT * FROM runner_registrations WHERE token_hash = ?")
873	.bind(&[hash(&a.token).into()])?
874	.first::<RegistrationRow>(None)
875	.await?;
876	let at = now_ms();
877	let Some(registration) = registration.filter(\|r\| r.expires_at.as_str() > rfc3339(at).as_str()) else {
878	return Ok(refused());
879	};
880	let name = match valid_name(&a.name) {
881	Ok(name) => name,
882	Err(problem) => return Ok(fail(FailureCode::Invalid, problem)),
883	};
884	let (Some(os), Some(arch)) = (model::os_label(&a.os), model::arch_label(&a.arch)) else {
885	return Ok(fail(FailureCode::Invalid, "A runner's os is linux, macos or windows, and its arch x64 or arm64."));
886	};
887	let labels = match model::runner_labels(&a.labels, os, arch) {
888	Ok(labels) => labels,
889	Err(problem) => return Ok(fail(FailureCode::Invalid, problem)),
890	};
891	let workspace = registration.workspace.clone();
892	// Abuse: so many machines, so fast, are not a team's.
893	let recent = self
894	.db
895	.prepare("SELECT COUNT(*) AS n FROM runners WHERE workspace = ? AND created_at > ?")
896	.bind(&[workspace.as_str().into(), rfc3339(at.saturating_sub(60 * 1000)).into()])?
897	.first::<Count>(None)
898	.await?
899	.map_or(0, \|c\| c.n);
900	if recent >= model::MAX_REGISTRATIONS_PER_MINUTE {
901	return Ok(fail(FailureCode::Conflict, "Too many runners registered in the last minute. Wait a moment and try again."));
902	}
903	let total = self
904	.db
905	.prepare("SELECT COUNT(*) AS n FROM runners WHERE workspace = ?")
906	.bind(&[workspace.as_str().into()])?
907	.first::<Count>(None)
908	.await?
909	.map_or(0, \|c\| c.n);
910	// A group: the one asked for, the token's, or the default.
911	let group = if registration.repo_id.is_some() {
912	None
913	} else {
914	match a.group.as_deref().map(str::trim).filter(\|g\| !g.is_empty()) {
915	Some(wanted) => match self.find_group(&workspace, wanted).await? {
916	Some(group) => Some(group.id),
917	None => return Ok(fail(FailureCode::NotFound, format!("There is no runner group called {wanted}."))),
918	},
919	None => match registration.group_id.clone() {
920	Some(id) => Some(id),
921	None => Some(self.default_group(&workspace).await?.id),
922	},
923	}
924	};
925	let existing = self
926	.db
927	.prepare("SELECT * FROM runners WHERE workspace = ? AND COALESCE(repo_id, '') = ? AND lower(name) = lower(?)")
928	.bind(&[workspace.as_str().into(), registration.repo_id.clone().unwrap_or_default().into(), name.as_str().into()])?
929	.first::<RunnerRow>(None)
930	.await?;
931	if let Some(existing) = &existing {
932	if !a.replace {
933	return Ok(fail(
934	FailureCode::Conflict,
935	format!("A runner called {name} is already registered here. Register with --replace to take its place, or choose another --name."),
936	));
937	}
938	self.forget_runner(existing, &format!("The runner {name} was replaced by a new registration.")).await?;
939	} else if total >= model::MAX_RUNNERS {
940	return Ok(fail(FailureCode::Conflict, format!("A workspace has at most {} self-hosted runners.", model::MAX_RUNNERS)));
941	}
942	// How it keeps work from its machine, if it said (older runners do not).
943	let isolation = a.isolation.as_deref().filter(\|level\| model::ISOLATIONS.contains(level));
944	let id = new_id("rnr", at);
945	let credential = format!("{}{}", model::CREDENTIAL_PREFIX, random_hex(32));
946	let created = rfc3339(at);
947	self.db
948	.prepare(
949	"INSERT INTO runners (id, workspace, repo_id, repo, group_id, name, labels, os, arch, version, ephemeral, credential_hash,
950	rotated_at, last_seen_at, created_at, created_by, isolation)
951	VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)",
952	)
953	.bind(&[
954	id.as_str().into(),
955	workspace.as_str().into(),
956	optional(registration.repo_id.as_deref()),
957	optional(registration.repo.as_deref()),
958	optional(group.as_deref()),
959	name.as_str().into(),
960	serde_json::to_string(&labels)?.into(),
961	os.into(),
962	arch.into(),
963	a.version.chars().take(40).collect::<String>().into(),
964	u32::from(a.ephemeral).into(),
965	hash(&credential).into(),
966	created.as_str().into(),
967	created.as_str().into(),
968	created.as_str().into(),
969	// The person who made the token: a desktop's runner is theirs.
970	optional(registration.created_by.as_deref()),
971	optional(isolation),
972	])?
973	.run()
974	.await?;
975	self.db
976	.prepare("UPDATE runner_registrations SET used = used + 1 WHERE token_hash = ?")
977	.bind(&[hash(&a.token).into()])?
978	.run()
979	.await?;
980	let row = self.runner_by_id(&id).await?.ok_or_else(\|\| worker::Error::RustError("the runner was not kept".into()))?;
981	self.audit(vec![entry(
982	runner_actor(&row),
983	"register_runner",
984	&workspace,
985	row.repo.clone(),
986	Some(name.clone()),
987	format!("Registered the self-hosted runner {name} ({os}, {arch}; labels {}).", labels.join(", ")),
988	)])
989	.await;
990	let groups = self.group_rows(&workspace).await?;
991	Ok(Outcome::Ok(Registered {
992	runner: self.runner_view(&row, &groups, at).await?,
993	credential,
994	}))
995	}
996
997	/// The runner a call is from, if its credential is that runner's (or
998	/// the one before a rotation, until the new one is used).
999	async fn authenticated(&self, auth: &RunnerAuth) -> Result<Option<(RunnerRow, bool)>> {
1000	if !auth.credential.starts_with(model::CREDENTIAL_PREFIX) {
1001	return Ok(None);
1002	}
1003	let Some(row) = self.runner_by_id(&auth.runner).await? else { return Ok(None) };
1004	let given = hash(&auth.credential);
1005	if g1t_secrets::same(&row.credential_hash, &given) {
1006	if row.previous_hash.is_some() {
1007	self.db.prepare("UPDATE runners SET previous_hash = NULL WHERE id = ?").bind(&[row.id.as_str().into()])?.run().await?;
1008	}
1009	return Ok(Some((row, true)));
1010	}
1011	if row.previous_hash.as_deref().is_some_and(\|previous\| g1t_secrets::same(previous, &given)) {
1012	return Ok(Some((row, false)));
1013	}
1014	Ok(None)
1015	}
1016
1017	fn unknown_runner<T>() -> Outcome<T> {
1018	fail(FailureCode::Unauthenticated, "This runner is not registered, or its credential is not valid. Register it again.")
1019	}
1020
1021	/// `runner_poll`.
1022	pub async fn runner_poll(&self, a: PollArgs) -> Result<Outcome<Poll>> {
1023	let Some((row, current)) = self.authenticated(&a.auth).await? else {
1024	return Ok(Self::unknown_runner());
1025	};
1026	let at = now_ms();
1027	let seen = rfc3339(at);
1028	// A new credential every day, handed over on a poll that used the
1029	// current one.
1030	let mut credential = None;
1031	let rotate_after = rfc3339(at.saturating_sub(model::ROTATE_AFTER_MS));
1032	if current && row.rotated_at.as_str() < rotate_after.as_str() {
1033	let fresh = format!("{}{}", model::CREDENTIAL_PREFIX, random_hex(32));
1034	self.db
1035	.prepare("UPDATE runners SET previous_hash = credential_hash, credential_hash = ?, rotated_at = ? WHERE id = ?")
1036	.bind(&[hash(&fresh).into(), seen.as_str().into(), row.id.as_str().into()])?
1037	.run()
1038	.await?;
1039	credential = Some(fresh);
1040	}
1041	// What it holds that it should stop, and the heartbeat for the rest.
1042	let mut cancel = Vec::new();
1043	let mut active: Option<(String, &'static str)> = None;
1044	for id in a.running.iter().take(10) {
1045	if let Some(job) = self.db.prepare("SELECT * FROM jobs WHERE id = ?").bind(&[id.as_str().into()])?.first::<JobRow>(None).await? {
1046	if job.status == "in_progress" && job.runner_id.as_deref() == Some(row.id.as_str()) {
1047	if seen_is_stale(job.seen_at.as_deref(), at) {
1048	self.db.prepare("UPDATE jobs SET seen_at = ? WHERE id = ?").bind(&[seen.as_str().into(), id.as_str().into()])?.run().await?;
1049	}
1050	active = Some((id.clone(), "workflow"));
1051	} else {
1052	cancel.push(id.clone());
1053	}
1054	continue;
1055	}
1056	match self.db.prepare("SELECT * FROM runner_tasks WHERE id = ?").bind(&[id.as_str().into()])?.first::<TaskRow>(None).await? {
1057	Some(task) if task.status == "in_progress" && task.runner_id.as_deref() == Some(row.id.as_str()) => {
1058	if seen_is_stale(task.seen_at.as_deref(), at) {
1059	self.db.prepare("UPDATE runner_tasks SET seen_at = ? WHERE id = ?").bind(&[seen.as_str().into(), id.as_str().into()])?.run().await?;
1060	}
1061	active = Some((id.clone(), "agent"));
1062	}
1063	_ => cancel.push(id.clone()),
1064	}
1065	}
1066	// Written when something changed, or once a minute to keep it online.
1067	let version: String = if a.version.is_empty() { row.version.clone() } else { a.version.chars().take(40).collect() };
1068	let work_id = active.as_ref().map(\|(id, _)\| id.as_str());
1069	let work_kind = active.as_ref().map(\|(_, kind)\| *kind);
1070	// How it is isolated, as it says now: a runner re-registered or
1071	// restarted with or without Docker says so on its next poll.
1072	let isolation = a
1073	.isolation
1074	.as_deref()
1075	.filter(\|level\| model::ISOLATIONS.contains(level))
1076	.map(str::to_owned)
1077	.or_else(\|\| row.isolation.clone());
1078	let changed = version != row.version
1079	\|\| work_id != row.work_id.as_deref()
1080	\|\| work_kind != row.work_kind.as_deref()
1081	\|\| isolation != row.isolation;
1082	if changed \|\| seen_is_stale(row.last_seen_at.as_deref(), at) {
1083	self.db
1084	.prepare("UPDATE runners SET last_seen_at = ?, version = ?, work_id = ?, work_kind = ?, isolation = ? WHERE id = ?")
1085	.bind(&[
1086	seen.as_str().into(),
1087	version.into(),
1088	optional(work_id),
1089	optional(work_kind),
1090	optional(isolation.as_deref()),
1091	row.id.as_str().into(),
1092	])?
1093	.run()
1094	.await?;
1095	}
1096	let row = RunnerRow { isolation, ..row };
1097	let mut poll = Poll { assignment: None, cancel, credential, removed: false };
1098	// Busy, or an ephemeral runner that has had its one job.
1099	if active.is_some() \|\| !a.running.is_empty() {
1100	return Ok(Outcome::Ok(poll));
1101	}
1102	if row.ephemeral != 0 && row.spent != 0 {
1103	self.forget_runner(&row, "The ephemeral runner finished its job.").await?;
1104	poll.removed = true;
1105	return Ok(Outcome::Ok(poll));
1106	}
1107	let groups = self.group_rows(&row.workspace).await?;
1108	let group = row.group_id.as_ref().and_then(\|id\| groups.iter().find(\|g\| &g.id == id)).cloned();
1109	let deadline = at + a.wait_ms.min(model::MAX_POLL_WAIT_MS);
1110	let mut looked = Looked::default();
1111	loop {
1112	if let Some(assignment) = self.claim(&row, group.as_ref(), &mut looked).await? {
1113	poll.assignment = Some(assignment);
1114	return Ok(Outcome::Ok(poll));
1115	}
1116	if now_ms() + POLL_EVERY_MS > deadline {
1117	return Ok(Outcome::Ok(poll));
1118	}
1119	worker::Delay::from(std::time::Duration::from_millis(POLL_EVERY_MS)).await;
1120	}
1121	}
1122
1123	/// Takes the oldest work this runner can do, if there is any.
1124	async fn claim(&self, runner: &RunnerRow, group: Option<&GroupRow>, looked: &mut Looked) -> Result<Option<Assignment>> {
1125	let labels = runner.labels();
1126	let group_name = group.map(\|g\| g.name.clone());
1127	// Workflow jobs.
1128	let mut sql = "SELECT jobs.*, runs.repo AS run_repo FROM jobs JOIN runs ON runs.id = jobs.run_id
1129	WHERE jobs.status = 'queued' AND jobs.labels IS NOT NULL AND lower(jobs.namespace) = ?"
1130	.to_owned();
1131	let mut binds: Vec<JsValue> = vec![runner.workspace.as_str().into()];
1132	if let Some(repo_id) = &runner.repo_id {
1133	sql.push_str(" AND jobs.repo_id = ?");
1134	binds.push(repo_id.as_str().into());
1135	}
1136	// One job of a concurrency group runs at a time.
1137	sql.push_str(
1138	" AND (jobs.concurrency_group IS NULL OR NOT EXISTS (SELECT 1 FROM jobs o WHERE o.repo_id = jobs.repo_id
1139	AND o.concurrency_group = jobs.concurrency_group AND o.status IN ('in_progress', 'calling')))",
1140	);
1141	sql.push_str(" ORDER BY jobs.rowid LIMIT 50");
1142	#[derive(Deserialize)]
1143	struct Queued {
1144	#[serde(flatten)]
1145	job: JobRow,
1146	run_repo: String,
1147	}
1148	let queued = self.db.prepare(sql).bind(&binds)?.all().await?.results::<Queued>()?;
1149	for Queued { job, run_repo } in queued {
1150	let stored: Vec<String> = job.labels.as_deref().and_then(\|l\| serde_json::from_str(l).ok()).unwrap_or_default();
1151	let wanted = Wanted::from_stored(&stored);
1152	if !wanted.matches(&labels, group_name.as_deref()) {
1153	continue;
1154	}
1155	if runner.repo_id.is_none() && !group.is_none_or(\|g\| g.allows(&run_repo)) {
1156	continue;
1157	}
1158	// A host-mode runner takes no public repository's job: it waits
1159	// for a sandboxed runner, and says so when none could take it.
1160	if runner.isolation.as_deref() == Some("host") {
1161	let visibility = self.visibility_once(looked, &job.repo_id).await;
1162	if !model::isolation_allows(runner.isolation.as_deref(), visibility.as_deref()) {
1163	if looked.noted.insert(job.id.clone()) {
1164	self.wait_for_sandbox(&job, &run_repo, &wanted).await?;
1165	}
1166	continue;
1167	}
1168	}
1169	if let Some(max) = job.max_parallel {
1170	let siblings = self
1171	.db
1172	.prepare("SELECT COUNT(*) AS n FROM jobs WHERE run_id = ? AND key = ? AND status = 'in_progress'")
1173	.bind(&[job.run_id.as_str().into(), job.key.as_str().into()])?
1174	.first::<Count>(None)
1175	.await?
1176	.map_or(0, \|c\| c.n);
1177	if siblings >= max {
1178	continue;
1179	}
1180	}
1181	let token = random_hex(24);
1182	let at = now();
1183	let claimed = self
1184	.db
1185	.prepare(
1186	"UPDATE jobs SET status = 'in_progress', token_hash = ?, runner_id = ?, runner_name = ?, reason = NULL, started_at = ?, seen_at = ?
1187	WHERE id = ? AND status = 'queued' RETURNING id",
1188	)
1189	.bind(&[
1190	sha256_hex(&token).into(),
1191	runner.id.as_str().into(),
1192	runner.name.as_str().into(),
1193	at.as_str().into(),
1194	at.as_str().into(),
1195	job.id.as_str().into(),
1196	])?
1197	.first::<Value>(None)
1198	.await?;
1199	if claimed.is_none() {
1200	continue;
1201	}
1202	self.db
1203	.batch(vec![
1204	self.db
1205	.prepare("UPDATE runs SET status = 'in_progress', started_at = COALESCE(started_at, ?) WHERE id = ? AND status = 'queued'")
1206	.bind(&[at.as_str().into(), job.run_id.as_str().into()])?,
1207	self.db
1208	.prepare("UPDATE runners SET work_id = ?, work_kind = 'workflow', spent = ephemeral WHERE id = ?")
1209	.bind(&[job.id.as_str().into(), runner.id.as_str().into()])?,
1210	])
1211	.await?;
1212	self.audit(vec![entry(
1213	runner_actor(runner),
1214	"runner.take_job",
1215	&runner.workspace,
1216	Some(run_repo.clone()),
1217	Some(job.name.clone()),
1218	format!("The self-hosted runner {} took the job {}.", runner.name, job.name),
1219	)])
1220	.await;
1221	// A job that deploys: its run's deployment is under way.
1222	self.job_started(&job.id).await?;
1223	let image = self.job_image(&job).await?;
1224	let visibility = self.visibility(&job.repo_id).await;
1225	return Ok(Some(Assignment {
1226	kind: "workflow".into(),
1227	id: job.id.clone(),
1228	name: job.name.clone(),
1229	repo: run_repo,
1230	timeout_minutes: job.timeout_minutes,
1231	image,
1232	token: Some(token),
1233	env: None,
1234	visibility,
1235	}));
1236	}
1237
1238	// Agent work.
1239	let tasks = self
1240	.db
1241	.prepare("SELECT * FROM runner_tasks WHERE status = 'queued' AND workspace = ? ORDER BY created_at LIMIT 20")
1242	.bind(&[runner.workspace.as_str().into()])?
1243	.all()
1244	.await?
1245	.results::<TaskRow>()?;
1246	for task in tasks {
1247	let wanted = Wanted::of(&serde_json::from_str::<Vec<String>>(&task.labels).unwrap_or_default(), None);
1248	if !wanted.matches(&labels, None) {
1249	continue;
1250	}
1251	// Nor a public repository's agent work.
1252	if runner.isolation.as_deref() == Some("host")
1253	&& let Some(repo_id) = &task.repo_id
1254	{
1255	let visibility = self.visibility_once(looked, repo_id).await;
1256	if !model::isolation_allows(runner.isolation.as_deref(), visibility.as_deref()) {
1257	continue;
1258	}
1259	}
1260	if let Some(repo_id) = &runner.repo_id
1261	&& task.repo_id.as_deref() != Some(repo_id.as_str())
1262	{
1263	continue;
1264	}
1265	if runner.repo_id.is_none() && !group.is_none_or(\|g\| g.allows(&task.repo)) {
1266	continue;
1267	}
1268	let Some(sealed) = task.env.clone() else { continue };
1269	let at = now();
1270	let claimed = self
1271	.db
1272	.prepare(
1273	"UPDATE runner_tasks SET status = 'in_progress', runner_id = ?, runner_name = ?, env = NULL, started_at = ?, seen_at = ?
1274	WHERE id = ? AND status = 'queued' RETURNING id",
1275	)
1276	.bind(&[runner.id.as_str().into(), runner.name.as_str().into(), at.as_str().into(), at.as_str().into(), task.id.as_str().into()])?
1277	.first::<Value>(None)
1278	.await?;
1279	if claimed.is_none() {
1280	continue;
1281	}
1282	self.db
1283	.prepare("UPDATE runners SET work_id = ?, work_kind = 'agent', spent = ephemeral WHERE id = ?")
1284	.bind(&[task.id.as_str().into(), runner.id.as_str().into()])?
1285	.run()
1286	.await?;
1287	let env: Map<String, Value> = self
1288	.sealer
1289	.as_ref()
1290	.and_then(\|sealer\| sealer.open(&sealed, &task.id))
1291	.and_then(\|text\| serde_json::from_str(&text).ok())
1292	.unwrap_or_default();
1293	if env.is_empty() {
1294	self.end_task(&TaskRow { status: "in_progress".into(), ..task.clone() }, 1, "The work's environment could not be opened.").await?;
1295	continue;
1296	}
1297	self.audit(vec![entry(
1298	runner_actor(runner),
1299	"runner.take_task",
1300	&runner.workspace,
1301	Some(task.repo.clone()),
1302	Some(task.kind.clone()),
1303	format!("The self-hosted runner {} took agent work: {}.", runner.name, task.title),
1304	)])
1305	.await;
1306	let visibility = match &task.repo_id {
1307	Some(repo_id) => self.visibility(repo_id).await,
1308	None => None,
1309	};
1310	return Ok(Some(Assignment {
1311	kind: "agent".into(),
1312	id: task.id.clone(),
1313	name: task.title.clone(),
1314	repo: task.repo.clone(),
1315	timeout_minutes: task.timeout_minutes,
1316	image: None,
1317	token: None,
1318	env: Some(env),
1319	visibility,
1320	}));
1321	}
1322	Ok(None)
1323	}
1324
1325	/// [`Self::visibility`], asked once per repository in a poll.
1326	async fn visibility_once(&self, looked: &mut Looked, repo_id: &str) -> Option<String> {
1327	if let Some(known) = looked.visibilities.get(repo_id) {
1328	return known.clone();
1329	}
1330	let found = self.visibility(repo_id).await;
1331	looked.visibilities.insert(repo_id.to_owned(), found.clone());
1332	found
1333	}
1334
1335	/// A public repository's job that a host-mode runner passed over: when
1336	/// no online runner that may take it could (its labels, group,
1337	/// repository, and isolation), its page says it waits for a sandboxed
1338	/// one. Written only when that changes what it says.
1339	async fn wait_for_sandbox(&self, job: &JobRow, run_repo: &str, wanted: &Wanted) -> Result<()> {
1340	let at = now_ms();
1341	let groups = self.group_rows(&job.namespace.to_lowercase()).await?;
1342	let runners = self.runner_rows(&job.namespace.to_lowercase()).await?;
1343	let sandboxed = runners.iter().any(\|other\| takes_public_job(other, &groups, &job.repo_id, run_repo, wanted, at));
1344	if sandboxed {
1345	return Ok(());
1346	}
1347	self.db
1348	.prepare("UPDATE jobs SET reason = ? WHERE id = ? AND status = 'queued' AND reason IS NOT ?")
1349	.bind(&[model::WAITING_FOR_SANDBOX.into(), job.id.as_str().into(), model::WAITING_FOR_SANDBOX.into()])?
1350	.run()
1351	.await?;
1352	Ok(())
1353	}
1354
1355	/// A repository's visibility for a runner, `public` or `private`; none
1356	/// when it cannot be read, and the runner decides without it.
1357	async fn visibility(&self, repo_id: &str) -> Option<String> {
1358	match self.repo_by_id(repo_id).await {
1359	Ok(Some((repo, _))) => Some(if repo.is_private { "private" } else { "public" }.to_owned()),
1360	_ => None,
1361	}
1362	}
1363
1364	/// The image a job's `container:` names, when it names one plainly.
1365	async fn job_image(&self, job: &JobRow) -> Result<Option<String>> {
1366	let Some(run) = self.run_row(&job.run_id).await? else { return Ok(None) };
1367	let raw = match job.callee() {
1368	Some((_, spec, _)) => spec.raw,
1369	None => match g1t_actions::workflow::parse(&run.source) {
1370	Ok(workflow) => match workflow.jobs.into_iter().find(\|j\| j.id == job.key) {
1371	Some(spec) => spec.raw,
1372	None => return Ok(None),
1373	},
1374	Err(_) => return Ok(None),
1375	},
1376	};
1377	let image = match raw.get("container") {
1378	Some(Value::String(image)) => Some(image.clone()),
1379	Some(Value::Object(container)) => container.get("image").and_then(Value::as_str).map(str::to_owned),
1380	_ => None,
1381	};
1382	Ok(image.filter(\|image\| !image.contains("${{") && !image.trim().is_empty()))
1383	}
1384
1385	/// `runner_finished`.
1386	pub async fn runner_finished(&self, a: FinishedArgs) -> Result<Outcome<bool>> {
1387	let Some((row, _)) = self.authenticated(&a.auth).await? else {
1388	return Ok(Self::unknown_runner());
1389	};
1390	let reason = a.reason.clone().filter(\|r\| !r.trim().is_empty()).map(\|r\| r.chars().take(2000).collect::<String>());
1391	let mut changed = false;
1392	if let Some(job) = self.db.prepare("SELECT * FROM jobs WHERE id = ?").bind(&[a.id.as_str().into()])?.first::<JobRow>(None).await? {
1393	if job.runner_id.as_deref() == Some(row.id.as_str()) && job.status == "in_progress" {
1394	let why = reason.unwrap_or_else(\|\| format!("The job's process on {} exited with code {} before it reported how the job went.", row.name, a.exit_code));
1395	self.finish_job(&job.id, "failure", Some(&why), None).await?;
1396	changed = true;
1397	}
1398	} else if let Some(task) = self.db.prepare("SELECT * FROM runner_tasks WHERE id = ?").bind(&[a.id.as_str().into()])?.first::<TaskRow>(None).await?
1399	&& task.runner_id.as_deref() == Some(row.id.as_str())
1400	&& task.status == "in_progress"
1401	{
1402	let why = reason.unwrap_or_else(\|\| if a.exit_code == 0 { String::new() } else { format!("The work exited with code {} on {}.", a.exit_code, row.name) });
1403	self.end_task(&task, a.exit_code, &why).await?;
1404	changed = true;
1405	}
1406	self.db
1407	.prepare("UPDATE runners SET work_id = NULL, work_kind = NULL, last_seen_at = ? WHERE id = ? AND work_id = ?")
1408	.bind(&[now().into(), row.id.as_str().into(), a.id.as_str().into()])?
1409	.run()
1410	.await?;
1411	Ok(Outcome::Ok(changed))
1412	}
1413
1414	/// `runner_remove_self`.
1415	pub async fn runner_remove_self(&self, a: RemoveSelfArgs) -> Result<Outcome<bool>> {
1416	let Some((row, _)) = self.authenticated(&a.auth).await? else {
1417	return Ok(Self::unknown_runner());
1418	};
1419	self.forget_runner(&row, &format!("The runner {} was removed from its machine.", row.name)).await?;
1420	self.audit(vec![entry(
1421	runner_actor(&row),
1422	"remove_runner",
1423	&row.workspace,
1424	row.repo.clone(),
1425	Some(row.name.clone()),
1426	format!("The self-hosted runner {} removed itself.", row.name),
1427	)])
1428	.await;
1429	Ok(Outcome::Ok(true))
1430	}
1431
1432	/// A self-hosted job finished, however it did: its runner is free, and
1433	/// its time goes on the workspace's usage as self-hosted, at $0.
1434	pub async fn released(&self, job: &JobRow) -> Result<()> {
1435	let Some(runner_id) = &job.runner_id else { return Ok(()) };
1436	self.db
1437	.prepare("UPDATE runners SET work_id = NULL, work_kind = NULL WHERE id = ? AND work_id = ?")
1438	.bind(&[runner_id.as_str().into(), job.id.as_str().into()])?
1439	.run()
1440	.await?;
1441	let (Some(started), Some(finished)) = (job.started_at.as_deref(), job.finished_at.as_deref().map(str::to_owned).or_else(\|\| Some(now()))) else {
1442	return Ok(());
1443	};
1444	let seconds = match (g1t_contracts::time::parse_rfc3339(started), g1t_contracts::time::parse_rfc3339(&finished)) {
1445	(Some(a), Some(b)) if b > a => ((b - a) / 1000).max(1),
1446	_ => 1,
1447	};
1448	let run = self.run_row(&job.run_id).await?;
1449	let repo = run.as_ref().map(\|r\| r.repo.clone());
1450	let recorded: Result<Outcome<bool>> = g1t_kit::call(
1451	&self.billing,
1452	"record_sandbox",
1453	&RecordSandboxArgs {
1454	workspace: job.namespace.to_lowercase(),
1455	seconds: u32::try_from(seconds).unwrap_or(u32::MAX),
1456	description: format!(
1457	"{} in {} on the self-hosted runner {}",
1458	job.name,
1459	repo.clone().unwrap_or_default(),
1460	job.runner_name.clone().unwrap_or_default()
1461	),
1462	repo,
1463	reference: format!("selfhosted/{}/{}", job.id, started),
1464	kind: Some(ComputeKind::Workflow),
1465	cpu_seconds: None,
1466	// A workflow job on a self-hosted runner: no agent ran it and nobody asked an agent for it.
1467	agent: None,
1468	asked_by: None,
1469	reservation_id: None,
1470	self_hosted: true,
1471	instance: None,
1472	},
1473	)
1474	.await;
1475	if let Err(error) = recorded {
1476	worker::console_error!("actions: self-hosted time not recorded for {}: {error}", job.id);
1477	}
1478	Ok(())
1479	}
1480
1481	/// `runner` and `RUNNER_*` for a job on a self-hosted runner.
1482	pub async fn runner_context_for(&self, runner_id: &str, variables: &mut Map<String, Value>) -> Result<Value> {
1483	let mut context = g1t_actions::events::runner_context();
1484	let Some(runner) = self.runner_by_id(runner_id).await? else { return Ok(context) };
1485	let os = match runner.os.as_str() {
1486	"macos" => "macOS",
1487	"windows" => "Windows",
1488	_ => "Linux",
1489	};
1490	let arch = if runner.arch == "arm64" { "ARM64" } else { "X64" };
1491	context["name"] = json!(runner.name);
1492	context["os"] = json!(os);
1493	context["arch"] = json!(arch);
1494	context["environment"] = json!("self-hosted");
1495	{
1496	let vars = variables;
1497	vars.insert("RUNNER_NAME".into(), json!(runner.name));
1498	vars.insert("RUNNER_OS".into(), json!(os));
1499	vars.insert("RUNNER_ARCH".into(), json!(arch));
1500	vars.insert("RUNNER_ENVIRONMENT".into(), json!("self-hosted"));
1501	}
1502	Ok(context)
1503	}
1504
1505	// --- Agent work, for the runner service ----------------------------------
1506
1507	/// `runner_route`: the labels g1t's own work in `repo` runs on, when
1508	/// its workspace (or the repository) sends it to self-hosted runners.
1509	pub async fn runner_route(&self, a: RouteArgs) -> Result<Option<Vec<String>>> {
1510	let repo_id = match self.repo_by_path(&a.repo).await? {
1511	Some(repo) => Some(repo.id),
1512	None => None,
1513	};
1514	let settings = self.effective_runner_settings(&a.workspace, repo_id.as_deref()).await?;
1515	Ok(settings.agents_on_self_hosted.then_some(settings.agent_labels))
1516	}
1517
1518	async fn repo_by_path(&self, path: &RepoPath) -> Result<Option<Repo>> {
1519	let Some(actor) = self.workspace_actor(&path.namespace).await? else { return Ok(None) };
1520	self.visible_repo(path, &Some(actor)).await
1521	}
1522
1523	/// `enqueue_task`.
1524	pub async fn enqueue_task(&self, a: EnqueueTaskArgs) -> Result<Outcome<String>> {
1525	let Some(sealer) = &self.sealer else {
1526	return Ok(fail(FailureCode::Conflict, "Agent work cannot be handed to self-hosted runners yet: g1t's key for it is not set."));
1527	};
1528	let id = new_id("rtk", now_ms());
1529	let sealed = sealer.seal(&serde_json::to_string(&a.env)?, &id);
1530	let repo_id = self.repo_by_path(&a.repo).await?.map(\|r\| r.id);
1531	let labels = match model::runner_labels(&a.labels, model::SELF_HOSTED, model::SELF_HOSTED) {
1532	Ok(labels) => labels,
1533	Err(problem) => return Ok(fail(FailureCode::Invalid, problem)),
1534	};
1535	let at = now();
1536	let inserted = self
1537	.db
1538	.prepare(
1539	"INSERT OR IGNORE INTO runner_tasks (id, sandbox, workspace, repo_id, repo, kind, title, labels, env, timeout_minutes, status, created_at, run_id)
1540	VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, 'queued', ?, ?) RETURNING id",
1541	)
1542	.bind(&[
1543	id.as_str().into(),
1544	a.sandbox.as_str().into(),
1545	a.workspace.to_lowercase().into(),
1546	optional(repo_id.as_deref()),
1547	format!("{}/{}", a.repo.namespace, a.repo.name).into(),
1548	a.kind.as_str().into(),
1549	a.title.chars().take(300).collect::<String>().into(),
1550	serde_json::to_string(&labels)?.into(),
1551	sealed.into(),
1552	a.timeout_minutes.max(1).into(),
1553	at.into(),
1554	optional(a.run_id.as_deref().filter(\|id\| !id.is_empty())),
1555	])?
1556	.first::<Value>(None)
1557	.await?;
1558	if inserted.is_none() {
1559	return Ok(fail(FailureCode::Conflict, "That sandbox already handed work to a self-hosted runner."));
1560	}
1561	Ok(Outcome::Ok(id))
1562	}
1563
1564	/// `cancel_task`: the sandbox stopped it. Its runner hears on its next
1565	/// poll; the sandbox does its own ending.
1566	pub async fn cancel_task(&self, a: CancelTaskArgs) -> Result<Outcome<bool>> {
1567	let done = self
1568	.db
1569	.prepare(
1570	"UPDATE runner_tasks SET status = 'completed', exit_code = -1, reason = ?, env = NULL, finished_at = ?
1571	WHERE sandbox = ? AND status != 'completed' RETURNING id",
1572	)
1573	.bind(&[optional(a.reason.as_deref()), now().into(), a.sandbox.as_str().into()])?
1574	.first::<Value>(None)
1575	.await?;
1576	Ok(Outcome::Ok(done.is_some()))
1577	}
1578
1579	/// Ends a task and tells the sandbox waiting on it, once.
1580	async fn end_task(&self, task: &TaskRow, exit_code: i32, reason: &str) -> Result<()> {
1581	let ended = self
1582	.db
1583	.prepare(
1584	"UPDATE runner_tasks SET status = 'completed', exit_code = ?, reason = ?, env = NULL, finished_at = ?
1585	WHERE id = ? AND status != 'completed' RETURNING id",
1586	)
1587	.bind(&[exit_code.into(), optional(Some(reason).filter(\|r\| !r.is_empty())), now().into(), task.id.as_str().into()])?
1588	.first::<Value>(None)
1589	.await?;
1590	if ended.is_none() {
1591	return Ok(());
1592	}
1593	if let Some(runner_id) = &task.runner_id {
1594	self.db
1595	.prepare("UPDATE runners SET work_id = NULL, work_kind = NULL WHERE id = ? AND work_id = ?")
1596	.bind(&[runner_id.as_str().into(), task.id.as_str().into()])?
1597	.run()
1598	.await?;
1599	}
1600	let told: Result<Value> = g1t_kit::call(
1601	&self.runner,
1602	"task_ended",
1603	&json!({
1604	"sandbox": task.sandbox,
1605	"exitCode": exit_code,
1606	"reason": if reason.is_empty() { Value::Null } else { json!(reason) },
1607	"runner": task.runner_name,
1608	}),
1609	)
1610	.await;
1611	if let Err(error) = told {
1612	worker::console_error!("actions: the sandbox waiting on task {} was not told: {error}", task.id);
1613	}
1614	Ok(())
1615	}
1616
1617	/// `runner_activity`: what runs for the workspace now, in g1t's
1618	/// sandboxes and handed to its own runners. Its owners only, as its
1619	/// runners are.
1620	pub async fn runner_activity(&self, a: RunnerActivityArgs) -> Result<Outcome<RunnerActivity>> {
1621	let owner = RunnersOwner { repo: None, workspace: Some(a.workspace.clone()) };
1622	let place = check!(self.runner_place(&a.actor, &owner, false).await?);
1623	let workspace = place.workspace;
1624	let cloud_jobs = self
1625	.db
1626	.prepare(
1627	"SELECT jobs.id, jobs.name, jobs.run_id, runs.repo, jobs.started_at FROM jobs JOIN runs ON runs.id = jobs.run_id
1628	WHERE lower(jobs.namespace) = ? AND jobs.status = 'in_progress' AND jobs.labels IS NULL
1629	ORDER BY jobs.started_at DESC LIMIT ?",
1630	)
1631	.bind(&[workspace.as_str().into(), ACTIVITY_LIMIT.into()])?
1632	.all()
1633	.await?
1634	.results::<CloudJob>()?;
1635	let cloud_jobs_queued = self.queued_jobs(&workspace, false).await?;
1636	let self_hosted_jobs_queued = self.queued_jobs(&workspace, true).await?;
1637	let handed_over = self
1638	.db
1639	.prepare(
1640	"SELECT id, run_id, kind, title, repo, status, runner_name, created_at, started_at FROM runner_tasks
1641	WHERE status IN ('queued', 'in_progress') AND workspace = ?
1642	ORDER BY created_at DESC LIMIT ?",
1643	)
1644	.bind(&[workspace.as_str().into(), ACTIVITY_LIMIT.into()])?
1645	.all()
1646	.await?
1647	.results::<HandedOverTask>()?;
1648	Ok(Outcome::Ok(RunnerActivity { cloud_jobs, cloud_jobs_queued, handed_over, self_hosted_jobs_queued }))
1649	}
1650
1651	/// The workspace's workflow jobs queued for its own runners
1652	/// (`labelled`), or for g1t's sandboxes.
1653	async fn queued_jobs(&self, workspace: &str, labelled: bool) -> Result<u32> {
1654	let sql = if labelled {
1655	"SELECT COUNT(*) AS n FROM jobs WHERE lower(namespace) = ? AND status = 'queued' AND labels IS NOT NULL"
1656	} else {
1657	"SELECT COUNT(*) AS n FROM jobs WHERE lower(namespace) = ? AND status = 'queued' AND labels IS NULL"
1658	};
1659	Ok(self.db.prepare(sql).bind(&[workspace.into()])?.first::<Count>(None).await?.map_or(0, \|c\| c.n))
1660	}
1661
1662	/// `stuck_jobs`: jobs of the viewer's workspaces that have waited ten
1663	/// minutes or more for a self-hosted runner while none that could take
1664	/// them is online, for Mission control's Needs you.
1665	pub async fn stuck_jobs(&self, a: StuckJobsArgs) -> Result<Vec<StuckJob>> {
1666	let Some(viewer) = a.viewer else { return Ok(Vec::new()) };
1667	let at = now_ms();
1668	let mut out = Vec::new();
1669	for membership in viewer.workspaces.iter().take(20) {
1670	let workspace = membership.slug.to_lowercase();
1671	#[derive(Deserialize)]
1672	struct Waiting {
1673	id: String,
1674	name: String,
1675	run_id: String,
1676	labels: String,
1677	queued_at: Option<String>,
1678	repo: String,
1679	repo_id: String,
1680	}
1681	let waiting = self
1682	.db
1683	.prepare(
1684	"SELECT jobs.id, jobs.name, jobs.run_id, jobs.labels, jobs.queued_at, runs.repo, jobs.repo_id FROM jobs JOIN runs ON runs.id = jobs.run_id
1685	WHERE jobs.status = 'queued' AND jobs.labels IS NOT NULL AND lower(jobs.namespace) = ? AND jobs.queued_at < ?
1686	ORDER BY jobs.rowid LIMIT 20",
1687	)
1688	.bind(&[workspace.as_str().into(), rfc3339(at.saturating_sub(10 * 60 * 1000)).into()])?
1689	.all()
1690	.await?
1691	.results::<Waiting>()?;
1692	if waiting.is_empty() {
1693	continue;
1694	}
1695	let groups = self.group_rows(&workspace).await?;
1696	let online: Vec<RunnerRow> = self.runner_rows(&workspace).await?.into_iter().filter(\|r\| r.online(at)).collect();
1697	// A repository's visibility is asked only when a host-mode runner
1698	// is online, since only those pass over a public repository's jobs.
1699	let any_host = online.iter().any(\|runner\| runner.isolation.as_deref() == Some("host"));
1700	let mut looked = Looked::default();
1701	for job in waiting {
1702	let wanted = Wanted::from_stored(&serde_json::from_str::<Vec<String>>(&job.labels).unwrap_or_default());
1703	let visibility = if any_host { self.visibility_once(&mut looked, &job.repo_id).await } else { None };
1704	let served = online.iter().any(\|runner\| {
1705	let group = runner.group_id.as_ref().and_then(\|id\| groups.iter().find(\|g\| &g.id == id));
1706	wanted.matches(&runner.labels(), group.map(\|g\| g.name.as_str()))
1707	&& model::isolation_allows(runner.isolation.as_deref(), visibility.as_deref())
1708	});
1709	if !served {
1710	out.push(StuckJob {
1711	id: job.id,
1712	name: job.name,
1713	run_id: job.run_id,
1714	repo: job.repo,
1715	labels: wanted.describe(),
1716	queued_at: job.queued_at.unwrap_or_default(),
1717	});
1718	}
1719	}
1720	}
1721	Ok(out)
1722	}
1723
1724	// --- Every minute ----------------------------------------------------------
1725
1726	/// Jobs and tasks that waited too long, tasks whose runner went quiet,
1727	/// runners offline for weeks, and registration tokens long expired.
1728	pub async fn sweep_runners(&self, at: u64) -> Result<()> {
1729	let before = \|ms: u64\| rfc3339(at.saturating_sub(ms));
1730	let waited = model::MAX_WAIT_HOURS * 60 * 60 * 1000;
1731	let stale = self
1732	.db
1733	.prepare("SELECT * FROM jobs WHERE status = 'queued' AND labels IS NOT NULL AND queued_at < ? LIMIT 50")
1734	.bind(&[before(waited).into()])?
1735	.all()
1736	.await?
1737	.results::<JobRow>()?;
1738	for job in stale {
1739	let stored: Vec<String> = job.labels.as_deref().and_then(\|l\| serde_json::from_str(l).ok()).unwrap_or_default();
1740	let why = format!(
1741	"No self-hosted runner with labels {} took it within {} hours.",
1742	Wanted::from_stored(&stored).describe(),
1743	model::MAX_WAIT_HOURS
1744	);
1745	self.finish_job(&job.id, "failure", Some(&why), None).await?;
1746	}
1747	let tasks = self
1748	.db
1749	.prepare(
1750	"SELECT * FROM runner_tasks WHERE (status = 'queued' AND created_at < ?) OR (status = 'in_progress' AND seen_at < ?) LIMIT 50",
1751	)
1752	.bind(&[before(waited).into(), before(SILENT_MS).into()])?
1753	.all()
1754	.await?
1755	.results::<TaskRow>()?;
1756	for task in tasks {
1757	let why = if task.status == "queued" {
1758	format!("No self-hosted runner with labels {} took it within {} hours.", Wanted::of(&serde_json::from_str::<Vec<String>>(&task.labels).unwrap_or_default(), None).describe(), model::MAX_WAIT_HOURS)
1759	} else {
1760	format!("The self-hosted runner {} stopped answering.", task.runner_name.clone().unwrap_or_default())
1761	};
1762	self.end_task(&task, 1, &why).await?;
1763	}
1764	let forgotten = self
1765	.db
1766	.prepare("SELECT * FROM runners WHERE COALESCE(last_seen_at, created_at) < ? LIMIT 50")
1767	.bind(&[before(FORGET_OFFLINE_MS).into()])?
1768	.all()
1769	.await?
1770	.results::<RunnerRow>()?;
1771	for row in forgotten {
1772	self.forget_runner(&row, "The runner was offline for 14 days.").await?;
1773	}
1774	// An ephemeral runner that took its job and then went quiet.
1775	let spent = self
1776	.db
1777	.prepare("SELECT * FROM runners WHERE ephemeral = 1 AND spent = 1 AND work_id IS NULL AND COALESCE(last_seen_at, created_at) < ? LIMIT 50")
1778	.bind(&[before(60 * 60 * 1000).into()])?
1779	.all()
1780	.await?
1781	.results::<RunnerRow>()?;
1782	for row in spent {
1783	self.forget_runner(&row, "The ephemeral runner finished its job.").await?;
1784	}
1785	self.db
1786	.prepare("DELETE FROM runner_registrations WHERE expires_at < ?")
1787	.bind(&[before(24 * 60 * 60 * 1000).into()])?
1788	.run()
1789	.await?;
1790	Ok(())
1791	}
1792	}
1793
1794	#[derive(Deserialize)]
1795	struct IdRow {
1796	id: String,
1797	}
1798
1799	/// Whether `runner` is online and could take a public repository's job:
1800	/// not host-mode, and its repository, group and labels let it.
1801	fn takes_public_job(runner: &RunnerRow, groups: &[GroupRow], repo_id: &str, run_repo: &str, wanted: &Wanted, at: u64) -> bool {
1802	let group = runner.group_id.as_ref().and_then(\|id\| groups.iter().find(\|g\| &g.id == id));
1803	runner.online(at)
1804	&& model::isolation_allows(runner.isolation.as_deref(), Some("public"))
1805	&& runner.repo_id.as_deref().is_none_or(\|repo\| repo == repo_id)
1806	&& (runner.repo_id.is_some() \|\| group.is_none_or(\|g\| g.allows(run_repo)))
1807	&& wanted.matches(&runner.labels(), group.map(\|g\| g.name.as_str()))
1808	}
1809
1810	fn group_view(group: &GroupRow, runners: &[RunnerRow]) -> RunnerGroup {
1811	RunnerGroup {
1812	id: group.id.clone(),
1813	name: group.name.clone(),
1814	default: group.is_default != 0,
1815	repositories: group.repositories(),
1816	runners: runners.iter().filter(\|r\| r.group_id.as_deref() == Some(group.id.as_str())).count() as u32,
1817	updated_at: group.updated_at.clone(),
1818	}
1819	}
1820
1821	#[cfg(test)]
1822	mod tests {
1823	use super::*;
1824
1825	fn group(repositories: &[&str]) -> GroupRow {
1826	GroupRow {
1827	id: "rng_1".into(),
1828	name: "Default".into(),
1829	is_default: 1,
1830	repositories: serde_json::to_string(&repositories).unwrap(),
1831	updated_at: "2026-10-06T00:00:00Z".into(),
1832	}
1833	}
1834
1835	#[test]
1836	fn being_seen_is_written_once_a_minute() {
1837	let at = 2_000_000_000_000;
1838	assert!(seen_is_stale(None, at));
1839	assert!(!seen_is_stale(Some(&rfc3339(at - 59_000)), at));
1840	assert!(seen_is_stale(Some(&rfc3339(at - 61_000)), at));
1841	// A runner written down at the last moment is still online until the
1842	// next write, a poll later.
1843	assert!(SEEN_RESOLUTION_MS + model::MAX_POLL_WAIT_MS < model::ONLINE_WITHIN_MS);
1844	assert!(SEEN_RESOLUTION_MS < crate::SILENT_MS);
1845	}
1846
1847	#[test]
1848	fn a_group_lets_its_repositories_use_it() {
1849	assert!(group(&[]).allows("acme/web"));
1850	assert!(group(&["web", "api"]).allows("acme/WEB"));
1851	assert!(group(&["web"]).allows("web"));
1852	assert!(!group(&["web"]).allows("acme/docs"));
1853	}
1854
1855	#[test]
1856	fn names_are_checked() {
1857	assert_eq!(valid_name(" build-01 ").unwrap(), "build-01");
1858	assert!(valid_name("").is_err());
1859	assert!(valid_name("has space").is_err());
1860	assert!(valid_name(&"x".repeat(65)).is_err());
1861	assert!(valid_group_name("GPU machines").is_ok());
1862	assert!(valid_group_name("").is_err());
1863	}
1864
1865	#[test]
1866	fn only_a_sandboxed_runner_that_could_take_it_serves_a_public_job() {
1867	let at = 1_000_000_000;
1868	let runner = \|isolation: Option<&str>\| RunnerRow {
1869	id: "rnr_1".into(),
1870	workspace: "acme".into(),
1871	repo_id: None,
1872	repo: None,
1873	group_id: None,
1874	name: "a".into(),
1875	labels: r#"["self-hosted","linux","x64"]"#.into(),
1876	os: "linux".into(),
1877	arch: "x64".into(),
1878	version: String::new(),
1879	ephemeral: 0,
1880	credential_hash: String::new(),
1881	previous_hash: None,
1882	rotated_at: String::new(),
1883	work_id: None,
1884	work_kind: None,
1885	spent: 0,
1886	last_seen_at: Some(rfc3339(at)),
1887	created_at: String::new(),
1888	created_by: None,
1889	isolation: isolation.map(str::to_owned),
1890	};
1891	let wanted = Wanted::of(&["self-hosted".to_owned(), "linux".to_owned()], None);
1892	let takes = \|row: &RunnerRow\| takes_public_job(row, &[], "rep_1", "acme/web", &wanted, at);
1893	assert!(!takes(&runner(Some("host"))));
1894	assert!(takes(&runner(Some("container"))));
1895	assert!(takes(&runner(Some("nested-container"))));
1896	assert!(takes(&runner(None)), "an older runner, which refuses for itself");
1897	assert!(!takes(&RunnerRow { repo_id: Some("rep_2".into()), ..runner(None) }), "another repository's");
1898	assert!(!takes(&RunnerRow { labels: r#"["self-hosted","macos"]"#.into(), ..runner(None) }));
1899	assert!(!takes(&RunnerRow { last_seen_at: None, ..runner(None) }), "offline");
1900	}
1901
1902	#[test]
1903	fn online_means_seen_in_the_last_ninety_seconds() {
1904	let row = RunnerRow {
1905	id: "rnr_1".into(),
1906	workspace: "acme".into(),
1907	repo_id: None,
1908	repo: None,
1909	group_id: None,
1910	name: "a".into(),
1911	labels: "[]".into(),
1912	os: "linux".into(),
1913	arch: "x64".into(),
1914	version: String::new(),
1915	ephemeral: 0,
1916	credential_hash: String::new(),
1917	previous_hash: None,
1918	rotated_at: String::new(),
1919	work_id: None,
1920	work_kind: None,
1921	spent: 0,
1922	last_seen_at: Some(rfc3339(1_000_000_000)),
1923	created_at: String::new(),
1924	created_by: None,
1925	isolation: None,
1926	};
1927	assert!(row.online(1_000_000_000 + 60_000));
1928	assert!(!row.online(1_000_000_000 + 120_000));
1929	assert_eq!(row.workspace, "acme");
1930	}
1931	}