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, DesktopCapability, 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	capabilities: d
730	.capabilities
731	.iter()
732	.take(model::MAX_DESKTOP_CAPABILITIES)
733	.map(\|c\| DesktopCapability {
734	id: clip(&c.id, 20),
735	label: clip(&c.label, 60),
736	tone: clip(&c.tone, 20),
737	title: clip(&c.title, 200),
738	cause: c.cause.as_deref().map(\|t\| clip(t, 1000)),
739	fix: c.fix.as_deref().map(\|t\| clip(t, 1000)),
740	})
741	.collect(),
742	};
743	let at = now_ms();
744	let known = self
745	.db
746	.prepare("SELECT COUNT(*) AS n FROM desktop_runners WHERE workspace = ? AND lower(machine) <> lower(?)")
747	.bind(&[place.workspace.as_str().into(), machine.as_str().into()])?
748	.first::<Count>(None)
749	.await?
750	.map_or(0, \|c\| c.n);
751	if known >= model::MAX_DESKTOPS {
752	// The oldest report makes room: a computer long gone.
753	self.db
754	.prepare("DELETE FROM desktop_runners WHERE rowid IN (SELECT rowid FROM desktop_runners WHERE workspace = ? ORDER BY reported_at LIMIT 1)")
755	.bind(&[place.workspace.as_str().into()])?
756	.run()
757	.await?;
758	}
759	self.db
760	.prepare(
761	"INSERT INTO desktop_runners (workspace, machine, person, backend, version, diagnosis, closed, reported_at)
762	VALUES (?, ?, ?, ?, ?, ?, ?, ?)
763	ON CONFLICT (workspace, machine) DO UPDATE SET person = ?3, backend = ?4, version = ?5, diagnosis = ?6, closed = ?7, reported_at = ?8",
764	)
765	.bind(&[
766	place.workspace.as_str().into(),
767	machine.as_str().into(),
768	a.actor.username.as_str().into(),
769	a.backend.as_str().into(),
770	optional(a.version.as_deref().map(\|v\| clip(v, 40)).as_deref()),
771	serde_json::to_string(&tidy(&a.diagnosis))?.into(),
772	serde_json::to_string(&tidy(&a.closed))?.into(),
773	rfc3339(at).into(),
774	])?
775	.run()
776	.await?;
777	Ok(Outcome::Ok(true))
778	}
779
780	/// `desktop_runners`: the workspace's desktops, as each app last said.
781	pub async fn desktop_runners(&self, a: DesktopRunnersArgs) -> Result<Outcome<Vec<DesktopRunner>>> {
782	let owner = RunnersOwner { repo: None, workspace: Some(a.workspace.clone()) };
783	let place = check!(self.runner_place(&a.actor, &owner, false).await?);
784	#[derive(Deserialize)]
785	struct Row {
786	workspace: String,
787	machine: String,
788	person: String,
789	backend: String,
790	version: Option<String>,
791	diagnosis: String,
792	closed: String,
793	reported_at: String,
794	}
795	let rows = self
796	.db
797	.prepare("SELECT * FROM desktop_runners WHERE workspace = ? ORDER BY lower(machine)")
798	.bind(&[place.workspace.as_str().into()])?
799	.all()
800	.await?
801	.results::<Row>()?;
802	let open_since = rfc3339(now_ms().saturating_sub(model::DESKTOP_OPEN_WITHIN_MS));
803	Ok(Outcome::Ok(
804	rows.into_iter()
805	.map(\|row\| DesktopRunner {
806	open: row.reported_at.as_str() > open_since.as_str(),
807	diagnosis: serde_json::from_str(&row.diagnosis).unwrap_or_default(),
808	closed: serde_json::from_str(&row.closed).unwrap_or_default(),
809	workspace: row.workspace,
810	machine: row.machine,
811	person: row.person,
812	backend: row.backend,
813	version: row.version,
814	reported_at: row.reported_at,
815	})
816	.collect(),
817	))
818	}
819
820	/// `set_runner_settings`.
821	pub async fn set_runner_settings(&self, a: SetRunnerSettingsArgs) -> Result<Outcome<RunnerSettings>> {
822	let place = check!(self.runner_place(&a.actor, &a.owner, true).await?);
823	let (owner, scope) = match &place.repo {
824	Some(repo) => (repo.id.clone(), "repository"),
825	None => (place.workspace.clone(), "workspace"),
826	};
827	if a.inherit && place.repo.is_some() {
828	self.db.prepare("DELETE FROM runner_settings WHERE owner = ?").bind(&[owner.as_str().into()])?.run().await?;
829	} else {
830	let current = self.effective_runner_settings(&place.workspace, place.repo.as_ref().map(\|r\| r.id.as_str())).await?;
831	let labels = match &a.agent_labels {
832	Some(given) => match model::runner_labels(given, model::SELF_HOSTED, model::SELF_HOSTED) {
833	Ok(labels) => labels,
834	Err(problem) => return Ok(fail(FailureCode::Invalid, problem)),
835	},
836	None => current.agent_labels.clone(),
837	};
838	let agents = a.agents_on_self_hosted.unwrap_or(current.agents_on_self_hosted);
839	let forks = a.fork_pull_requests.unwrap_or(current.fork_pull_requests);
840	self.db
841	.prepare(
842	"INSERT INTO runner_settings (owner, scope, agents, agent_labels, fork_pulls, updated_at, updated_by)
843	VALUES (?1, ?2, ?3, ?4, ?5, ?6, ?7)
844	ON CONFLICT (owner) DO UPDATE SET agents = ?3, agent_labels = ?4, fork_pulls = ?5, updated_at = ?6, updated_by = ?7",
845	)
846	.bind(&[
847	owner.as_str().into(),
848	scope.into(),
849	u32::from(agents).into(),
850	serde_json::to_string(&labels)?.into(),
851	u32::from(forks).into(),
852	now().into(),
853	a.actor.username.as_str().into(),
854	])?
855	.run()
856	.await?;
857	}
858	let settings = self.effective_runner_settings(&place.workspace, place.repo.as_ref().map(\|r\| r.id.as_str())).await?;
859	self.audit(vec![entry(
860	AuditActor::of(&a.actor),
861	"update_runner_settings",
862	&place.workspace,
863	place.repo.as_ref().map(\|r\| format!("{}/{}", r.namespace, r.name)),
864	None,
865	format!(
866	"Agents on self-hosted runners: {}; labels {}; pull requests from forks: {}.",
867	if settings.agents_on_self_hosted { "on" } else { "off" },
868	settings.agent_labels.join(", "),
869	if settings.fork_pull_requests { "allowed" } else { "not allowed" }
870	),
871	)])
872	.await;
873	Ok(Outcome::Ok(settings))
874	}
875
876	// --- The runner's side ---------------------------------------------------
877
878	/// `runner_register`.
879	pub async fn runner_register(&self, a: RegisterArgs) -> Result<Outcome<Registered>> {
880	let refused = \|\| fail(FailureCode::Unauthenticated, "That registration token is not valid, or it expired. Make a new one under Settings, Runners.");
881	if !a.token.starts_with(model::REGISTRATION_PREFIX) {
882	return Ok(refused());
883	}
884	let registration = self
885	.db
886	.prepare("SELECT * FROM runner_registrations WHERE token_hash = ?")
887	.bind(&[hash(&a.token).into()])?
888	.first::<RegistrationRow>(None)
889	.await?;
890	let at = now_ms();
891	let Some(registration) = registration.filter(\|r\| r.expires_at.as_str() > rfc3339(at).as_str()) else {
892	return Ok(refused());
893	};
894	let name = match valid_name(&a.name) {
895	Ok(name) => name,
896	Err(problem) => return Ok(fail(FailureCode::Invalid, problem)),
897	};
898	let (Some(os), Some(arch)) = (model::os_label(&a.os), model::arch_label(&a.arch)) else {
899	return Ok(fail(FailureCode::Invalid, "A runner's os is linux, macos or windows, and its arch x64 or arm64."));
900	};
901	let labels = match model::runner_labels(&a.labels, os, arch) {
902	Ok(labels) => labels,
903	Err(problem) => return Ok(fail(FailureCode::Invalid, problem)),
904	};
905	let workspace = registration.workspace.clone();
906	// Abuse: so many machines, so fast, are not a team's.
907	let recent = self
908	.db
909	.prepare("SELECT COUNT(*) AS n FROM runners WHERE workspace = ? AND created_at > ?")
910	.bind(&[workspace.as_str().into(), rfc3339(at.saturating_sub(60 * 1000)).into()])?
911	.first::<Count>(None)
912	.await?
913	.map_or(0, \|c\| c.n);
914	if recent >= model::MAX_REGISTRATIONS_PER_MINUTE {
915	return Ok(fail(FailureCode::Conflict, "Too many runners registered in the last minute. Wait a moment and try again."));
916	}
917	let total = self
918	.db
919	.prepare("SELECT COUNT(*) AS n FROM runners WHERE workspace = ?")
920	.bind(&[workspace.as_str().into()])?
921	.first::<Count>(None)
922	.await?
923	.map_or(0, \|c\| c.n);
924	// A group: the one asked for, the token's, or the default.
925	let group = if registration.repo_id.is_some() {
926	None
927	} else {
928	match a.group.as_deref().map(str::trim).filter(\|g\| !g.is_empty()) {
929	Some(wanted) => match self.find_group(&workspace, wanted).await? {
930	Some(group) => Some(group.id),
931	None => return Ok(fail(FailureCode::NotFound, format!("There is no runner group called {wanted}."))),
932	},
933	None => match registration.group_id.clone() {
934	Some(id) => Some(id),
935	None => Some(self.default_group(&workspace).await?.id),
936	},
937	}
938	};
939	let existing = self
940	.db
941	.prepare("SELECT * FROM runners WHERE workspace = ? AND COALESCE(repo_id, '') = ? AND lower(name) = lower(?)")
942	.bind(&[workspace.as_str().into(), registration.repo_id.clone().unwrap_or_default().into(), name.as_str().into()])?
943	.first::<RunnerRow>(None)
944	.await?;
945	if let Some(existing) = &existing {
946	if !a.replace {
947	return Ok(fail(
948	FailureCode::Conflict,
949	format!("A runner called {name} is already registered here. Register with --replace to take its place, or choose another --name."),
950	));
951	}
952	self.forget_runner(existing, &format!("The runner {name} was replaced by a new registration.")).await?;
953	} else if total >= model::MAX_RUNNERS {
954	return Ok(fail(FailureCode::Conflict, format!("A workspace has at most {} self-hosted runners.", model::MAX_RUNNERS)));
955	}
956	// How it keeps work from its machine, if it said (older runners do not).
957	let isolation = a.isolation.as_deref().filter(\|level\| model::ISOLATIONS.contains(level));
958	let id = new_id("rnr", at);
959	let credential = format!("{}{}", model::CREDENTIAL_PREFIX, random_hex(32));
960	let created = rfc3339(at);
961	self.db
962	.prepare(
963	"INSERT INTO runners (id, workspace, repo_id, repo, group_id, name, labels, os, arch, version, ephemeral, credential_hash,
964	rotated_at, last_seen_at, created_at, created_by, isolation)
965	VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)",
966	)
967	.bind(&[
968	id.as_str().into(),
969	workspace.as_str().into(),
970	optional(registration.repo_id.as_deref()),
971	optional(registration.repo.as_deref()),
972	optional(group.as_deref()),
973	name.as_str().into(),
974	serde_json::to_string(&labels)?.into(),
975	os.into(),
976	arch.into(),
977	a.version.chars().take(40).collect::<String>().into(),
978	u32::from(a.ephemeral).into(),
979	hash(&credential).into(),
980	created.as_str().into(),
981	created.as_str().into(),
982	created.as_str().into(),
983	// The person who made the token: a desktop's runner is theirs.
984	optional(registration.created_by.as_deref()),
985	optional(isolation),
986	])?
987	.run()
988	.await?;
989	self.db
990	.prepare("UPDATE runner_registrations SET used = used + 1 WHERE token_hash = ?")
991	.bind(&[hash(&a.token).into()])?
992	.run()
993	.await?;
994	let row = self.runner_by_id(&id).await?.ok_or_else(\|\| worker::Error::RustError("the runner was not kept".into()))?;
995	self.audit(vec![entry(
996	runner_actor(&row),
997	"register_runner",
998	&workspace,
999	row.repo.clone(),
1000	Some(name.clone()),
1001	format!("Registered the self-hosted runner {name} ({os}, {arch}; labels {}).", labels.join(", ")),
1002	)])
1003	.await;
1004	let groups = self.group_rows(&workspace).await?;
1005	Ok(Outcome::Ok(Registered {
1006	runner: self.runner_view(&row, &groups, at).await?,
1007	credential,
1008	}))
1009	}
1010
1011	/// The runner a call is from, if its credential is that runner's (or
1012	/// the one before a rotation, until the new one is used).
1013	async fn authenticated(&self, auth: &RunnerAuth) -> Result<Option<(RunnerRow, bool)>> {
1014	if !auth.credential.starts_with(model::CREDENTIAL_PREFIX) {
1015	return Ok(None);
1016	}
1017	let Some(row) = self.runner_by_id(&auth.runner).await? else { return Ok(None) };
1018	let given = hash(&auth.credential);
1019	if g1t_secrets::same(&row.credential_hash, &given) {
1020	if row.previous_hash.is_some() {
1021	self.db.prepare("UPDATE runners SET previous_hash = NULL WHERE id = ?").bind(&[row.id.as_str().into()])?.run().await?;
1022	}
1023	return Ok(Some((row, true)));
1024	}
1025	if row.previous_hash.as_deref().is_some_and(\|previous\| g1t_secrets::same(previous, &given)) {
1026	return Ok(Some((row, false)));
1027	}
1028	Ok(None)
1029	}
1030
1031	fn unknown_runner<T>() -> Outcome<T> {
1032	fail(FailureCode::Unauthenticated, "This runner is not registered, or its credential is not valid. Register it again.")
1033	}
1034
1035	/// `runner_poll`.
1036	pub async fn runner_poll(&self, a: PollArgs) -> Result<Outcome<Poll>> {
1037	let Some((row, current)) = self.authenticated(&a.auth).await? else {
1038	return Ok(Self::unknown_runner());
1039	};
1040	let at = now_ms();
1041	let seen = rfc3339(at);
1042	// A new credential every day, handed over on a poll that used the
1043	// current one.
1044	let mut credential = None;
1045	let rotate_after = rfc3339(at.saturating_sub(model::ROTATE_AFTER_MS));
1046	if current && row.rotated_at.as_str() < rotate_after.as_str() {
1047	let fresh = format!("{}{}", model::CREDENTIAL_PREFIX, random_hex(32));
1048	self.db
1049	.prepare("UPDATE runners SET previous_hash = credential_hash, credential_hash = ?, rotated_at = ? WHERE id = ?")
1050	.bind(&[hash(&fresh).into(), seen.as_str().into(), row.id.as_str().into()])?
1051	.run()
1052	.await?;
1053	credential = Some(fresh);
1054	}
1055	// What it holds that it should stop, and the heartbeat for the rest.
1056	let mut cancel = Vec::new();
1057	let mut active: Option<(String, &'static str)> = None;
1058	for id in a.running.iter().take(10) {
1059	if let Some(job) = self.db.prepare("SELECT * FROM jobs WHERE id = ?").bind(&[id.as_str().into()])?.first::<JobRow>(None).await? {
1060	if job.status == "in_progress" && job.runner_id.as_deref() == Some(row.id.as_str()) {
1061	if seen_is_stale(job.seen_at.as_deref(), at) {
1062	self.db.prepare("UPDATE jobs SET seen_at = ? WHERE id = ?").bind(&[seen.as_str().into(), id.as_str().into()])?.run().await?;
1063	}
1064	active = Some((id.clone(), "workflow"));
1065	} else {
1066	cancel.push(id.clone());
1067	}
1068	continue;
1069	}
1070	match self.db.prepare("SELECT * FROM runner_tasks WHERE id = ?").bind(&[id.as_str().into()])?.first::<TaskRow>(None).await? {
1071	Some(task) if task.status == "in_progress" && task.runner_id.as_deref() == Some(row.id.as_str()) => {
1072	if seen_is_stale(task.seen_at.as_deref(), at) {
1073	self.db.prepare("UPDATE runner_tasks SET seen_at = ? WHERE id = ?").bind(&[seen.as_str().into(), id.as_str().into()])?.run().await?;
1074	}
1075	active = Some((id.clone(), "agent"));
1076	}
1077	_ => cancel.push(id.clone()),
1078	}
1079	}
1080	// Written when something changed, or once a minute to keep it online.
1081	let version: String = if a.version.is_empty() { row.version.clone() } else { a.version.chars().take(40).collect() };
1082	let work_id = active.as_ref().map(\|(id, _)\| id.as_str());
1083	let work_kind = active.as_ref().map(\|(_, kind)\| *kind);
1084	// How it is isolated, as it says now: a runner re-registered or
1085	// restarted with or without Docker says so on its next poll.
1086	let isolation = a
1087	.isolation
1088	.as_deref()
1089	.filter(\|level\| model::ISOLATIONS.contains(level))
1090	.map(str::to_owned)
1091	.or_else(\|\| row.isolation.clone());
1092	let changed = version != row.version
1093	\|\| work_id != row.work_id.as_deref()
1094	\|\| work_kind != row.work_kind.as_deref()
1095	\|\| isolation != row.isolation;
1096	if changed \|\| seen_is_stale(row.last_seen_at.as_deref(), at) {
1097	self.db
1098	.prepare("UPDATE runners SET last_seen_at = ?, version = ?, work_id = ?, work_kind = ?, isolation = ? WHERE id = ?")
1099	.bind(&[
1100	seen.as_str().into(),
1101	version.into(),
1102	optional(work_id),
1103	optional(work_kind),
1104	optional(isolation.as_deref()),
1105	row.id.as_str().into(),
1106	])?
1107	.run()
1108	.await?;
1109	}
1110	let row = RunnerRow { isolation, ..row };
1111	let mut poll = Poll { assignment: None, cancel, credential, removed: false };
1112	// Busy, or an ephemeral runner that has had its one job.
1113	if active.is_some() \|\| !a.running.is_empty() {
1114	return Ok(Outcome::Ok(poll));
1115	}
1116	if row.ephemeral != 0 && row.spent != 0 {
1117	self.forget_runner(&row, "The ephemeral runner finished its job.").await?;
1118	poll.removed = true;
1119	return Ok(Outcome::Ok(poll));
1120	}
1121	let groups = self.group_rows(&row.workspace).await?;
1122	let group = row.group_id.as_ref().and_then(\|id\| groups.iter().find(\|g\| &g.id == id)).cloned();
1123	let deadline = at + a.wait_ms.min(model::MAX_POLL_WAIT_MS);
1124	let mut looked = Looked::default();
1125	loop {
1126	if let Some(assignment) = self.claim(&row, group.as_ref(), &mut looked).await? {
1127	poll.assignment = Some(assignment);
1128	return Ok(Outcome::Ok(poll));
1129	}
1130	if now_ms() + POLL_EVERY_MS > deadline {
1131	return Ok(Outcome::Ok(poll));
1132	}
1133	worker::Delay::from(std::time::Duration::from_millis(POLL_EVERY_MS)).await;
1134	}
1135	}
1136
1137	/// Takes the oldest work this runner can do, if there is any.
1138	async fn claim(&self, runner: &RunnerRow, group: Option<&GroupRow>, looked: &mut Looked) -> Result<Option<Assignment>> {
1139	let labels = runner.labels();
1140	let group_name = group.map(\|g\| g.name.clone());
1141	// Workflow jobs.
1142	let mut sql = "SELECT jobs.*, runs.repo AS run_repo FROM jobs JOIN runs ON runs.id = jobs.run_id
1143	WHERE jobs.status = 'queued' AND jobs.labels IS NOT NULL AND lower(jobs.namespace) = ?"
1144	.to_owned();
1145	let mut binds: Vec<JsValue> = vec![runner.workspace.as_str().into()];
1146	if let Some(repo_id) = &runner.repo_id {
1147	sql.push_str(" AND jobs.repo_id = ?");
1148	binds.push(repo_id.as_str().into());
1149	}
1150	// One job of a concurrency group runs at a time.
1151	sql.push_str(
1152	" AND (jobs.concurrency_group IS NULL OR NOT EXISTS (SELECT 1 FROM jobs o WHERE o.repo_id = jobs.repo_id
1153	AND o.concurrency_group = jobs.concurrency_group AND o.status IN ('in_progress', 'calling')))",
1154	);
1155	sql.push_str(" ORDER BY jobs.rowid LIMIT 50");
1156	#[derive(Deserialize)]
1157	struct Queued {
1158	#[serde(flatten)]
1159	job: JobRow,
1160	run_repo: String,
1161	}
1162	let queued = self.db.prepare(sql).bind(&binds)?.all().await?.results::<Queued>()?;
1163	for Queued { job, run_repo } in queued {
1164	let stored: Vec<String> = job.labels.as_deref().and_then(\|l\| serde_json::from_str(l).ok()).unwrap_or_default();
1165	let wanted = Wanted::from_stored(&stored);
1166	if !wanted.matches(&labels, group_name.as_deref()) {
1167	continue;
1168	}
1169	if runner.repo_id.is_none() && !group.is_none_or(\|g\| g.allows(&run_repo)) {
1170	continue;
1171	}
1172	// A host-mode runner takes no public repository's job: it waits
1173	// for a sandboxed runner, and says so when none could take it.
1174	if runner.isolation.as_deref() == Some("host") {
1175	let visibility = self.visibility_once(looked, &job.repo_id).await;
1176	if !model::isolation_allows(runner.isolation.as_deref(), visibility.as_deref()) {
1177	if looked.noted.insert(job.id.clone()) {
1178	self.wait_for_sandbox(&job, &run_repo, &wanted).await?;
1179	}
1180	continue;
1181	}
1182	}
1183	if let Some(max) = job.max_parallel {
1184	let siblings = self
1185	.db
1186	.prepare("SELECT COUNT(*) AS n FROM jobs WHERE run_id = ? AND key = ? AND status = 'in_progress'")
1187	.bind(&[job.run_id.as_str().into(), job.key.as_str().into()])?
1188	.first::<Count>(None)
1189	.await?
1190	.map_or(0, \|c\| c.n);
1191	if siblings >= max {
1192	continue;
1193	}
1194	}
1195	let token = random_hex(24);
1196	let at = now();
1197	let claimed = self
1198	.db
1199	.prepare(
1200	"UPDATE jobs SET status = 'in_progress', token_hash = ?, runner_id = ?, runner_name = ?, reason = NULL, started_at = ?, seen_at = ?
1201	WHERE id = ? AND status = 'queued' RETURNING id",
1202	)
1203	.bind(&[
1204	sha256_hex(&token).into(),
1205	runner.id.as_str().into(),
1206	runner.name.as_str().into(),
1207	at.as_str().into(),
1208	at.as_str().into(),
1209	job.id.as_str().into(),
1210	])?
1211	.first::<Value>(None)
1212	.await?;
1213	if claimed.is_none() {
1214	continue;
1215	}
1216	self.db
1217	.batch(vec![
1218	self.db
1219	.prepare("UPDATE runs SET status = 'in_progress', started_at = COALESCE(started_at, ?) WHERE id = ? AND status = 'queued'")
1220	.bind(&[at.as_str().into(), job.run_id.as_str().into()])?,
1221	self.db
1222	.prepare("UPDATE runners SET work_id = ?, work_kind = 'workflow', spent = ephemeral WHERE id = ?")
1223	.bind(&[job.id.as_str().into(), runner.id.as_str().into()])?,
1224	])
1225	.await?;
1226	self.audit(vec![entry(
1227	runner_actor(runner),
1228	"runner.take_job",
1229	&runner.workspace,
1230	Some(run_repo.clone()),
1231	Some(job.name.clone()),
1232	format!("The self-hosted runner {} took the job {}.", runner.name, job.name),
1233	)])
1234	.await;
1235	// A job that deploys: its run's deployment is under way.
1236	self.job_started(&job.id).await?;
1237	let image = self.job_image(&job).await?;
1238	let visibility = self.visibility(&job.repo_id).await;
1239	return Ok(Some(Assignment {
1240	kind: "workflow".into(),
1241	id: job.id.clone(),
1242	name: job.name.clone(),
1243	repo: run_repo,
1244	timeout_minutes: job.timeout_minutes,
1245	image,
1246	token: Some(token),
1247	env: None,
1248	visibility,
1249	}));
1250	}
1251
1252	// Agent work.
1253	let tasks = self
1254	.db
1255	.prepare("SELECT * FROM runner_tasks WHERE status = 'queued' AND workspace = ? ORDER BY created_at LIMIT 20")
1256	.bind(&[runner.workspace.as_str().into()])?
1257	.all()
1258	.await?
1259	.results::<TaskRow>()?;
1260	for task in tasks {
1261	let wanted = Wanted::of(&serde_json::from_str::<Vec<String>>(&task.labels).unwrap_or_default(), None);
1262	if !wanted.matches(&labels, None) {
1263	continue;
1264	}
1265	// Nor a public repository's agent work.
1266	if runner.isolation.as_deref() == Some("host")
1267	&& let Some(repo_id) = &task.repo_id
1268	{
1269	let visibility = self.visibility_once(looked, repo_id).await;
1270	if !model::isolation_allows(runner.isolation.as_deref(), visibility.as_deref()) {
1271	continue;
1272	}
1273	}
1274	if let Some(repo_id) = &runner.repo_id
1275	&& task.repo_id.as_deref() != Some(repo_id.as_str())
1276	{
1277	continue;
1278	}
1279	if runner.repo_id.is_none() && !group.is_none_or(\|g\| g.allows(&task.repo)) {
1280	continue;
1281	}
1282	let Some(sealed) = task.env.clone() else { continue };
1283	let at = now();
1284	let claimed = self
1285	.db
1286	.prepare(
1287	"UPDATE runner_tasks SET status = 'in_progress', runner_id = ?, runner_name = ?, env = NULL, started_at = ?, seen_at = ?
1288	WHERE id = ? AND status = 'queued' RETURNING id",
1289	)
1290	.bind(&[runner.id.as_str().into(), runner.name.as_str().into(), at.as_str().into(), at.as_str().into(), task.id.as_str().into()])?
1291	.first::<Value>(None)
1292	.await?;
1293	if claimed.is_none() {
1294	continue;
1295	}
1296	self.db
1297	.prepare("UPDATE runners SET work_id = ?, work_kind = 'agent', spent = ephemeral WHERE id = ?")
1298	.bind(&[task.id.as_str().into(), runner.id.as_str().into()])?
1299	.run()
1300	.await?;
1301	let env: Map<String, Value> = self
1302	.sealer
1303	.as_ref()
1304	.and_then(\|sealer\| sealer.open(&sealed, &task.id))
1305	.and_then(\|text\| serde_json::from_str(&text).ok())
1306	.unwrap_or_default();
1307	if env.is_empty() {
1308	self.end_task(&TaskRow { status: "in_progress".into(), ..task.clone() }, 1, "The work's environment could not be opened.").await?;
1309	continue;
1310	}
1311	self.audit(vec![entry(
1312	runner_actor(runner),
1313	"runner.take_task",
1314	&runner.workspace,
1315	Some(task.repo.clone()),
1316	Some(task.kind.clone()),
1317	format!("The self-hosted runner {} took agent work: {}.", runner.name, task.title),
1318	)])
1319	.await;
1320	let visibility = match &task.repo_id {
1321	Some(repo_id) => self.visibility(repo_id).await,
1322	None => None,
1323	};
1324	return Ok(Some(Assignment {
1325	kind: "agent".into(),
1326	id: task.id.clone(),
1327	name: task.title.clone(),
1328	repo: task.repo.clone(),
1329	timeout_minutes: task.timeout_minutes,
1330	image: None,
1331	token: None,
1332	env: Some(env),
1333	visibility,
1334	}));
1335	}
1336	Ok(None)
1337	}
1338
1339	/// [`Self::visibility`], asked once per repository in a poll.
1340	async fn visibility_once(&self, looked: &mut Looked, repo_id: &str) -> Option<String> {
1341	if let Some(known) = looked.visibilities.get(repo_id) {
1342	return known.clone();
1343	}
1344	let found = self.visibility(repo_id).await;
1345	looked.visibilities.insert(repo_id.to_owned(), found.clone());
1346	found
1347	}
1348
1349	/// A public repository's job that a host-mode runner passed over: when
1350	/// no online runner that may take it could (its labels, group,
1351	/// repository, and isolation), its page says it waits for a sandboxed
1352	/// one. Written only when that changes what it says.
1353	async fn wait_for_sandbox(&self, job: &JobRow, run_repo: &str, wanted: &Wanted) -> Result<()> {
1354	let at = now_ms();
1355	let groups = self.group_rows(&job.namespace.to_lowercase()).await?;
1356	let runners = self.runner_rows(&job.namespace.to_lowercase()).await?;
1357	let sandboxed = runners.iter().any(\|other\| takes_public_job(other, &groups, &job.repo_id, run_repo, wanted, at));
1358	if sandboxed {
1359	return Ok(());
1360	}
1361	self.db
1362	.prepare("UPDATE jobs SET reason = ? WHERE id = ? AND status = 'queued' AND reason IS NOT ?")
1363	.bind(&[model::WAITING_FOR_SANDBOX.into(), job.id.as_str().into(), model::WAITING_FOR_SANDBOX.into()])?
1364	.run()
1365	.await?;
1366	Ok(())
1367	}
1368
1369	/// A repository's visibility for a runner, `public` or `private`; none
1370	/// when it cannot be read, and the runner decides without it.
1371	async fn visibility(&self, repo_id: &str) -> Option<String> {
1372	match self.repo_by_id(repo_id).await {
1373	Ok(Some((repo, _))) => Some(if repo.is_private { "private" } else { "public" }.to_owned()),
1374	_ => None,
1375	}
1376	}
1377
1378	/// The image a job's `container:` names, when it names one plainly.
1379	async fn job_image(&self, job: &JobRow) -> Result<Option<String>> {
1380	let Some(run) = self.run_row(&job.run_id).await? else { return Ok(None) };
1381	let raw = match job.callee() {
1382	Some((_, spec, _)) => spec.raw,
1383	None => match g1t_actions::workflow::parse(&run.source) {
1384	Ok(workflow) => match workflow.jobs.into_iter().find(\|j\| j.id == job.key) {
1385	Some(spec) => spec.raw,
1386	None => return Ok(None),
1387	},
1388	Err(_) => return Ok(None),
1389	},
1390	};
1391	let image = match raw.get("container") {
1392	Some(Value::String(image)) => Some(image.clone()),
1393	Some(Value::Object(container)) => container.get("image").and_then(Value::as_str).map(str::to_owned),
1394	_ => None,
1395	};
1396	Ok(image.filter(\|image\| !image.contains("${{") && !image.trim().is_empty()))
1397	}
1398
1399	/// `runner_finished`.
1400	pub async fn runner_finished(&self, a: FinishedArgs) -> Result<Outcome<bool>> {
1401	let Some((row, _)) = self.authenticated(&a.auth).await? else {
1402	return Ok(Self::unknown_runner());
1403	};
1404	let reason = a.reason.clone().filter(\|r\| !r.trim().is_empty()).map(\|r\| r.chars().take(2000).collect::<String>());
1405	let mut changed = false;
1406	if let Some(job) = self.db.prepare("SELECT * FROM jobs WHERE id = ?").bind(&[a.id.as_str().into()])?.first::<JobRow>(None).await? {
1407	if job.runner_id.as_deref() == Some(row.id.as_str()) && job.status == "in_progress" {
1408	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));
1409	self.finish_job(&job.id, "failure", Some(&why), None).await?;
1410	changed = true;
1411	}
1412	} else if let Some(task) = self.db.prepare("SELECT * FROM runner_tasks WHERE id = ?").bind(&[a.id.as_str().into()])?.first::<TaskRow>(None).await?
1413	&& task.runner_id.as_deref() == Some(row.id.as_str())
1414	&& task.status == "in_progress"
1415	{
1416	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) });
1417	self.end_task(&task, a.exit_code, &why).await?;
1418	changed = true;
1419	}
1420	self.db
1421	.prepare("UPDATE runners SET work_id = NULL, work_kind = NULL, last_seen_at = ? WHERE id = ? AND work_id = ?")
1422	.bind(&[now().into(), row.id.as_str().into(), a.id.as_str().into()])?
1423	.run()
1424	.await?;
1425	Ok(Outcome::Ok(changed))
1426	}
1427
1428	/// `runner_remove_self`.
1429	pub async fn runner_remove_self(&self, a: RemoveSelfArgs) -> Result<Outcome<bool>> {
1430	let Some((row, _)) = self.authenticated(&a.auth).await? else {
1431	return Ok(Self::unknown_runner());
1432	};
1433	self.forget_runner(&row, &format!("The runner {} was removed from its machine.", row.name)).await?;
1434	self.audit(vec![entry(
1435	runner_actor(&row),
1436	"remove_runner",
1437	&row.workspace,
1438	row.repo.clone(),
1439	Some(row.name.clone()),
1440	format!("The self-hosted runner {} removed itself.", row.name),
1441	)])
1442	.await;
1443	Ok(Outcome::Ok(true))
1444	}
1445
1446	/// A self-hosted job finished, however it did: its runner is free, and
1447	/// its time goes on the workspace's usage as self-hosted, at $0.
1448	pub async fn released(&self, job: &JobRow) -> Result<()> {
1449	let Some(runner_id) = &job.runner_id else { return Ok(()) };
1450	self.db
1451	.prepare("UPDATE runners SET work_id = NULL, work_kind = NULL WHERE id = ? AND work_id = ?")
1452	.bind(&[runner_id.as_str().into(), job.id.as_str().into()])?
1453	.run()
1454	.await?;
1455	let (Some(started), Some(finished)) = (job.started_at.as_deref(), job.finished_at.as_deref().map(str::to_owned).or_else(\|\| Some(now()))) else {
1456	return Ok(());
1457	};
1458	let seconds = match (g1t_contracts::time::parse_rfc3339(started), g1t_contracts::time::parse_rfc3339(&finished)) {
1459	(Some(a), Some(b)) if b > a => ((b - a) / 1000).max(1),
1460	_ => 1,
1461	};
1462	let run = self.run_row(&job.run_id).await?;
1463	let repo = run.as_ref().map(\|r\| r.repo.clone());
1464	let recorded: Result<Outcome<bool>> = g1t_kit::call(
1465	&self.billing,
1466	"record_sandbox",
1467	&RecordSandboxArgs {
1468	workspace: job.namespace.to_lowercase(),
1469	seconds: u32::try_from(seconds).unwrap_or(u32::MAX),
1470	description: format!(
1471	"{} in {} on the self-hosted runner {}",
1472	job.name,
1473	repo.clone().unwrap_or_default(),
1474	job.runner_name.clone().unwrap_or_default()
1475	),
1476	repo,
1477	reference: format!("selfhosted/{}/{}", job.id, started),
1478	kind: Some(ComputeKind::Workflow),
1479	cpu_seconds: None,
1480	// A workflow job on a self-hosted runner: no agent ran it and nobody asked an agent for it.
1481	agent: None,
1482	asked_by: None,
1483	reservation_id: None,
1484	self_hosted: true,
1485	instance: None,
1486	},
1487	)
1488	.await;
1489	if let Err(error) = recorded {
1490	worker::console_error!("actions: self-hosted time not recorded for {}: {error}", job.id);
1491	}
1492	Ok(())
1493	}
1494
1495	/// `runner` and `RUNNER_*` for a job on a self-hosted runner.
1496	pub async fn runner_context_for(&self, runner_id: &str, variables: &mut Map<String, Value>) -> Result<Value> {
1497	let mut context = g1t_actions::events::runner_context();
1498	let Some(runner) = self.runner_by_id(runner_id).await? else { return Ok(context) };
1499	let os = match runner.os.as_str() {
1500	"macos" => "macOS",
1501	"windows" => "Windows",
1502	_ => "Linux",
1503	};
1504	let arch = if runner.arch == "arm64" { "ARM64" } else { "X64" };
1505	context["name"] = json!(runner.name);
1506	context["os"] = json!(os);
1507	context["arch"] = json!(arch);
1508	context["environment"] = json!("self-hosted");
1509	{
1510	let vars = variables;
1511	vars.insert("RUNNER_NAME".into(), json!(runner.name));
1512	vars.insert("RUNNER_OS".into(), json!(os));
1513	vars.insert("RUNNER_ARCH".into(), json!(arch));
1514	vars.insert("RUNNER_ENVIRONMENT".into(), json!("self-hosted"));
1515	}
1516	Ok(context)
1517	}
1518
1519	// --- Agent work, for the runner service ----------------------------------
1520
1521	/// `runner_route`: the labels g1t's own work in `repo` runs on, when
1522	/// its workspace (or the repository) sends it to self-hosted runners.
1523	pub async fn runner_route(&self, a: RouteArgs) -> Result<Option<Vec<String>>> {
1524	let repo_id = match self.repo_by_path(&a.repo).await? {
1525	Some(repo) => Some(repo.id),
1526	None => None,
1527	};
1528	let settings = self.effective_runner_settings(&a.workspace, repo_id.as_deref()).await?;
1529	Ok(settings.agents_on_self_hosted.then_some(settings.agent_labels))
1530	}
1531
1532	async fn repo_by_path(&self, path: &RepoPath) -> Result<Option<Repo>> {
1533	let Some(actor) = self.workspace_actor(&path.namespace).await? else { return Ok(None) };
1534	self.visible_repo(path, &Some(actor)).await
1535	}
1536
1537	/// `enqueue_task`.
1538	pub async fn enqueue_task(&self, a: EnqueueTaskArgs) -> Result<Outcome<String>> {
1539	let Some(sealer) = &self.sealer else {
1540	return Ok(fail(FailureCode::Conflict, "Agent work cannot be handed to self-hosted runners yet: g1t's key for it is not set."));
1541	};
1542	let id = new_id("rtk", now_ms());
1543	let sealed = sealer.seal(&serde_json::to_string(&a.env)?, &id);
1544	let repo_id = self.repo_by_path(&a.repo).await?.map(\|r\| r.id);
1545	let labels = match model::runner_labels(&a.labels, model::SELF_HOSTED, model::SELF_HOSTED) {
1546	Ok(labels) => labels,
1547	Err(problem) => return Ok(fail(FailureCode::Invalid, problem)),
1548	};
1549	let at = now();
1550	let inserted = self
1551	.db
1552	.prepare(
1553	"INSERT OR IGNORE INTO runner_tasks (id, sandbox, workspace, repo_id, repo, kind, title, labels, env, timeout_minutes, status, created_at, run_id)
1554	VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, 'queued', ?, ?) RETURNING id",
1555	)
1556	.bind(&[
1557	id.as_str().into(),
1558	a.sandbox.as_str().into(),
1559	a.workspace.to_lowercase().into(),
1560	optional(repo_id.as_deref()),
1561	format!("{}/{}", a.repo.namespace, a.repo.name).into(),
1562	a.kind.as_str().into(),
1563	a.title.chars().take(300).collect::<String>().into(),
1564	serde_json::to_string(&labels)?.into(),
1565	sealed.into(),
1566	a.timeout_minutes.max(1).into(),
1567	at.into(),
1568	optional(a.run_id.as_deref().filter(\|id\| !id.is_empty())),
1569	])?
1570	.first::<Value>(None)
1571	.await?;
1572	if inserted.is_none() {
1573	return Ok(fail(FailureCode::Conflict, "That sandbox already handed work to a self-hosted runner."));
1574	}
1575	Ok(Outcome::Ok(id))
1576	}
1577
1578	/// `cancel_task`: the sandbox stopped it. Its runner hears on its next
1579	/// poll; the sandbox does its own ending.
1580	pub async fn cancel_task(&self, a: CancelTaskArgs) -> Result<Outcome<bool>> {
1581	let done = self
1582	.db
1583	.prepare(
1584	"UPDATE runner_tasks SET status = 'completed', exit_code = -1, reason = ?, env = NULL, finished_at = ?
1585	WHERE sandbox = ? AND status != 'completed' RETURNING id",
1586	)
1587	.bind(&[optional(a.reason.as_deref()), now().into(), a.sandbox.as_str().into()])?
1588	.first::<Value>(None)
1589	.await?;
1590	Ok(Outcome::Ok(done.is_some()))
1591	}
1592
1593	/// Ends a task and tells the sandbox waiting on it, once.
1594	async fn end_task(&self, task: &TaskRow, exit_code: i32, reason: &str) -> Result<()> {
1595	let ended = self
1596	.db
1597	.prepare(
1598	"UPDATE runner_tasks SET status = 'completed', exit_code = ?, reason = ?, env = NULL, finished_at = ?
1599	WHERE id = ? AND status != 'completed' RETURNING id",
1600	)
1601	.bind(&[exit_code.into(), optional(Some(reason).filter(\|r\| !r.is_empty())), now().into(), task.id.as_str().into()])?
1602	.first::<Value>(None)
1603	.await?;
1604	if ended.is_none() {
1605	return Ok(());
1606	}
1607	if let Some(runner_id) = &task.runner_id {
1608	self.db
1609	.prepare("UPDATE runners SET work_id = NULL, work_kind = NULL WHERE id = ? AND work_id = ?")
1610	.bind(&[runner_id.as_str().into(), task.id.as_str().into()])?
1611	.run()
1612	.await?;
1613	}
1614	let told: Result<Value> = g1t_kit::call(
1615	&self.runner,
1616	"task_ended",
1617	&json!({
1618	"sandbox": task.sandbox,
1619	"exitCode": exit_code,
1620	"reason": if reason.is_empty() { Value::Null } else { json!(reason) },
1621	"runner": task.runner_name,
1622	}),
1623	)
1624	.await;
1625	if let Err(error) = told {
1626	worker::console_error!("actions: the sandbox waiting on task {} was not told: {error}", task.id);
1627	}
1628	Ok(())
1629	}
1630
1631	/// `runner_activity`: what runs for the workspace now, in g1t's
1632	/// sandboxes and handed to its own runners. Its owners only, as its
1633	/// runners are.
1634	pub async fn runner_activity(&self, a: RunnerActivityArgs) -> Result<Outcome<RunnerActivity>> {
1635	let owner = RunnersOwner { repo: None, workspace: Some(a.workspace.clone()) };
1636	let place = check!(self.runner_place(&a.actor, &owner, false).await?);
1637	let workspace = place.workspace;
1638	let cloud_jobs = self
1639	.db
1640	.prepare(
1641	"SELECT jobs.id, jobs.name, jobs.run_id, runs.repo, jobs.started_at FROM jobs JOIN runs ON runs.id = jobs.run_id
1642	WHERE lower(jobs.namespace) = ? AND jobs.status = 'in_progress' AND jobs.labels IS NULL
1643	ORDER BY jobs.started_at DESC LIMIT ?",
1644	)
1645	.bind(&[workspace.as_str().into(), ACTIVITY_LIMIT.into()])?
1646	.all()
1647	.await?
1648	.results::<CloudJob>()?;
1649	let cloud_jobs_queued = self.queued_jobs(&workspace, false).await?;
1650	let self_hosted_jobs_queued = self.queued_jobs(&workspace, true).await?;
1651	let handed_over = self
1652	.db
1653	.prepare(
1654	"SELECT id, run_id, kind, title, repo, status, runner_name, created_at, started_at FROM runner_tasks
1655	WHERE status IN ('queued', 'in_progress') AND workspace = ?
1656	ORDER BY created_at DESC LIMIT ?",
1657	)
1658	.bind(&[workspace.as_str().into(), ACTIVITY_LIMIT.into()])?
1659	.all()
1660	.await?
1661	.results::<HandedOverTask>()?;
1662	Ok(Outcome::Ok(RunnerActivity { cloud_jobs, cloud_jobs_queued, handed_over, self_hosted_jobs_queued }))
1663	}
1664
1665	/// The workspace's workflow jobs queued for its own runners
1666	/// (`labelled`), or for g1t's sandboxes.
1667	async fn queued_jobs(&self, workspace: &str, labelled: bool) -> Result<u32> {
1668	let sql = if labelled {
1669	"SELECT COUNT(*) AS n FROM jobs WHERE lower(namespace) = ? AND status = 'queued' AND labels IS NOT NULL"
1670	} else {
1671	"SELECT COUNT(*) AS n FROM jobs WHERE lower(namespace) = ? AND status = 'queued' AND labels IS NULL"
1672	};
1673	Ok(self.db.prepare(sql).bind(&[workspace.into()])?.first::<Count>(None).await?.map_or(0, \|c\| c.n))
1674	}
1675
1676	/// `stuck_jobs`: jobs of the viewer's workspaces that have waited ten
1677	/// minutes or more for a self-hosted runner while none that could take
1678	/// them is online, for Mission control's Needs you.
1679	pub async fn stuck_jobs(&self, a: StuckJobsArgs) -> Result<Vec<StuckJob>> {
1680	let Some(viewer) = a.viewer else { return Ok(Vec::new()) };
1681	let at = now_ms();
1682	let mut out = Vec::new();
1683	for membership in viewer.workspaces.iter().take(20) {
1684	let workspace = membership.slug.to_lowercase();
1685	#[derive(Deserialize)]
1686	struct Waiting {
1687	id: String,
1688	name: String,
1689	run_id: String,
1690	labels: String,
1691	queued_at: Option<String>,
1692	repo: String,
1693	repo_id: String,
1694	}
1695	let waiting = self
1696	.db
1697	.prepare(
1698	"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
1699	WHERE jobs.status = 'queued' AND jobs.labels IS NOT NULL AND lower(jobs.namespace) = ? AND jobs.queued_at < ?
1700	ORDER BY jobs.rowid LIMIT 20",
1701	)
1702	.bind(&[workspace.as_str().into(), rfc3339(at.saturating_sub(10 * 60 * 1000)).into()])?
1703	.all()
1704	.await?
1705	.results::<Waiting>()?;
1706	if waiting.is_empty() {
1707	continue;
1708	}
1709	let groups = self.group_rows(&workspace).await?;
1710	let online: Vec<RunnerRow> = self.runner_rows(&workspace).await?.into_iter().filter(\|r\| r.online(at)).collect();
1711	// A repository's visibility is asked only when a host-mode runner
1712	// is online, since only those pass over a public repository's jobs.
1713	let any_host = online.iter().any(\|runner\| runner.isolation.as_deref() == Some("host"));
1714	let mut looked = Looked::default();
1715	for job in waiting {
1716	let wanted = Wanted::from_stored(&serde_json::from_str::<Vec<String>>(&job.labels).unwrap_or_default());
1717	let visibility = if any_host { self.visibility_once(&mut looked, &job.repo_id).await } else { None };
1718	let served = online.iter().any(\|runner\| {
1719	let group = runner.group_id.as_ref().and_then(\|id\| groups.iter().find(\|g\| &g.id == id));
1720	wanted.matches(&runner.labels(), group.map(\|g\| g.name.as_str()))
1721	&& model::isolation_allows(runner.isolation.as_deref(), visibility.as_deref())
1722	});
1723	if !served {
1724	out.push(StuckJob {
1725	id: job.id,
1726	name: job.name,
1727	run_id: job.run_id,
1728	repo: job.repo,
1729	labels: wanted.describe(),
1730	queued_at: job.queued_at.unwrap_or_default(),
1731	});
1732	}
1733	}
1734	}
1735	Ok(out)
1736	}
1737
1738	// --- Every minute ----------------------------------------------------------
1739
1740	/// Jobs and tasks that waited too long, tasks whose runner went quiet,
1741	/// runners offline for weeks, and registration tokens long expired.
1742	pub async fn sweep_runners(&self, at: u64) -> Result<()> {
1743	let before = \|ms: u64\| rfc3339(at.saturating_sub(ms));
1744	let waited = model::MAX_WAIT_HOURS * 60 * 60 * 1000;
1745	let stale = self
1746	.db
1747	.prepare("SELECT * FROM jobs WHERE status = 'queued' AND labels IS NOT NULL AND queued_at < ? LIMIT 50")
1748	.bind(&[before(waited).into()])?
1749	.all()
1750	.await?
1751	.results::<JobRow>()?;
1752	for job in stale {
1753	let stored: Vec<String> = job.labels.as_deref().and_then(\|l\| serde_json::from_str(l).ok()).unwrap_or_default();
1754	let why = format!(
1755	"No self-hosted runner with labels {} took it within {} hours.",
1756	Wanted::from_stored(&stored).describe(),
1757	model::MAX_WAIT_HOURS
1758	);
1759	self.finish_job(&job.id, "failure", Some(&why), None).await?;
1760	}
1761	let tasks = self
1762	.db
1763	.prepare(
1764	"SELECT * FROM runner_tasks WHERE (status = 'queued' AND created_at < ?) OR (status = 'in_progress' AND seen_at < ?) LIMIT 50",
1765	)
1766	.bind(&[before(waited).into(), before(SILENT_MS).into()])?
1767	.all()
1768	.await?
1769	.results::<TaskRow>()?;
1770	for task in tasks {
1771	let why = if task.status == "queued" {
1772	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)
1773	} else {
1774	format!("The self-hosted runner {} stopped answering.", task.runner_name.clone().unwrap_or_default())
1775	};
1776	self.end_task(&task, 1, &why).await?;
1777	}
1778	let forgotten = self
1779	.db
1780	.prepare("SELECT * FROM runners WHERE COALESCE(last_seen_at, created_at) < ? LIMIT 50")
1781	.bind(&[before(FORGET_OFFLINE_MS).into()])?
1782	.all()
1783	.await?
1784	.results::<RunnerRow>()?;
1785	for row in forgotten {
1786	self.forget_runner(&row, "The runner was offline for 14 days.").await?;
1787	}
1788	// An ephemeral runner that took its job and then went quiet.
1789	let spent = self
1790	.db
1791	.prepare("SELECT * FROM runners WHERE ephemeral = 1 AND spent = 1 AND work_id IS NULL AND COALESCE(last_seen_at, created_at) < ? LIMIT 50")
1792	.bind(&[before(60 * 60 * 1000).into()])?
1793	.all()
1794	.await?
1795	.results::<RunnerRow>()?;
1796	for row in spent {
1797	self.forget_runner(&row, "The ephemeral runner finished its job.").await?;
1798	}
1799	self.db
1800	.prepare("DELETE FROM runner_registrations WHERE expires_at < ?")
1801	.bind(&[before(24 * 60 * 60 * 1000).into()])?
1802	.run()
1803	.await?;
1804	Ok(())
1805	}
1806	}
1807
1808	#[derive(Deserialize)]
1809	struct IdRow {
1810	id: String,
1811	}
1812
1813	/// Whether `runner` is online and could take a public repository's job:
1814	/// not host-mode, and its repository, group and labels let it.
1815	fn takes_public_job(runner: &RunnerRow, groups: &[GroupRow], repo_id: &str, run_repo: &str, wanted: &Wanted, at: u64) -> bool {
1816	let group = runner.group_id.as_ref().and_then(\|id\| groups.iter().find(\|g\| &g.id == id));
1817	runner.online(at)
1818	&& model::isolation_allows(runner.isolation.as_deref(), Some("public"))
1819	&& runner.repo_id.as_deref().is_none_or(\|repo\| repo == repo_id)
1820	&& (runner.repo_id.is_some() \|\| group.is_none_or(\|g\| g.allows(run_repo)))
1821	&& wanted.matches(&runner.labels(), group.map(\|g\| g.name.as_str()))
1822	}
1823
1824	fn group_view(group: &GroupRow, runners: &[RunnerRow]) -> RunnerGroup {
1825	RunnerGroup {
1826	id: group.id.clone(),
1827	name: group.name.clone(),
1828	default: group.is_default != 0,
1829	repositories: group.repositories(),
1830	runners: runners.iter().filter(\|r\| r.group_id.as_deref() == Some(group.id.as_str())).count() as u32,
1831	updated_at: group.updated_at.clone(),
1832	}
1833	}
1834
1835	#[cfg(test)]
1836	mod tests {
1837	use super::*;
1838
1839	fn group(repositories: &[&str]) -> GroupRow {
1840	GroupRow {
1841	id: "rng_1".into(),
1842	name: "Default".into(),
1843	is_default: 1,
1844	repositories: serde_json::to_string(&repositories).unwrap(),
1845	updated_at: "2026-10-06T00:00:00Z".into(),
1846	}
1847	}
1848
1849	#[test]
1850	fn being_seen_is_written_once_a_minute() {
1851	let at = 2_000_000_000_000;
1852	assert!(seen_is_stale(None, at));
1853	assert!(!seen_is_stale(Some(&rfc3339(at - 59_000)), at));
1854	assert!(seen_is_stale(Some(&rfc3339(at - 61_000)), at));
1855	// A runner written down at the last moment is still online until the
1856	// next write, a poll later.
1857	assert!(SEEN_RESOLUTION_MS + model::MAX_POLL_WAIT_MS < model::ONLINE_WITHIN_MS);
1858	assert!(SEEN_RESOLUTION_MS < crate::SILENT_MS);
1859	}
1860
1861	#[test]
1862	fn a_group_lets_its_repositories_use_it() {
1863	assert!(group(&[]).allows("acme/web"));
1864	assert!(group(&["web", "api"]).allows("acme/WEB"));
1865	assert!(group(&["web"]).allows("web"));
1866	assert!(!group(&["web"]).allows("acme/docs"));
1867	}
1868
1869	#[test]
1870	fn names_are_checked() {
1871	assert_eq!(valid_name(" build-01 ").unwrap(), "build-01");
1872	assert!(valid_name("").is_err());
1873	assert!(valid_name("has space").is_err());
1874	assert!(valid_name(&"x".repeat(65)).is_err());
1875	assert!(valid_group_name("GPU machines").is_ok());
1876	assert!(valid_group_name("").is_err());
1877	}
1878
1879	#[test]
1880	fn only_a_sandboxed_runner_that_could_take_it_serves_a_public_job() {
1881	let at = 1_000_000_000;
1882	let runner = \|isolation: Option<&str>\| RunnerRow {
1883	id: "rnr_1".into(),
1884	workspace: "acme".into(),
1885	repo_id: None,
1886	repo: None,
1887	group_id: None,
1888	name: "a".into(),
1889	labels: r#"["self-hosted","linux","x64"]"#.into(),
1890	os: "linux".into(),
1891	arch: "x64".into(),
1892	version: String::new(),
1893	ephemeral: 0,
1894	credential_hash: String::new(),
1895	previous_hash: None,
1896	rotated_at: String::new(),
1897	work_id: None,
1898	work_kind: None,
1899	spent: 0,
1900	last_seen_at: Some(rfc3339(at)),
1901	created_at: String::new(),
1902	created_by: None,
1903	isolation: isolation.map(str::to_owned),
1904	};
1905	let wanted = Wanted::of(&["self-hosted".to_owned(), "linux".to_owned()], None);
1906	let takes = \|row: &RunnerRow\| takes_public_job(row, &[], "rep_1", "acme/web", &wanted, at);
1907	assert!(!takes(&runner(Some("host"))));
1908	assert!(takes(&runner(Some("container"))));
1909	assert!(takes(&runner(Some("nested-container"))));
1910	assert!(takes(&runner(None)), "an older runner, which refuses for itself");
1911	assert!(!takes(&RunnerRow { repo_id: Some("rep_2".into()), ..runner(None) }), "another repository's");
1912	assert!(!takes(&RunnerRow { labels: r#"["self-hosted","macos"]"#.into(), ..runner(None) }));
1913	assert!(!takes(&RunnerRow { last_seen_at: None, ..runner(None) }), "offline");
1914	}
1915
1916	#[test]
1917	fn online_means_seen_in_the_last_ninety_seconds() {
1918	let row = RunnerRow {
1919	id: "rnr_1".into(),
1920	workspace: "acme".into(),
1921	repo_id: None,
1922	repo: None,
1923	group_id: None,
1924	name: "a".into(),
1925	labels: "[]".into(),
1926	os: "linux".into(),
1927	arch: "x64".into(),
1928	version: String::new(),
1929	ephemeral: 0,
1930	credential_hash: String::new(),
1931	previous_hash: None,
1932	rotated_at: String::new(),
1933	work_id: None,
1934	work_kind: None,
1935	spent: 0,
1936	last_seen_at: Some(rfc3339(1_000_000_000)),
1937	created_at: String::new(),
1938	created_by: None,
1939	isolation: None,
1940	};
1941	assert!(row.online(1_000_000_000 + 60_000));
1942	assert!(!row.online(1_000_000_000 + 120_000));
1943	assert_eq!(row.workspace, "acme");
1944	}
1945	}