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	isolation: row.isolation.clone(),
395	})
396	}
397
398	// --- People's side -----------------------------------------------------
399
400	/// `runners`: a workspace's runners, or a repository's: its own and
401	/// the workspace's that its group lets it use.
402	pub async fn runners(&self, a: ListRunnersArgs) -> Result<Outcome<Vec<Runner>>> {
403	let place = check!(self.runner_place(&a.actor, &a.owner, false).await?);
404	self.default_group(&place.workspace).await?;
405	let groups = self.group_rows(&place.workspace).await?;
406	let at = now_ms();
407	let mut out = Vec::new();
408	for row in self.runner_rows(&place.workspace).await? {
409	let shown = match &place.repo {
410	None => row.repo_id.is_none(),
411	Some(repo) => match &row.repo_id {
412	Some(id) => id == &repo.id,
413	None => groups.iter().find(\|g\| Some(&g.id) == row.group_id.as_ref()).is_none_or(\|g\| g.allows(&repo.name)),
414	},
415	};
416	if shown {
417	out.push(self.runner_view(&row, &groups, at).await?);
418	}
419	}
420	Ok(Outcome::Ok(out))
421	}
422
423	/// `create_registration_token`.
424	pub async fn create_registration_token(&self, a: CreateRegistrationTokenArgs) -> Result<Outcome<RegistrationToken>> {
425	let place = check!(self.runner_place(&a.actor, &a.owner, true).await?);
426	let made = self
427	.db
428	.prepare("SELECT COUNT(*) AS n FROM runner_registrations WHERE workspace = ? AND created_at > ?")
429	.bind(&[place.workspace.as_str().into(), rfc3339(now_ms().saturating_sub(60 * 60 * 1000)).into()])?
430	.first::<Count>(None)
431	.await?
432	.map_or(0, \|c\| c.n);
433	if made >= model::MAX_TOKENS_PER_HOUR {
434	return Ok(fail(FailureCode::Conflict, "Too many registration tokens in the last hour. Each lasts an hour and registers any number of runners."));
435	}
436	let group = match (&place.repo, a.group.as_deref().map(str::trim).filter(\|g\| !g.is_empty())) {
437	(Some(_), Some(_)) => return Ok(fail(FailureCode::Invalid, "A repository's own runners are in no group.")),
438	(Some(_), None) => None,
439	(None, Some(wanted)) => match self.find_group(&place.workspace, wanted).await? {
440	Some(group) => Some(group),
441	None => return Ok(fail(FailureCode::NotFound, format!("There is no runner group called {wanted}."))),
442	},
443	(None, None) => Some(self.default_group(&place.workspace).await?),
444	};
445	let token = format!("{}{}", model::REGISTRATION_PREFIX, random_hex(24));
446	let at = now_ms();
447	let expires_at = rfc3339(at + model::REGISTRATION_TTL_SECONDS * 1000);
448	let repo_name = place.repo.as_ref().map(\|r\| format!("{}/{}", r.namespace, r.name));
449	self.db
450	.prepare(
451	"INSERT INTO runner_registrations (token_hash, workspace, repo_id, repo, group_id, created_by, created_at, expires_at)
452	VALUES (?, ?, ?, ?, ?, ?, ?, ?)",
453	)
454	.bind(&[
455	hash(&token).into(),
456	place.workspace.as_str().into(),
457	optional(place.repo.as_ref().map(\|r\| r.id.as_str())),
458	optional(repo_name.as_deref()),
459	optional(group.as_ref().map(\|g\| g.id.as_str())),
460	a.actor.username.as_str().into(),
461	rfc3339(at).into(),
462	expires_at.as_str().into(),
463	])?
464	.run()
465	.await?;
466	self.audit(vec![entry(
467	AuditActor::of(&a.actor),
468	"create_runner_registration_token",
469	&place.workspace,
470	repo_name.clone(),
471	None,
472	format!("Made a runner registration token{}.", group.as_ref().map(\|g\| format!(" for group {}", g.name)).unwrap_or_default()),
473	)])
474	.await;
475	Ok(Outcome::Ok(RegistrationToken {
476	token,
477	expires_at,
478	workspace: place.workspace,
479	repo: repo_name,
480	group: group.map(\|g\| g.name),
481	url: crate::SITE.to_owned(),
482	}))
483	}
484
485	/// `remove_runner`: a person removing one.
486	pub async fn remove_runner(&self, a: RemoveRunnerArgs) -> Result<Outcome<bool>> {
487	let place = check!(self.runner_place(&a.actor, &a.owner, true).await?);
488	let Some(row) = self.runner_by_id(&a.id).await? else {
489	return Ok(fail(FailureCode::NotFound, "There is no such runner."));
490	};
491	let ours = row.workspace == place.workspace
492	&& match &place.repo {
493	Some(repo) => row.repo_id.as_deref() == Some(repo.id.as_str()),
494	None => row.repo_id.is_none(),
495	};
496	if !ours {
497	return Ok(fail(FailureCode::NotFound, "There is no such runner."));
498	}
499	self.forget_runner(&row, &format!("{} removed the runner {}.", a.actor.username, row.name)).await?;
500	self.audit(vec![entry(
501	AuditActor::of(&a.actor),
502	"remove_runner",
503	&row.workspace,
504	row.repo.clone(),
505	Some(row.name.clone()),
506	format!("Removed the self-hosted runner {}.", row.name),
507	)])
508	.await;
509	Ok(Outcome::Ok(true))
510	}
511
512	/// Removes a runner: what it was running fails, and its credential
513	/// stops working at once.
514	async fn forget_runner(&self, row: &RunnerRow, why: &str) -> Result<()> {
515	self.db.prepare("DELETE FROM runners WHERE id = ?").bind(&[row.id.as_str().into()])?.run().await?;
516	self.abandon_work(&row.id, why).await
517	}
518
519	/// Fails whatever a runner was running.
520	async fn abandon_work(&self, runner_id: &str, why: &str) -> Result<()> {
521	let jobs = self
522	.db
523	.prepare("SELECT id FROM jobs WHERE runner_id = ? AND status = 'in_progress'")
524	.bind(&[runner_id.into()])?
525	.all()
526	.await?
527	.results::<IdRow>()?;
528	for job in jobs {
529	self.finish_job(&job.id, "failure", Some(why), None).await?;
530	}
531	let tasks = self
532	.db
533	.prepare("SELECT * FROM runner_tasks WHERE runner_id = ? AND status = 'in_progress'")
534	.bind(&[runner_id.into()])?
535	.all()
536	.await?
537	.results::<TaskRow>()?;
538	for task in tasks {
539	self.end_task(&task, 1, why).await?;
540	}
541	Ok(())
542	}
543
544	/// `runner_groups`.
545	pub async fn runner_groups(&self, a: RunnerGroupsArgs) -> Result<Outcome<Vec<RunnerGroup>>> {
546	let owner = RunnersOwner { repo: None, workspace: Some(a.workspace.clone()) };
547	let place = check!(self.runner_place(&a.actor, &owner, false).await?);
548	self.default_group(&place.workspace).await?;
549	let groups = self.group_rows(&place.workspace).await?;
550	let runners = self.runner_rows(&place.workspace).await?;
551	Ok(Outcome::Ok(groups.iter().map(\|g\| group_view(g, &runners)).collect()))
552	}
553
554	/// `set_runner_group`.
555	pub async fn set_runner_group(&self, a: SetRunnerGroupArgs) -> Result<Outcome<RunnerGroup>> {
556	let owner = RunnersOwner { repo: None, workspace: Some(a.workspace.clone()) };
557	let place = check!(self.runner_place(&a.actor, &owner, true).await?);
558	self.default_group(&place.workspace).await?;
559	let groups = self.group_rows(&place.workspace).await?;
560	let name = match a.name.as_deref().map(valid_group_name).transpose() {
561	Ok(name) => name,
562	Err(problem) => return Ok(fail(FailureCode::Invalid, problem)),
563	};
564	let repositories: Option<Vec<String>> = a.repositories.map(\|list\| {
565	let mut out: Vec<String> = Vec::new();
566	for repo in list {
567	let name = repo.trim().rsplit('/').next().unwrap_or_default().to_owned();
568	if !name.is_empty() && !out.iter().any(\|r: &String\| r.eq_ignore_ascii_case(&name)) {
569	out.push(name);
570	}
571	}
572	out
573	});
574	if let Some(name) = &name
575	&& groups.iter().any(\|g\| g.name.eq_ignore_ascii_case(name) && Some(&g.id) != a.id.as_ref())
576	{
577	return Ok(fail(FailureCode::Conflict, format!("There is already a group called {name}.")));
578	}
579	let at = now();
580	let id = match &a.id {
581	Some(id) => {
582	let Some(group) = groups.iter().find(\|g\| &g.id == id) else {
583	return Ok(fail(FailureCode::NotFound, "There is no such runner group."));
584	};
585	self.db
586	.prepare("UPDATE runner_groups SET name = ?, repositories = ?, updated_at = ? WHERE id = ?")
587	.bind(&[
588	name.clone().unwrap_or(group.name.clone()).into(),
589	serde_json::to_string(&repositories.clone().unwrap_or(group.repositories()))?.into(),
590	at.as_str().into(),
591	id.as_str().into(),
592	])?
593	.run()
594	.await?;
595	id.clone()
596	}
597	None => {
598	let Some(name) = name.clone() else {
599	return Ok(fail(FailureCode::Invalid, "A new group needs a name."));
600	};
601	if groups.len() >= 50 {
602	return Ok(fail(FailureCode::Conflict, "A workspace has at most 50 runner groups."));
603	}
604	let id = new_id("rng", now_ms());
605	self.db
606	.prepare(
607	"INSERT INTO runner_groups (id, workspace, name, is_default, repositories, created_at, updated_at) VALUES (?, ?, ?, 0, ?, ?, ?)",
608	)
609	.bind(&[
610	id.as_str().into(),
611	place.workspace.as_str().into(),
612	name.into(),
613	serde_json::to_string(&repositories.clone().unwrap_or_default())?.into(),
614	at.as_str().into(),
615	at.as_str().into(),
616	])?
617	.run()
618	.await?;
619	id
620	}
621	};
622	let groups = self.group_rows(&place.workspace).await?;
623	let runners = self.runner_rows(&place.workspace).await?;
624	let Some(group) = groups.iter().find(\|g\| g.id == id) else {
625	return Ok(fail(FailureCode::NotFound, "There is no such runner group."));
626	};
627	self.audit(vec![entry(
628	AuditActor::of(&a.actor),
629	if a.id.is_some() { "update_runner_group" } else { "create_runner_group" },
630	&place.workspace,
631	None,
632	Some(group.name.clone()),
633	format!(
634	"Runner group {}: {}.",
635	group.name,
636	if group.repositories().is_empty() { "every repository".to_owned() } else { group.repositories().join(", ") }
637	),
638	)])
639	.await;
640	Ok(Outcome::Ok(group_view(group, &runners)))
641	}
642
643	/// `delete_runner_group`.
644	pub async fn delete_runner_group(&self, a: DeleteRunnerGroupArgs) -> Result<Outcome<bool>> {
645	let owner = RunnersOwner { repo: None, workspace: Some(a.workspace.clone()) };
646	let place = check!(self.runner_place(&a.actor, &owner, true).await?);
647	let default = self.default_group(&place.workspace).await?;
648	let Some(group) = self.group_rows(&place.workspace).await?.into_iter().find(\|g\| g.id == a.id) else {
649	return Ok(fail(FailureCode::NotFound, "There is no such runner group."));
650	};
651	if group.is_default != 0 {
652	return Ok(fail(FailureCode::Conflict, "The default group cannot be deleted."));
653	}
654	self.db
655	.batch(vec![
656	self.db.prepare("UPDATE runners SET group_id = ? WHERE group_id = ?").bind(&[default.id.as_str().into(), group.id.as_str().into()])?,
657	self.db.prepare("DELETE FROM runner_groups WHERE id = ?").bind(&[group.id.as_str().into()])?,
658	])
659	.await?;
660	self.audit(vec![entry(
661	AuditActor::of(&a.actor),
662	"delete_runner_group",
663	&place.workspace,
664	None,
665	Some(group.name.clone()),
666	format!("Deleted the runner group {}; its runners joined {}.", group.name, default.name),
667	)])
668	.await;
669	Ok(Outcome::Ok(true))
670	}
671
672	async fn settings_row(&self, owner: &str) -> Result<Option<SettingsRow>> {
673	self.db.prepare("SELECT * FROM runner_settings WHERE owner = ?").bind(&[owner.into()])?.first::<SettingsRow>(None).await
674	}
675
676	fn settings_of(row: Option<SettingsRow>, inherited: bool) -> RunnerSettings {
677	match row {
678	Some(row) => RunnerSettings {
679	agents_on_self_hosted: row.agents != 0,
680	agent_labels: serde_json::from_str(&row.agent_labels).unwrap_or_else(\|_\| vec![model::SELF_HOSTED.to_owned()]),
681	fork_pull_requests: row.fork_pulls != 0,
682	inherited,
683	},
684	None => RunnerSettings {
685	agent_labels: vec![model::SELF_HOSTED.to_owned()],
686	inherited,
687	..RunnerSettings::default()
688	},
689	}
690	}
691
692	/// The settings that apply in a workspace, or in one of its
693	/// repositories: the repository's own if it has them.
694	pub async fn effective_runner_settings(&self, workspace: &str, repo_id: Option<&str>) -> Result<RunnerSettings> {
695	if let Some(repo_id) = repo_id
696	&& let Some(own) = self.settings_row(repo_id).await?
697	{
698	return Ok(Self::settings_of(Some(own), false));
699	}
700	let inherited = repo_id.is_some();
701	Ok(Self::settings_of(self.settings_row(&workspace.to_lowercase()).await?, inherited))
702	}
703
704	/// `runner_settings`.
705	pub async fn runner_settings(&self, a: RunnerSettingsArgs) -> Result<Outcome<RunnerSettings>> {
706	let place = check!(self.runner_place(&a.actor, &a.owner, false).await?);
707	Ok(Outcome::Ok(self.effective_runner_settings(&place.workspace, place.repo.as_ref().map(\|r\| r.id.as_str())).await?))
708	}
709
710	/// `report_desktop`: what a person's desktop app says of its runner,
711	/// in a workspace they own. Kept per computer, replaced each time.
712	pub async fn report_desktop(&self, a: ReportDesktopArgs) -> Result<Outcome<bool>> {
713	let owner = RunnersOwner { repo: None, workspace: Some(a.workspace.clone()) };
714	let place = check!(self.runner_place(&a.actor, &owner, true).await?);
715	let machine = match valid_name(&a.machine) {
716	Ok(name) => name,
717	Err(problem) => return Ok(fail(FailureCode::Invalid, problem)),
718	};
719	if !matches!(a.backend.as_str(), "sandboxed" \| "docker" \| "host") {
720	return Ok(fail(FailureCode::Invalid, "A desktop runs jobs sandboxed, in Docker or on the host."));
721	}
722	let clip = \|text: &str, most: usize\| text.chars().take(most).collect::<String>();
723	let tidy = \|d: &DesktopDiagnosis\| DesktopDiagnosis {
724	code: clip(&d.code, 40),
725	tone: clip(&d.tone, 20),
726	title: clip(&d.title, 200),
727	cause: d.cause.as_deref().map(\|t\| clip(t, 1000)),
728	fix: d.fix.as_deref().map(\|t\| clip(t, 1000)),
729	};
730	let at = now_ms();
731	let known = self
732	.db
733	.prepare("SELECT COUNT(*) AS n FROM desktop_runners WHERE workspace = ? AND lower(machine) <> lower(?)")
734	.bind(&[place.workspace.as_str().into(), machine.as_str().into()])?
735	.first::<Count>(None)
736	.await?
737	.map_or(0, \|c\| c.n);
738	if known >= model::MAX_DESKTOPS {
739	// The oldest report makes room: a computer long gone.
740	self.db
741	.prepare("DELETE FROM desktop_runners WHERE rowid IN (SELECT rowid FROM desktop_runners WHERE workspace = ? ORDER BY reported_at LIMIT 1)")
742	.bind(&[place.workspace.as_str().into()])?
743	.run()
744	.await?;
745	}
746	self.db
747	.prepare(
748	"INSERT INTO desktop_runners (workspace, machine, person, backend, version, diagnosis, closed, reported_at)
749	VALUES (?, ?, ?, ?, ?, ?, ?, ?)
750	ON CONFLICT (workspace, machine) DO UPDATE SET person = ?3, backend = ?4, version = ?5, diagnosis = ?6, closed = ?7, reported_at = ?8",
751	)
752	.bind(&[
753	place.workspace.as_str().into(),
754	machine.as_str().into(),
755	a.actor.username.as_str().into(),
756	a.backend.as_str().into(),
757	optional(a.version.as_deref().map(\|v\| clip(v, 40)).as_deref()),
758	serde_json::to_string(&tidy(&a.diagnosis))?.into(),
759	serde_json::to_string(&tidy(&a.closed))?.into(),
760	rfc3339(at).into(),
761	])?
762	.run()
763	.await?;
764	Ok(Outcome::Ok(true))
765	}
766
767	/// `desktop_runners`: the workspace's desktops, as each app last said.
768	pub async fn desktop_runners(&self, a: DesktopRunnersArgs) -> Result<Outcome<Vec<DesktopRunner>>> {
769	let owner = RunnersOwner { repo: None, workspace: Some(a.workspace.clone()) };
770	let place = check!(self.runner_place(&a.actor, &owner, false).await?);
771	#[derive(Deserialize)]
772	struct Row {
773	workspace: String,
774	machine: String,
775	person: String,
776	backend: String,
777	version: Option<String>,
778	diagnosis: String,
779	closed: String,
780	reported_at: String,
781	}
782	let rows = self
783	.db
784	.prepare("SELECT * FROM desktop_runners WHERE workspace = ? ORDER BY lower(machine)")
785	.bind(&[place.workspace.as_str().into()])?
786	.all()
787	.await?
788	.results::<Row>()?;
789	let open_since = rfc3339(now_ms().saturating_sub(model::DESKTOP_OPEN_WITHIN_MS));
790	Ok(Outcome::Ok(
791	rows.into_iter()
792	.map(\|row\| DesktopRunner {
793	open: row.reported_at.as_str() > open_since.as_str(),
794	diagnosis: serde_json::from_str(&row.diagnosis).unwrap_or_default(),
795	closed: serde_json::from_str(&row.closed).unwrap_or_default(),
796	workspace: row.workspace,
797	machine: row.machine,
798	person: row.person,
799	backend: row.backend,
800	version: row.version,
801	reported_at: row.reported_at,
802	})
803	.collect(),
804	))
805	}
806
807	/// `set_runner_settings`.
808	pub async fn set_runner_settings(&self, a: SetRunnerSettingsArgs) -> Result<Outcome<RunnerSettings>> {
809	let place = check!(self.runner_place(&a.actor, &a.owner, true).await?);
810	let (owner, scope) = match &place.repo {
811	Some(repo) => (repo.id.clone(), "repository"),
812	None => (place.workspace.clone(), "workspace"),
813	};
814	if a.inherit && place.repo.is_some() {
815	self.db.prepare("DELETE FROM runner_settings WHERE owner = ?").bind(&[owner.as_str().into()])?.run().await?;
816	} else {
817	let current = self.effective_runner_settings(&place.workspace, place.repo.as_ref().map(\|r\| r.id.as_str())).await?;
818	let labels = match &a.agent_labels {
819	Some(given) => match model::runner_labels(given, model::SELF_HOSTED, model::SELF_HOSTED) {
820	Ok(labels) => labels,
821	Err(problem) => return Ok(fail(FailureCode::Invalid, problem)),
822	},
823	None => current.agent_labels.clone(),
824	};
825	let agents = a.agents_on_self_hosted.unwrap_or(current.agents_on_self_hosted);
826	let forks = a.fork_pull_requests.unwrap_or(current.fork_pull_requests);
827	self.db
828	.prepare(
829	"INSERT INTO runner_settings (owner, scope, agents, agent_labels, fork_pulls, updated_at, updated_by)
830	VALUES (?1, ?2, ?3, ?4, ?5, ?6, ?7)
831	ON CONFLICT (owner) DO UPDATE SET agents = ?3, agent_labels = ?4, fork_pulls = ?5, updated_at = ?6, updated_by = ?7",
832	)
833	.bind(&[
834	owner.as_str().into(),
835	scope.into(),
836	u32::from(agents).into(),
837	serde_json::to_string(&labels)?.into(),
838	u32::from(forks).into(),
839	now().into(),
840	a.actor.username.as_str().into(),
841	])?
842	.run()
843	.await?;
844	}
845	let settings = self.effective_runner_settings(&place.workspace, place.repo.as_ref().map(\|r\| r.id.as_str())).await?;
846	self.audit(vec![entry(
847	AuditActor::of(&a.actor),
848	"update_runner_settings",
849	&place.workspace,
850	place.repo.as_ref().map(\|r\| format!("{}/{}", r.namespace, r.name)),
851	None,
852	format!(
853	"Agents on self-hosted runners: {}; labels {}; pull requests from forks: {}.",
854	if settings.agents_on_self_hosted { "on" } else { "off" },
855	settings.agent_labels.join(", "),
856	if settings.fork_pull_requests { "allowed" } else { "not allowed" }
857	),
858	)])
859	.await;
860	Ok(Outcome::Ok(settings))
861	}
862
863	// --- The runner's side ---------------------------------------------------
864
865	/// `runner_register`.
866	pub async fn runner_register(&self, a: RegisterArgs) -> Result<Outcome<Registered>> {
867	let refused = \|\| fail(FailureCode::Unauthenticated, "That registration token is not valid, or it expired. Make a new one under Settings, Runners.");
868	if !a.token.starts_with(model::REGISTRATION_PREFIX) {
869	return Ok(refused());
870	}
871	let registration = self
872	.db
873	.prepare("SELECT * FROM runner_registrations WHERE token_hash = ?")
874	.bind(&[hash(&a.token).into()])?
875	.first::<RegistrationRow>(None)
876	.await?;
877	let at = now_ms();
878	let Some(registration) = registration.filter(\|r\| r.expires_at.as_str() > rfc3339(at).as_str()) else {
879	return Ok(refused());
880	};
881	let name = match valid_name(&a.name) {
882	Ok(name) => name,
883	Err(problem) => return Ok(fail(FailureCode::Invalid, problem)),
884	};
885	let (Some(os), Some(arch)) = (model::os_label(&a.os), model::arch_label(&a.arch)) else {
886	return Ok(fail(FailureCode::Invalid, "A runner's os is linux, macos or windows, and its arch x64 or arm64."));
887	};
888	let labels = match model::runner_labels(&a.labels, os, arch) {
889	Ok(labels) => labels,
890	Err(problem) => return Ok(fail(FailureCode::Invalid, problem)),
891	};
892	let workspace = registration.workspace.clone();
893	// Abuse: so many machines, so fast, are not a team's.
894	let recent = self
895	.db
896	.prepare("SELECT COUNT(*) AS n FROM runners WHERE workspace = ? AND created_at > ?")
897	.bind(&[workspace.as_str().into(), rfc3339(at.saturating_sub(60 * 1000)).into()])?
898	.first::<Count>(None)
899	.await?
900	.map_or(0, \|c\| c.n);
901	if recent >= model::MAX_REGISTRATIONS_PER_MINUTE {
902	return Ok(fail(FailureCode::Conflict, "Too many runners registered in the last minute. Wait a moment and try again."));
903	}
904	let total = self
905	.db
906	.prepare("SELECT COUNT(*) AS n FROM runners WHERE workspace = ?")
907	.bind(&[workspace.as_str().into()])?
908	.first::<Count>(None)
909	.await?
910	.map_or(0, \|c\| c.n);
911	// A group: the one asked for, the token's, or the default.
912	let group = if registration.repo_id.is_some() {
913	None
914	} else {
915	match a.group.as_deref().map(str::trim).filter(\|g\| !g.is_empty()) {
916	Some(wanted) => match self.find_group(&workspace, wanted).await? {
917	Some(group) => Some(group.id),
918	None => return Ok(fail(FailureCode::NotFound, format!("There is no runner group called {wanted}."))),
919	},
920	None => match registration.group_id.clone() {
921	Some(id) => Some(id),
922	None => Some(self.default_group(&workspace).await?.id),
923	},
924	}
925	};
926	let existing = self
927	.db
928	.prepare("SELECT * FROM runners WHERE workspace = ? AND COALESCE(repo_id, '') = ? AND lower(name) = lower(?)")
929	.bind(&[workspace.as_str().into(), registration.repo_id.clone().unwrap_or_default().into(), name.as_str().into()])?
930	.first::<RunnerRow>(None)
931	.await?;
932	if let Some(existing) = &existing {
933	if !a.replace {
934	return Ok(fail(
935	FailureCode::Conflict,
936	format!("A runner called {name} is already registered here. Register with --replace to take its place, or choose another --name."),
937	));
938	}
939	self.forget_runner(existing, &format!("The runner {name} was replaced by a new registration.")).await?;
940	} else if total >= model::MAX_RUNNERS {
941	return Ok(fail(FailureCode::Conflict, format!("A workspace has at most {} self-hosted runners.", model::MAX_RUNNERS)));
942	}
943	// How it keeps work from its machine, if it said (older runners do not).
944	let isolation = a.isolation.as_deref().filter(\|level\| model::ISOLATIONS.contains(level));
945	let id = new_id("rnr", at);
946	let credential = format!("{}{}", model::CREDENTIAL_PREFIX, random_hex(32));
947	let created = rfc3339(at);
948	self.db
949	.prepare(
950	"INSERT INTO runners (id, workspace, repo_id, repo, group_id, name, labels, os, arch, version, ephemeral, credential_hash,
951	rotated_at, last_seen_at, created_at, created_by, isolation)
952	VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)",
953	)
954	.bind(&[
955	id.as_str().into(),
956	workspace.as_str().into(),
957	optional(registration.repo_id.as_deref()),
958	optional(registration.repo.as_deref()),
959	optional(group.as_deref()),
960	name.as_str().into(),
961	serde_json::to_string(&labels)?.into(),
962	os.into(),
963	arch.into(),
964	a.version.chars().take(40).collect::<String>().into(),
965	u32::from(a.ephemeral).into(),
966	hash(&credential).into(),
967	created.as_str().into(),
968	created.as_str().into(),
969	created.as_str().into(),
970	// The person who made the token: a desktop's runner is theirs.
971	optional(registration.created_by.as_deref()),
972	optional(isolation),
973	])?
974	.run()
975	.await?;
976	self.db
977	.prepare("UPDATE runner_registrations SET used = used + 1 WHERE token_hash = ?")
978	.bind(&[hash(&a.token).into()])?
979	.run()
980	.await?;
981	let row = self.runner_by_id(&id).await?.ok_or_else(\|\| worker::Error::RustError("the runner was not kept".into()))?;
982	self.audit(vec![entry(
983	runner_actor(&row),
984	"register_runner",
985	&workspace,
986	row.repo.clone(),
987	Some(name.clone()),
988	format!("Registered the self-hosted runner {name} ({os}, {arch}; labels {}).", labels.join(", ")),
989	)])
990	.await;
991	let groups = self.group_rows(&workspace).await?;
992	Ok(Outcome::Ok(Registered {
993	runner: self.runner_view(&row, &groups, at).await?,
994	credential,
995	}))
996	}
997
998	/// The runner a call is from, if its credential is that runner's (or
999	/// the one before a rotation, until the new one is used).
1000	async fn authenticated(&self, auth: &RunnerAuth) -> Result<Option<(RunnerRow, bool)>> {
1001	if !auth.credential.starts_with(model::CREDENTIAL_PREFIX) {
1002	return Ok(None);
1003	}
1004	let Some(row) = self.runner_by_id(&auth.runner).await? else { return Ok(None) };
1005	let given = hash(&auth.credential);
1006	if g1t_secrets::same(&row.credential_hash, &given) {
1007	if row.previous_hash.is_some() {
1008	self.db.prepare("UPDATE runners SET previous_hash = NULL WHERE id = ?").bind(&[row.id.as_str().into()])?.run().await?;
1009	}
1010	return Ok(Some((row, true)));
1011	}
1012	if row.previous_hash.as_deref().is_some_and(\|previous\| g1t_secrets::same(previous, &given)) {
1013	return Ok(Some((row, false)));
1014	}
1015	Ok(None)
1016	}
1017
1018	fn unknown_runner<T>() -> Outcome<T> {
1019	fail(FailureCode::Unauthenticated, "This runner is not registered, or its credential is not valid. Register it again.")
1020	}
1021
1022	/// `runner_poll`.
1023	pub async fn runner_poll(&self, a: PollArgs) -> Result<Outcome<Poll>> {
1024	let Some((row, current)) = self.authenticated(&a.auth).await? else {
1025	return Ok(Self::unknown_runner());
1026	};
1027	let at = now_ms();
1028	let seen = rfc3339(at);
1029	// A new credential every day, handed over on a poll that used the
1030	// current one.
1031	let mut credential = None;
1032	let rotate_after = rfc3339(at.saturating_sub(model::ROTATE_AFTER_MS));
1033	if current && row.rotated_at.as_str() < rotate_after.as_str() {
1034	let fresh = format!("{}{}", model::CREDENTIAL_PREFIX, random_hex(32));
1035	self.db
1036	.prepare("UPDATE runners SET previous_hash = credential_hash, credential_hash = ?, rotated_at = ? WHERE id = ?")
1037	.bind(&[hash(&fresh).into(), seen.as_str().into(), row.id.as_str().into()])?
1038	.run()
1039	.await?;
1040	credential = Some(fresh);
1041	}
1042	// What it holds that it should stop, and the heartbeat for the rest.
1043	let mut cancel = Vec::new();
1044	let mut active: Option<(String, &'static str)> = None;
1045	for id in a.running.iter().take(10) {
1046	if let Some(job) = self.db.prepare("SELECT * FROM jobs WHERE id = ?").bind(&[id.as_str().into()])?.first::<JobRow>(None).await? {
1047	if job.status == "in_progress" && job.runner_id.as_deref() == Some(row.id.as_str()) {
1048	if seen_is_stale(job.seen_at.as_deref(), at) {
1049	self.db.prepare("UPDATE jobs SET seen_at = ? WHERE id = ?").bind(&[seen.as_str().into(), id.as_str().into()])?.run().await?;
1050	}
1051	active = Some((id.clone(), "workflow"));
1052	} else {
1053	cancel.push(id.clone());
1054	}
1055	continue;
1056	}
1057	match self.db.prepare("SELECT * FROM runner_tasks WHERE id = ?").bind(&[id.as_str().into()])?.first::<TaskRow>(None).await? {
1058	Some(task) if task.status == "in_progress" && task.runner_id.as_deref() == Some(row.id.as_str()) => {
1059	if seen_is_stale(task.seen_at.as_deref(), at) {
1060	self.db.prepare("UPDATE runner_tasks SET seen_at = ? WHERE id = ?").bind(&[seen.as_str().into(), id.as_str().into()])?.run().await?;
1061	}
1062	active = Some((id.clone(), "agent"));
1063	}
1064	_ => cancel.push(id.clone()),
1065	}
1066	}
1067	// Written when something changed, or once a minute to keep it online.
1068	let version: String = if a.version.is_empty() { row.version.clone() } else { a.version.chars().take(40).collect() };
1069	let work_id = active.as_ref().map(\|(id, _)\| id.as_str());
1070	let work_kind = active.as_ref().map(\|(_, kind)\| *kind);
1071	// How it is isolated, as it says now: a runner re-registered or
1072	// restarted with or without Docker says so on its next poll.
1073	let isolation = a
1074	.isolation
1075	.as_deref()
1076	.filter(\|level\| model::ISOLATIONS.contains(level))
1077	.map(str::to_owned)
1078	.or_else(\|\| row.isolation.clone());
1079	let changed = version != row.version
1080	\|\| work_id != row.work_id.as_deref()
1081	\|\| work_kind != row.work_kind.as_deref()
1082	\|\| isolation != row.isolation;
1083	if changed \|\| seen_is_stale(row.last_seen_at.as_deref(), at) {
1084	self.db
1085	.prepare("UPDATE runners SET last_seen_at = ?, version = ?, work_id = ?, work_kind = ?, isolation = ? WHERE id = ?")
1086	.bind(&[
1087	seen.as_str().into(),
1088	version.into(),
1089	optional(work_id),
1090	optional(work_kind),
1091	optional(isolation.as_deref()),
1092	row.id.as_str().into(),
1093	])?
1094	.run()
1095	.await?;
1096	}
1097	let row = RunnerRow { isolation, ..row };
1098	let mut poll = Poll { assignment: None, cancel, credential, removed: false };
1099	// Busy, or an ephemeral runner that has had its one job.
1100	if active.is_some() \|\| !a.running.is_empty() {
1101	return Ok(Outcome::Ok(poll));
1102	}
1103	if row.ephemeral != 0 && row.spent != 0 {
1104	self.forget_runner(&row, "The ephemeral runner finished its job.").await?;
1105	poll.removed = true;
1106	return Ok(Outcome::Ok(poll));
1107	}
1108	let groups = self.group_rows(&row.workspace).await?;
1109	let group = row.group_id.as_ref().and_then(\|id\| groups.iter().find(\|g\| &g.id == id)).cloned();
1110	let deadline = at + a.wait_ms.min(model::MAX_POLL_WAIT_MS);
1111	let mut looked = Looked::default();
1112	loop {
1113	if let Some(assignment) = self.claim(&row, group.as_ref(), &mut looked).await? {
1114	poll.assignment = Some(assignment);
1115	return Ok(Outcome::Ok(poll));
1116	}
1117	if now_ms() + POLL_EVERY_MS > deadline {
1118	return Ok(Outcome::Ok(poll));
1119	}
1120	worker::Delay::from(std::time::Duration::from_millis(POLL_EVERY_MS)).await;
1121	}
1122	}
1123
1124	/// Takes the oldest work this runner can do, if there is any.
1125	async fn claim(&self, runner: &RunnerRow, group: Option<&GroupRow>, looked: &mut Looked) -> Result<Option<Assignment>> {
1126	let labels = runner.labels();
1127	let group_name = group.map(\|g\| g.name.clone());
1128	// Workflow jobs.
1129	let mut sql = "SELECT jobs.*, runs.repo AS run_repo FROM jobs JOIN runs ON runs.id = jobs.run_id
1130	WHERE jobs.status = 'queued' AND jobs.labels IS NOT NULL AND lower(jobs.namespace) = ?"
1131	.to_owned();
1132	let mut binds: Vec<JsValue> = vec![runner.workspace.as_str().into()];
1133	if let Some(repo_id) = &runner.repo_id {
1134	sql.push_str(" AND jobs.repo_id = ?");
1135	binds.push(repo_id.as_str().into());
1136	}
1137	// One job of a concurrency group runs at a time.
1138	sql.push_str(
1139	" AND (jobs.concurrency_group IS NULL OR NOT EXISTS (SELECT 1 FROM jobs o WHERE o.repo_id = jobs.repo_id
1140	AND o.concurrency_group = jobs.concurrency_group AND o.status IN ('in_progress', 'calling')))",
1141	);
1142	sql.push_str(" ORDER BY jobs.rowid LIMIT 50");
1143	#[derive(Deserialize)]
1144	struct Queued {
1145	#[serde(flatten)]
1146	job: JobRow,
1147	run_repo: String,
1148	}
1149	let queued = self.db.prepare(sql).bind(&binds)?.all().await?.results::<Queued>()?;
1150	for Queued { job, run_repo } in queued {
1151	let stored: Vec<String> = job.labels.as_deref().and_then(\|l\| serde_json::from_str(l).ok()).unwrap_or_default();
1152	let wanted = Wanted::from_stored(&stored);
1153	if !wanted.matches(&labels, group_name.as_deref()) {
1154	continue;
1155	}
1156	if runner.repo_id.is_none() && !group.is_none_or(\|g\| g.allows(&run_repo)) {
1157	continue;
1158	}
1159	// A host-mode runner takes no public repository's job: it waits
1160	// for a sandboxed runner, and says so when none could take it.
1161	if runner.isolation.as_deref() == Some("host") {
1162	let visibility = self.visibility_once(looked, &job.repo_id).await;
1163	if !model::isolation_allows(runner.isolation.as_deref(), visibility.as_deref()) {
1164	if looked.noted.insert(job.id.clone()) {
1165	self.wait_for_sandbox(&job, &run_repo, &wanted).await?;
1166	}
1167	continue;
1168	}
1169	}
1170	if let Some(max) = job.max_parallel {
1171	let siblings = self
1172	.db
1173	.prepare("SELECT COUNT(*) AS n FROM jobs WHERE run_id = ? AND key = ? AND status = 'in_progress'")
1174	.bind(&[job.run_id.as_str().into(), job.key.as_str().into()])?
1175	.first::<Count>(None)
1176	.await?
1177	.map_or(0, \|c\| c.n);
1178	if siblings >= max {
1179	continue;
1180	}
1181	}
1182	let token = random_hex(24);
1183	let at = now();
1184	let claimed = self
1185	.db
1186	.prepare(
1187	"UPDATE jobs SET status = 'in_progress', token_hash = ?, runner_id = ?, runner_name = ?, reason = NULL, started_at = ?, seen_at = ?
1188	WHERE id = ? AND status = 'queued' RETURNING id",
1189	)
1190	.bind(&[
1191	sha256_hex(&token).into(),
1192	runner.id.as_str().into(),
1193	runner.name.as_str().into(),
1194	at.as_str().into(),
1195	at.as_str().into(),
1196	job.id.as_str().into(),
1197	])?
1198	.first::<Value>(None)
1199	.await?;
1200	if claimed.is_none() {
1201	continue;
1202	}
1203	self.db
1204	.batch(vec![
1205	self.db
1206	.prepare("UPDATE runs SET status = 'in_progress', started_at = COALESCE(started_at, ?) WHERE id = ? AND status = 'queued'")
1207	.bind(&[at.as_str().into(), job.run_id.as_str().into()])?,
1208	self.db
1209	.prepare("UPDATE runners SET work_id = ?, work_kind = 'workflow', spent = ephemeral WHERE id = ?")
1210	.bind(&[job.id.as_str().into(), runner.id.as_str().into()])?,
1211	])
1212	.await?;
1213	self.audit(vec![entry(
1214	runner_actor(runner),
1215	"runner.take_job",
1216	&runner.workspace,
1217	Some(run_repo.clone()),
1218	Some(job.name.clone()),
1219	format!("The self-hosted runner {} took the job {}.", runner.name, job.name),
1220	)])
1221	.await;
1222	// A job that deploys: its run's deployment is under way.
1223	self.job_started(&job.id).await?;
1224	let image = self.job_image(&job).await?;
1225	let visibility = self.visibility(&job.repo_id).await;
1226	return Ok(Some(Assignment {
1227	kind: "workflow".into(),
1228	id: job.id.clone(),
1229	name: job.name.clone(),
1230	repo: run_repo,
1231	timeout_minutes: job.timeout_minutes,
1232	image,
1233	token: Some(token),
1234	env: None,
1235	visibility,
1236	}));
1237	}
1238
1239	// Agent work.
1240	let tasks = self
1241	.db
1242	.prepare("SELECT * FROM runner_tasks WHERE status = 'queued' AND workspace = ? ORDER BY created_at LIMIT 20")
1243	.bind(&[runner.workspace.as_str().into()])?
1244	.all()
1245	.await?
1246	.results::<TaskRow>()?;
1247	for task in tasks {
1248	let wanted = Wanted::of(&serde_json::from_str::<Vec<String>>(&task.labels).unwrap_or_default(), None);
1249	if !wanted.matches(&labels, None) {
1250	continue;
1251	}
1252	// Nor a public repository's agent work.
1253	if runner.isolation.as_deref() == Some("host")
1254	&& let Some(repo_id) = &task.repo_id
1255	{
1256	let visibility = self.visibility_once(looked, repo_id).await;
1257	if !model::isolation_allows(runner.isolation.as_deref(), visibility.as_deref()) {
1258	continue;
1259	}
1260	}
1261	if let Some(repo_id) = &runner.repo_id
1262	&& task.repo_id.as_deref() != Some(repo_id.as_str())
1263	{
1264	continue;
1265	}
1266	if runner.repo_id.is_none() && !group.is_none_or(\|g\| g.allows(&task.repo)) {
1267	continue;
1268	}
1269	let Some(sealed) = task.env.clone() else { continue };
1270	let at = now();
1271	let claimed = self
1272	.db
1273	.prepare(
1274	"UPDATE runner_tasks SET status = 'in_progress', runner_id = ?, runner_name = ?, env = NULL, started_at = ?, seen_at = ?
1275	WHERE id = ? AND status = 'queued' RETURNING id",
1276	)
1277	.bind(&[runner.id.as_str().into(), runner.name.as_str().into(), at.as_str().into(), at.as_str().into(), task.id.as_str().into()])?
1278	.first::<Value>(None)
1279	.await?;
1280	if claimed.is_none() {
1281	continue;
1282	}
1283	self.db
1284	.prepare("UPDATE runners SET work_id = ?, work_kind = 'agent', spent = ephemeral WHERE id = ?")
1285	.bind(&[task.id.as_str().into(), runner.id.as_str().into()])?
1286	.run()
1287	.await?;
1288	let env: Map<String, Value> = self
1289	.sealer
1290	.as_ref()
1291	.and_then(\|sealer\| sealer.open(&sealed, &task.id))
1292	.and_then(\|text\| serde_json::from_str(&text).ok())
1293	.unwrap_or_default();
1294	if env.is_empty() {
1295	self.end_task(&TaskRow { status: "in_progress".into(), ..task.clone() }, 1, "The work's environment could not be opened.").await?;
1296	continue;
1297	}
1298	self.audit(vec![entry(
1299	runner_actor(runner),
1300	"runner.take_task",
1301	&runner.workspace,
1302	Some(task.repo.clone()),
1303	Some(task.kind.clone()),
1304	format!("The self-hosted runner {} took agent work: {}.", runner.name, task.title),
1305	)])
1306	.await;
1307	let visibility = match &task.repo_id {
1308	Some(repo_id) => self.visibility(repo_id).await,
1309	None => None,
1310	};
1311	return Ok(Some(Assignment {
1312	kind: "agent".into(),
1313	id: task.id.clone(),
1314	name: task.title.clone(),
1315	repo: task.repo.clone(),
1316	timeout_minutes: task.timeout_minutes,
1317	image: None,
1318	token: None,
1319	env: Some(env),
1320	visibility,
1321	}));
1322	}
1323	Ok(None)
1324	}
1325
1326	/// [`Self::visibility`], asked once per repository in a poll.
1327	async fn visibility_once(&self, looked: &mut Looked, repo_id: &str) -> Option<String> {
1328	if let Some(known) = looked.visibilities.get(repo_id) {
1329	return known.clone();
1330	}
1331	let found = self.visibility(repo_id).await;
1332	looked.visibilities.insert(repo_id.to_owned(), found.clone());
1333	found
1334	}
1335
1336	/// A public repository's job that a host-mode runner passed over: when
1337	/// no online runner that may take it could (its labels, group,
1338	/// repository, and isolation), its page says it waits for a sandboxed
1339	/// one. Written only when that changes what it says.
1340	async fn wait_for_sandbox(&self, job: &JobRow, run_repo: &str, wanted: &Wanted) -> Result<()> {
1341	let at = now_ms();
1342	let groups = self.group_rows(&job.namespace.to_lowercase()).await?;
1343	let runners = self.runner_rows(&job.namespace.to_lowercase()).await?;
1344	let sandboxed = runners.iter().any(\|other\| takes_public_job(other, &groups, &job.repo_id, run_repo, wanted, at));
1345	if sandboxed {
1346	return Ok(());
1347	}
1348	self.db
1349	.prepare("UPDATE jobs SET reason = ? WHERE id = ? AND status = 'queued' AND reason IS NOT ?")
1350	.bind(&[model::WAITING_FOR_SANDBOX.into(), job.id.as_str().into(), model::WAITING_FOR_SANDBOX.into()])?
1351	.run()
1352	.await?;
1353	Ok(())
1354	}
1355
1356	/// A repository's visibility for a runner, `public` or `private`; none
1357	/// when it cannot be read, and the runner decides without it.
1358	async fn visibility(&self, repo_id: &str) -> Option<String> {
1359	match self.repo_by_id(repo_id).await {
1360	Ok(Some((repo, _))) => Some(if repo.is_private { "private" } else { "public" }.to_owned()),
1361	_ => None,
1362	}
1363	}
1364
1365	/// The image a job's `container:` names, when it names one plainly.
1366	async fn job_image(&self, job: &JobRow) -> Result<Option<String>> {
1367	let Some(run) = self.run_row(&job.run_id).await? else { return Ok(None) };
1368	let raw = match job.callee() {
1369	Some((_, spec, _)) => spec.raw,
1370	None => match g1t_actions::workflow::parse(&run.source) {
1371	Ok(workflow) => match workflow.jobs.into_iter().find(\|j\| j.id == job.key) {
1372	Some(spec) => spec.raw,
1373	None => return Ok(None),
1374	},
1375	Err(_) => return Ok(None),
1376	},
1377	};
1378	let image = match raw.get("container") {
1379	Some(Value::String(image)) => Some(image.clone()),
1380	Some(Value::Object(container)) => container.get("image").and_then(Value::as_str).map(str::to_owned),
1381	_ => None,
1382	};
1383	Ok(image.filter(\|image\| !image.contains("${{") && !image.trim().is_empty()))
1384	}
1385
1386	/// `runner_finished`.
1387	pub async fn runner_finished(&self, a: FinishedArgs) -> Result<Outcome<bool>> {
1388	let Some((row, _)) = self.authenticated(&a.auth).await? else {
1389	return Ok(Self::unknown_runner());
1390	};
1391	let reason = a.reason.clone().filter(\|r\| !r.trim().is_empty()).map(\|r\| r.chars().take(2000).collect::<String>());
1392	let mut changed = false;
1393	if let Some(job) = self.db.prepare("SELECT * FROM jobs WHERE id = ?").bind(&[a.id.as_str().into()])?.first::<JobRow>(None).await? {
1394	if job.runner_id.as_deref() == Some(row.id.as_str()) && job.status == "in_progress" {
1395	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));
1396	self.finish_job(&job.id, "failure", Some(&why), None).await?;
1397	changed = true;
1398	}
1399	} else if let Some(task) = self.db.prepare("SELECT * FROM runner_tasks WHERE id = ?").bind(&[a.id.as_str().into()])?.first::<TaskRow>(None).await?
1400	&& task.runner_id.as_deref() == Some(row.id.as_str())
1401	&& task.status == "in_progress"
1402	{
1403	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) });
1404	self.end_task(&task, a.exit_code, &why).await?;
1405	changed = true;
1406	}
1407	self.db
1408	.prepare("UPDATE runners SET work_id = NULL, work_kind = NULL, last_seen_at = ? WHERE id = ? AND work_id = ?")
1409	.bind(&[now().into(), row.id.as_str().into(), a.id.as_str().into()])?
1410	.run()
1411	.await?;
1412	Ok(Outcome::Ok(changed))
1413	}
1414
1415	/// `runner_remove_self`.
1416	pub async fn runner_remove_self(&self, a: RemoveSelfArgs) -> Result<Outcome<bool>> {
1417	let Some((row, _)) = self.authenticated(&a.auth).await? else {
1418	return Ok(Self::unknown_runner());
1419	};
1420	self.forget_runner(&row, &format!("The runner {} was removed from its machine.", row.name)).await?;
1421	self.audit(vec![entry(
1422	runner_actor(&row),
1423	"remove_runner",
1424	&row.workspace,
1425	row.repo.clone(),
1426	Some(row.name.clone()),
1427	format!("The self-hosted runner {} removed itself.", row.name),
1428	)])
1429	.await;
1430	Ok(Outcome::Ok(true))
1431	}
1432
1433	/// A self-hosted job finished, however it did: its runner is free, and
1434	/// its time goes on the workspace's usage as self-hosted, at $0.
1435	pub async fn released(&self, job: &JobRow) -> Result<()> {
1436	let Some(runner_id) = &job.runner_id else { return Ok(()) };
1437	self.db
1438	.prepare("UPDATE runners SET work_id = NULL, work_kind = NULL WHERE id = ? AND work_id = ?")
1439	.bind(&[runner_id.as_str().into(), job.id.as_str().into()])?
1440	.run()
1441	.await?;
1442	let (Some(started), Some(finished)) = (job.started_at.as_deref(), job.finished_at.as_deref().map(str::to_owned).or_else(\|\| Some(now()))) else {
1443	return Ok(());
1444	};
1445	let seconds = match (g1t_contracts::time::parse_rfc3339(started), g1t_contracts::time::parse_rfc3339(&finished)) {
1446	(Some(a), Some(b)) if b > a => ((b - a) / 1000).max(1),
1447	_ => 1,
1448	};
1449	let run = self.run_row(&job.run_id).await?;
1450	let repo = run.as_ref().map(\|r\| r.repo.clone());
1451	let recorded: Result<Outcome<bool>> = g1t_kit::call(
1452	&self.billing,
1453	"record_sandbox",
1454	&RecordSandboxArgs {
1455	workspace: job.namespace.to_lowercase(),
1456	seconds: u32::try_from(seconds).unwrap_or(u32::MAX),
1457	description: format!(
1458	"{} in {} on the self-hosted runner {}",
1459	job.name,
1460	repo.clone().unwrap_or_default(),
1461	job.runner_name.clone().unwrap_or_default()
1462	),
1463	repo,
1464	reference: format!("selfhosted/{}/{}", job.id, started),
1465	kind: Some(ComputeKind::Workflow),
1466	cpu_seconds: None,
1467	// A workflow job on a self-hosted runner: no agent ran it and nobody asked an agent for it.
1468	agent: None,
1469	asked_by: None,
1470	reservation_id: None,
1471	self_hosted: true,
1472	instance: None,
1473	},
1474	)
1475	.await;
1476	if let Err(error) = recorded {
1477	worker::console_error!("actions: self-hosted time not recorded for {}: {error}", job.id);
1478	}
1479	Ok(())
1480	}
1481
1482	/// `runner` and `RUNNER_*` for a job on a self-hosted runner.
1483	pub async fn runner_context_for(&self, runner_id: &str, variables: &mut Map<String, Value>) -> Result<Value> {
1484	let mut context = g1t_actions::events::runner_context();
1485	let Some(runner) = self.runner_by_id(runner_id).await? else { return Ok(context) };
1486	let os = match runner.os.as_str() {
1487	"macos" => "macOS",
1488	"windows" => "Windows",
1489	_ => "Linux",
1490	};
1491	let arch = if runner.arch == "arm64" { "ARM64" } else { "X64" };
1492	context["name"] = json!(runner.name);
1493	context["os"] = json!(os);
1494	context["arch"] = json!(arch);
1495	context["environment"] = json!("self-hosted");
1496	{
1497	let vars = variables;
1498	vars.insert("RUNNER_NAME".into(), json!(runner.name));
1499	vars.insert("RUNNER_OS".into(), json!(os));
1500	vars.insert("RUNNER_ARCH".into(), json!(arch));
1501	vars.insert("RUNNER_ENVIRONMENT".into(), json!("self-hosted"));
1502	}
1503	Ok(context)
1504	}
1505
1506	// --- Agent work, for the runner service ----------------------------------
1507
1508	/// `runner_route`: the labels g1t's own work in `repo` runs on, when
1509	/// its workspace (or the repository) sends it to self-hosted runners.
1510	pub async fn runner_route(&self, a: RouteArgs) -> Result<Option<Vec<String>>> {
1511	let repo_id = match self.repo_by_path(&a.repo).await? {
1512	Some(repo) => Some(repo.id),
1513	None => None,
1514	};
1515	let settings = self.effective_runner_settings(&a.workspace, repo_id.as_deref()).await?;
1516	Ok(settings.agents_on_self_hosted.then_some(settings.agent_labels))
1517	}
1518
1519	async fn repo_by_path(&self, path: &RepoPath) -> Result<Option<Repo>> {
1520	let Some(actor) = self.workspace_actor(&path.namespace).await? else { return Ok(None) };
1521	self.visible_repo(path, &Some(actor)).await
1522	}
1523
1524	/// `enqueue_task`.
1525	pub async fn enqueue_task(&self, a: EnqueueTaskArgs) -> Result<Outcome<String>> {
1526	let Some(sealer) = &self.sealer else {
1527	return Ok(fail(FailureCode::Conflict, "Agent work cannot be handed to self-hosted runners yet: g1t's key for it is not set."));
1528	};
1529	let id = new_id("rtk", now_ms());
1530	let sealed = sealer.seal(&serde_json::to_string(&a.env)?, &id);
1531	let repo_id = self.repo_by_path(&a.repo).await?.map(\|r\| r.id);
1532	let labels = match model::runner_labels(&a.labels, model::SELF_HOSTED, model::SELF_HOSTED) {
1533	Ok(labels) => labels,
1534	Err(problem) => return Ok(fail(FailureCode::Invalid, problem)),
1535	};
1536	let at = now();
1537	let inserted = self
1538	.db
1539	.prepare(
1540	"INSERT OR IGNORE INTO runner_tasks (id, sandbox, workspace, repo_id, repo, kind, title, labels, env, timeout_minutes, status, created_at, run_id)
1541	VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, 'queued', ?, ?) RETURNING id",
1542	)
1543	.bind(&[
1544	id.as_str().into(),
1545	a.sandbox.as_str().into(),
1546	a.workspace.to_lowercase().into(),
1547	optional(repo_id.as_deref()),
1548	format!("{}/{}", a.repo.namespace, a.repo.name).into(),
1549	a.kind.as_str().into(),
1550	a.title.chars().take(300).collect::<String>().into(),
1551	serde_json::to_string(&labels)?.into(),
1552	sealed.into(),
1553	a.timeout_minutes.max(1).into(),
1554	at.into(),
1555	optional(a.run_id.as_deref().filter(\|id\| !id.is_empty())),
1556	])?
1557	.first::<Value>(None)
1558	.await?;
1559	if inserted.is_none() {
1560	return Ok(fail(FailureCode::Conflict, "That sandbox already handed work to a self-hosted runner."));
1561	}
1562	Ok(Outcome::Ok(id))
1563	}
1564
1565	/// `cancel_task`: the sandbox stopped it. Its runner hears on its next
1566	/// poll; the sandbox does its own ending.
1567	pub async fn cancel_task(&self, a: CancelTaskArgs) -> Result<Outcome<bool>> {
1568	let done = self
1569	.db
1570	.prepare(
1571	"UPDATE runner_tasks SET status = 'completed', exit_code = -1, reason = ?, env = NULL, finished_at = ?
1572	WHERE sandbox = ? AND status != 'completed' RETURNING id",
1573	)
1574	.bind(&[optional(a.reason.as_deref()), now().into(), a.sandbox.as_str().into()])?
1575	.first::<Value>(None)
1576	.await?;
1577	Ok(Outcome::Ok(done.is_some()))
1578	}
1579
1580	/// Ends a task and tells the sandbox waiting on it, once.
1581	async fn end_task(&self, task: &TaskRow, exit_code: i32, reason: &str) -> Result<()> {
1582	let ended = self
1583	.db
1584	.prepare(
1585	"UPDATE runner_tasks SET status = 'completed', exit_code = ?, reason = ?, env = NULL, finished_at = ?
1586	WHERE id = ? AND status != 'completed' RETURNING id",
1587	)
1588	.bind(&[exit_code.into(), optional(Some(reason).filter(\|r\| !r.is_empty())), now().into(), task.id.as_str().into()])?
1589	.first::<Value>(None)
1590	.await?;
1591	if ended.is_none() {
1592	return Ok(());
1593	}
1594	if let Some(runner_id) = &task.runner_id {
1595	self.db
1596	.prepare("UPDATE runners SET work_id = NULL, work_kind = NULL WHERE id = ? AND work_id = ?")
1597	.bind(&[runner_id.as_str().into(), task.id.as_str().into()])?
1598	.run()
1599	.await?;
1600	}
1601	let told: Result<Value> = g1t_kit::call(
1602	&self.runner,
1603	"task_ended",
1604	&json!({
1605	"sandbox": task.sandbox,
1606	"exitCode": exit_code,
1607	"reason": if reason.is_empty() { Value::Null } else { json!(reason) },
1608	"runner": task.runner_name,
1609	}),
1610	)
1611	.await;
1612	if let Err(error) = told {
1613	worker::console_error!("actions: the sandbox waiting on task {} was not told: {error}", task.id);
1614	}
1615	Ok(())
1616	}
1617
1618	/// `runner_activity`: what runs for the workspace now, in g1t's
1619	/// sandboxes and handed to its own runners. Its owners only, as its
1620	/// runners are.
1621	pub async fn runner_activity(&self, a: RunnerActivityArgs) -> Result<Outcome<RunnerActivity>> {
1622	let owner = RunnersOwner { repo: None, workspace: Some(a.workspace.clone()) };
1623	let place = check!(self.runner_place(&a.actor, &owner, false).await?);
1624	let workspace = place.workspace;
1625	let cloud_jobs = self
1626	.db
1627	.prepare(
1628	"SELECT jobs.id, jobs.name, jobs.run_id, runs.repo, jobs.started_at FROM jobs JOIN runs ON runs.id = jobs.run_id
1629	WHERE lower(jobs.namespace) = ? AND jobs.status = 'in_progress' AND jobs.labels IS NULL
1630	ORDER BY jobs.started_at DESC LIMIT ?",
1631	)
1632	.bind(&[workspace.as_str().into(), ACTIVITY_LIMIT.into()])?
1633	.all()
1634	.await?
1635	.results::<CloudJob>()?;
1636	let cloud_jobs_queued = self.queued_jobs(&workspace, false).await?;
1637	let self_hosted_jobs_queued = self.queued_jobs(&workspace, true).await?;
1638	let handed_over = self
1639	.db
1640	.prepare(
1641	"SELECT id, run_id, kind, title, repo, status, runner_name, created_at, started_at FROM runner_tasks
1642	WHERE status IN ('queued', 'in_progress') AND workspace = ?
1643	ORDER BY created_at DESC LIMIT ?",
1644	)
1645	.bind(&[workspace.as_str().into(), ACTIVITY_LIMIT.into()])?
1646	.all()
1647	.await?
1648	.results::<HandedOverTask>()?;
1649	Ok(Outcome::Ok(RunnerActivity { cloud_jobs, cloud_jobs_queued, handed_over, self_hosted_jobs_queued }))
1650	}
1651
1652	/// The workspace's workflow jobs queued for its own runners
1653	/// (`labelled`), or for g1t's sandboxes.
1654	async fn queued_jobs(&self, workspace: &str, labelled: bool) -> Result<u32> {
1655	let sql = if labelled {
1656	"SELECT COUNT(*) AS n FROM jobs WHERE lower(namespace) = ? AND status = 'queued' AND labels IS NOT NULL"
1657	} else {
1658	"SELECT COUNT(*) AS n FROM jobs WHERE lower(namespace) = ? AND status = 'queued' AND labels IS NULL"
1659	};
1660	Ok(self.db.prepare(sql).bind(&[workspace.into()])?.first::<Count>(None).await?.map_or(0, \|c\| c.n))
1661	}
1662
1663	/// `stuck_jobs`: jobs of the viewer's workspaces that have waited ten
1664	/// minutes or more for a self-hosted runner while none that could take
1665	/// them is online, for Mission control's Needs you.
1666	pub async fn stuck_jobs(&self, a: StuckJobsArgs) -> Result<Vec<StuckJob>> {
1667	let Some(viewer) = a.viewer else { return Ok(Vec::new()) };
1668	let at = now_ms();
1669	let mut out = Vec::new();
1670	for membership in viewer.workspaces.iter().take(20) {
1671	let workspace = membership.slug.to_lowercase();
1672	#[derive(Deserialize)]
1673	struct Waiting {
1674	id: String,
1675	name: String,
1676	run_id: String,
1677	labels: String,
1678	queued_at: Option<String>,
1679	repo: String,
1680	repo_id: String,
1681	}
1682	let waiting = self
1683	.db
1684	.prepare(
1685	"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
1686	WHERE jobs.status = 'queued' AND jobs.labels IS NOT NULL AND lower(jobs.namespace) = ? AND jobs.queued_at < ?
1687	ORDER BY jobs.rowid LIMIT 20",
1688	)
1689	.bind(&[workspace.as_str().into(), rfc3339(at.saturating_sub(10 * 60 * 1000)).into()])?
1690	.all()
1691	.await?
1692	.results::<Waiting>()?;
1693	if waiting.is_empty() {
1694	continue;
1695	}
1696	let groups = self.group_rows(&workspace).await?;
1697	let online: Vec<RunnerRow> = self.runner_rows(&workspace).await?.into_iter().filter(\|r\| r.online(at)).collect();
1698	// A repository's visibility is asked only when a host-mode runner
1699	// is online, since only those pass over a public repository's jobs.
1700	let any_host = online.iter().any(\|runner\| runner.isolation.as_deref() == Some("host"));
1701	let mut looked = Looked::default();
1702	for job in waiting {
1703	let wanted = Wanted::from_stored(&serde_json::from_str::<Vec<String>>(&job.labels).unwrap_or_default());
1704	let visibility = if any_host { self.visibility_once(&mut looked, &job.repo_id).await } else { None };
1705	let served = online.iter().any(\|runner\| {
1706	let group = runner.group_id.as_ref().and_then(\|id\| groups.iter().find(\|g\| &g.id == id));
1707	wanted.matches(&runner.labels(), group.map(\|g\| g.name.as_str()))
1708	&& model::isolation_allows(runner.isolation.as_deref(), visibility.as_deref())
1709	});
1710	if !served {
1711	out.push(StuckJob {
1712	id: job.id,
1713	name: job.name,
1714	run_id: job.run_id,
1715	repo: job.repo,
1716	labels: wanted.describe(),
1717	queued_at: job.queued_at.unwrap_or_default(),
1718	});
1719	}
1720	}
1721	}
1722	Ok(out)
1723	}
1724
1725	// --- Every minute ----------------------------------------------------------
1726
1727	/// Jobs and tasks that waited too long, tasks whose runner went quiet,
1728	/// runners offline for weeks, and registration tokens long expired.
1729	pub async fn sweep_runners(&self, at: u64) -> Result<()> {
1730	let before = \|ms: u64\| rfc3339(at.saturating_sub(ms));
1731	let waited = model::MAX_WAIT_HOURS * 60 * 60 * 1000;
1732	let stale = self
1733	.db
1734	.prepare("SELECT * FROM jobs WHERE status = 'queued' AND labels IS NOT NULL AND queued_at < ? LIMIT 50")
1735	.bind(&[before(waited).into()])?
1736	.all()
1737	.await?
1738	.results::<JobRow>()?;
1739	for job in stale {
1740	let stored: Vec<String> = job.labels.as_deref().and_then(\|l\| serde_json::from_str(l).ok()).unwrap_or_default();
1741	let why = format!(
1742	"No self-hosted runner with labels {} took it within {} hours.",
1743	Wanted::from_stored(&stored).describe(),
1744	model::MAX_WAIT_HOURS
1745	);
1746	self.finish_job(&job.id, "failure", Some(&why), None).await?;
1747	}
1748	let tasks = self
1749	.db
1750	.prepare(
1751	"SELECT * FROM runner_tasks WHERE (status = 'queued' AND created_at < ?) OR (status = 'in_progress' AND seen_at < ?) LIMIT 50",
1752	)
1753	.bind(&[before(waited).into(), before(SILENT_MS).into()])?
1754	.all()
1755	.await?
1756	.results::<TaskRow>()?;
1757	for task in tasks {
1758	let why = if task.status == "queued" {
1759	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)
1760	} else {
1761	format!("The self-hosted runner {} stopped answering.", task.runner_name.clone().unwrap_or_default())
1762	};
1763	self.end_task(&task, 1, &why).await?;
1764	}
1765	let forgotten = self
1766	.db
1767	.prepare("SELECT * FROM runners WHERE COALESCE(last_seen_at, created_at) < ? LIMIT 50")
1768	.bind(&[before(FORGET_OFFLINE_MS).into()])?
1769	.all()
1770	.await?
1771	.results::<RunnerRow>()?;
1772	for row in forgotten {
1773	self.forget_runner(&row, "The runner was offline for 14 days.").await?;
1774	}
1775	// An ephemeral runner that took its job and then went quiet.
1776	let spent = self
1777	.db
1778	.prepare("SELECT * FROM runners WHERE ephemeral = 1 AND spent = 1 AND work_id IS NULL AND COALESCE(last_seen_at, created_at) < ? LIMIT 50")
1779	.bind(&[before(60 * 60 * 1000).into()])?
1780	.all()
1781	.await?
1782	.results::<RunnerRow>()?;
1783	for row in spent {
1784	self.forget_runner(&row, "The ephemeral runner finished its job.").await?;
1785	}
1786	self.db
1787	.prepare("DELETE FROM runner_registrations WHERE expires_at < ?")
1788	.bind(&[before(24 * 60 * 60 * 1000).into()])?
1789	.run()
1790	.await?;
1791	Ok(())
1792	}
1793	}
1794
1795	#[derive(Deserialize)]
1796	struct IdRow {
1797	id: String,
1798	}
1799
1800	/// Whether `runner` is online and could take a public repository's job:
1801	/// not host-mode, and its repository, group and labels let it.
1802	fn takes_public_job(runner: &RunnerRow, groups: &[GroupRow], repo_id: &str, run_repo: &str, wanted: &Wanted, at: u64) -> bool {
1803	let group = runner.group_id.as_ref().and_then(\|id\| groups.iter().find(\|g\| &g.id == id));
1804	runner.online(at)
1805	&& model::isolation_allows(runner.isolation.as_deref(), Some("public"))
1806	&& runner.repo_id.as_deref().is_none_or(\|repo\| repo == repo_id)
1807	&& (runner.repo_id.is_some() \|\| group.is_none_or(\|g\| g.allows(run_repo)))
1808	&& wanted.matches(&runner.labels(), group.map(\|g\| g.name.as_str()))
1809	}
1810
1811	fn group_view(group: &GroupRow, runners: &[RunnerRow]) -> RunnerGroup {
1812	RunnerGroup {
1813	id: group.id.clone(),
1814	name: group.name.clone(),
1815	default: group.is_default != 0,
1816	repositories: group.repositories(),
1817	runners: runners.iter().filter(\|r\| r.group_id.as_deref() == Some(group.id.as_str())).count() as u32,
1818	updated_at: group.updated_at.clone(),
1819	}
1820	}
1821
1822	#[cfg(test)]
1823	mod tests {
1824	use super::*;
1825
1826	fn group(repositories: &[&str]) -> GroupRow {
1827	GroupRow {
1828	id: "rng_1".into(),
1829	name: "Default".into(),
1830	is_default: 1,
1831	repositories: serde_json::to_string(&repositories).unwrap(),
1832	updated_at: "2026-10-06T00:00:00Z".into(),
1833	}
1834	}
1835
1836	#[test]
1837	fn being_seen_is_written_once_a_minute() {
1838	let at = 2_000_000_000_000;
1839	assert!(seen_is_stale(None, at));
1840	assert!(!seen_is_stale(Some(&rfc3339(at - 59_000)), at));
1841	assert!(seen_is_stale(Some(&rfc3339(at - 61_000)), at));
1842	// A runner written down at the last moment is still online until the
1843	// next write, a poll later.
1844	assert!(SEEN_RESOLUTION_MS + model::MAX_POLL_WAIT_MS < model::ONLINE_WITHIN_MS);
1845	assert!(SEEN_RESOLUTION_MS < crate::SILENT_MS);
1846	}
1847
1848	#[test]
1849	fn a_group_lets_its_repositories_use_it() {
1850	assert!(group(&[]).allows("acme/web"));
1851	assert!(group(&["web", "api"]).allows("acme/WEB"));
1852	assert!(group(&["web"]).allows("web"));
1853	assert!(!group(&["web"]).allows("acme/docs"));
1854	}
1855
1856	#[test]
1857	fn names_are_checked() {
1858	assert_eq!(valid_name(" build-01 ").unwrap(), "build-01");
1859	assert!(valid_name("").is_err());
1860	assert!(valid_name("has space").is_err());
1861	assert!(valid_name(&"x".repeat(65)).is_err());
1862	assert!(valid_group_name("GPU machines").is_ok());
1863	assert!(valid_group_name("").is_err());
1864	}
1865
1866	#[test]
1867	fn only_a_sandboxed_runner_that_could_take_it_serves_a_public_job() {
1868	let at = 1_000_000_000;
1869	let runner = \|isolation: Option<&str>\| RunnerRow {
1870	id: "rnr_1".into(),
1871	workspace: "acme".into(),
1872	repo_id: None,
1873	repo: None,
1874	group_id: None,
1875	name: "a".into(),
1876	labels: r#"["self-hosted","linux","x64"]"#.into(),
1877	os: "linux".into(),
1878	arch: "x64".into(),
1879	version: String::new(),
1880	ephemeral: 0,
1881	credential_hash: String::new(),
1882	previous_hash: None,
1883	rotated_at: String::new(),
1884	work_id: None,
1885	work_kind: None,
1886	spent: 0,
1887	last_seen_at: Some(rfc3339(at)),
1888	created_at: String::new(),
1889	created_by: None,
1890	isolation: isolation.map(str::to_owned),
1891	};
1892	let wanted = Wanted::of(&["self-hosted".to_owned(), "linux".to_owned()], None);
1893	let takes = \|row: &RunnerRow\| takes_public_job(row, &[], "rep_1", "acme/web", &wanted, at);
1894	assert!(!takes(&runner(Some("host"))));
1895	assert!(takes(&runner(Some("container"))));
1896	assert!(takes(&runner(Some("nested-container"))));
1897	assert!(takes(&runner(None)), "an older runner, which refuses for itself");
1898	assert!(!takes(&RunnerRow { repo_id: Some("rep_2".into()), ..runner(None) }), "another repository's");
1899	assert!(!takes(&RunnerRow { labels: r#"["self-hosted","macos"]"#.into(), ..runner(None) }));
1900	assert!(!takes(&RunnerRow { last_seen_at: None, ..runner(None) }), "offline");
1901	}
1902
1903	#[test]
1904	fn online_means_seen_in_the_last_ninety_seconds() {
1905	let row = RunnerRow {
1906	id: "rnr_1".into(),
1907	workspace: "acme".into(),
1908	repo_id: None,
1909	repo: None,
1910	group_id: None,
1911	name: "a".into(),
1912	labels: "[]".into(),
1913	os: "linux".into(),
1914	arch: "x64".into(),
1915	version: String::new(),
1916	ephemeral: 0,
1917	credential_hash: String::new(),
1918	previous_hash: None,
1919	rotated_at: String::new(),
1920	work_id: None,
1921	work_kind: None,
1922	spent: 0,
1923	last_seen_at: Some(rfc3339(1_000_000_000)),
1924	created_at: String::new(),
1925	created_by: None,
1926	isolation: None,
1927	};
1928	assert!(row.online(1_000_000_000 + 60_000));
1929	assert!(!row.online(1_000_000_000 + 120_000));
1930	assert_eq!(row.workspace, "acme");
1931	}
1932	}