flagon-io/g1t

public

Where people and agents ship software together. The open-source git platform for the whole job: issues, agents, checks and deploys to the edge.

g1t.sh

git agents cloudflare rust forge mcp

1,986 linesCode BlameRaw

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, DesktopJob, DesktopRunner, DesktopUpdate, 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	// A workspace g1t staff put a hold on registers no new runners.
427	if self.standing(&place.workspace).await.is_some_and(\|standing\| standing.held) {
428	return Ok(fail(
429	FailureCode::Paused,
430	"g1t staff put a hold on this workspace, so it cannot register new runners. Write to hey@flagon.io if you think this is a mistake.",
431	));
432	}
433	let made = self
434	.db
435	.prepare("SELECT COUNT(*) AS n FROM runner_registrations WHERE workspace = ? AND created_at > ?")
436	.bind(&[place.workspace.as_str().into(), rfc3339(now_ms().saturating_sub(60 * 60 * 1000)).into()])?
437	.first::<Count>(None)
438	.await?
439	.map_or(0, \|c\| c.n);
440	if made >= model::MAX_TOKENS_PER_HOUR {
441	return Ok(fail(FailureCode::Conflict, "Too many registration tokens in the last hour. Each lasts an hour and registers any number of runners."));
442	}
443	let group = match (&place.repo, a.group.as_deref().map(str::trim).filter(\|g\| !g.is_empty())) {
444	(Some(_), Some(_)) => return Ok(fail(FailureCode::Invalid, "A repository's own runners are in no group.")),
445	(Some(_), None) => None,
446	(None, Some(wanted)) => match self.find_group(&place.workspace, wanted).await? {
447	Some(group) => Some(group),
448	None => return Ok(fail(FailureCode::NotFound, format!("There is no runner group called {wanted}."))),
449	},
450	(None, None) => Some(self.default_group(&place.workspace).await?),
451	};
452	let token = format!("{}{}", model::REGISTRATION_PREFIX, random_hex(24));
453	let at = now_ms();
454	let expires_at = rfc3339(at + model::REGISTRATION_TTL_SECONDS * 1000);
455	let repo_name = place.repo.as_ref().map(\|r\| format!("{}/{}", r.namespace, r.name));
456	self.db
457	.prepare(
458	"INSERT INTO runner_registrations (token_hash, workspace, repo_id, repo, group_id, created_by, created_at, expires_at)
459	VALUES (?, ?, ?, ?, ?, ?, ?, ?)",
460	)
461	.bind(&[
462	hash(&token).into(),
463	place.workspace.as_str().into(),
464	optional(place.repo.as_ref().map(\|r\| r.id.as_str())),
465	optional(repo_name.as_deref()),
466	optional(group.as_ref().map(\|g\| g.id.as_str())),
467	a.actor.username.as_str().into(),
468	rfc3339(at).into(),
469	expires_at.as_str().into(),
470	])?
471	.run()
472	.await?;
473	self.audit(vec![entry(
474	AuditActor::of(&a.actor),
475	"create_runner_registration_token",
476	&place.workspace,
477	repo_name.clone(),
478	None,
479	format!("Made a runner registration token{}.", group.as_ref().map(\|g\| format!(" for group {}", g.name)).unwrap_or_default()),
480	)])
481	.await;
482	Ok(Outcome::Ok(RegistrationToken {
483	token,
484	expires_at,
485	workspace: place.workspace,
486	repo: repo_name,
487	group: group.map(\|g\| g.name),
488	url: crate::SITE.to_owned(),
489	}))
490	}
491
492	/// `remove_runner`: a person removing one.
493	pub async fn remove_runner(&self, a: RemoveRunnerArgs) -> Result<Outcome<bool>> {
494	let place = check!(self.runner_place(&a.actor, &a.owner, true).await?);
495	let Some(row) = self.runner_by_id(&a.id).await? else {
496	return Ok(fail(FailureCode::NotFound, "There is no such runner."));
497	};
498	let ours = row.workspace == place.workspace
499	&& match &place.repo {
500	Some(repo) => row.repo_id.as_deref() == Some(repo.id.as_str()),
501	None => row.repo_id.is_none(),
502	};
503	if !ours {
504	return Ok(fail(FailureCode::NotFound, "There is no such runner."));
505	}
506	self.forget_runner(&row, &format!("{} removed the runner {}.", a.actor.username, row.name)).await?;
507	self.audit(vec![entry(
508	AuditActor::of(&a.actor),
509	"remove_runner",
510	&row.workspace,
511	row.repo.clone(),
512	Some(row.name.clone()),
513	format!("Removed the self-hosted runner {}.", row.name),
514	)])
515	.await;
516	Ok(Outcome::Ok(true))
517	}
518
519	/// Removes a runner: what it was running fails, and its credential
520	/// stops working at once.
521	async fn forget_runner(&self, row: &RunnerRow, why: &str) -> Result<()> {
522	self.db.prepare("DELETE FROM runners WHERE id = ?").bind(&[row.id.as_str().into()])?.run().await?;
523	self.abandon_work(&row.id, why).await
524	}
525
526	/// Fails whatever a runner was running.
527	async fn abandon_work(&self, runner_id: &str, why: &str) -> Result<()> {
528	let jobs = self
529	.db
530	.prepare("SELECT id FROM jobs WHERE runner_id = ? AND status = 'in_progress'")
531	.bind(&[runner_id.into()])?
532	.all()
533	.await?
534	.results::<IdRow>()?;
535	for job in jobs {
536	self.finish_job(&job.id, "failure", Some(why), None).await?;
537	}
538	let tasks = self
539	.db
540	.prepare("SELECT * FROM runner_tasks WHERE runner_id = ? AND status = 'in_progress'")
541	.bind(&[runner_id.into()])?
542	.all()
543	.await?
544	.results::<TaskRow>()?;
545	for task in tasks {
546	self.end_task(&task, 1, why).await?;
547	}
548	Ok(())
549	}
550
551	/// `runner_groups`.
552	pub async fn runner_groups(&self, a: RunnerGroupsArgs) -> Result<Outcome<Vec<RunnerGroup>>> {
553	let owner = RunnersOwner { repo: None, workspace: Some(a.workspace.clone()) };
554	let place = check!(self.runner_place(&a.actor, &owner, false).await?);
555	self.default_group(&place.workspace).await?;
556	let groups = self.group_rows(&place.workspace).await?;
557	let runners = self.runner_rows(&place.workspace).await?;
558	Ok(Outcome::Ok(groups.iter().map(\|g\| group_view(g, &runners)).collect()))
559	}
560
561	/// `set_runner_group`.
562	pub async fn set_runner_group(&self, a: SetRunnerGroupArgs) -> Result<Outcome<RunnerGroup>> {
563	let owner = RunnersOwner { repo: None, workspace: Some(a.workspace.clone()) };
564	let place = check!(self.runner_place(&a.actor, &owner, true).await?);
565	self.default_group(&place.workspace).await?;
566	let groups = self.group_rows(&place.workspace).await?;
567	let name = match a.name.as_deref().map(valid_group_name).transpose() {
568	Ok(name) => name,
569	Err(problem) => return Ok(fail(FailureCode::Invalid, problem)),
570	};
571	let repositories: Option<Vec<String>> = a.repositories.map(\|list\| {
572	let mut out: Vec<String> = Vec::new();
573	for repo in list {
574	let name = repo.trim().rsplit('/').next().unwrap_or_default().to_owned();
575	if !name.is_empty() && !out.iter().any(\|r: &String\| r.eq_ignore_ascii_case(&name)) {
576	out.push(name);
577	}
578	}
579	out
580	});
581	if let Some(name) = &name
582	&& groups.iter().any(\|g\| g.name.eq_ignore_ascii_case(name) && Some(&g.id) != a.id.as_ref())
583	{
584	return Ok(fail(FailureCode::Conflict, format!("There is already a group called {name}.")));
585	}
586	let at = now();
587	let id = match &a.id {
588	Some(id) => {
589	let Some(group) = groups.iter().find(\|g\| &g.id == id) else {
590	return Ok(fail(FailureCode::NotFound, "There is no such runner group."));
591	};
592	self.db
593	.prepare("UPDATE runner_groups SET name = ?, repositories = ?, updated_at = ? WHERE id = ?")
594	.bind(&[
595	name.clone().unwrap_or(group.name.clone()).into(),
596	serde_json::to_string(&repositories.clone().unwrap_or(group.repositories()))?.into(),
597	at.as_str().into(),
598	id.as_str().into(),
599	])?
600	.run()
601	.await?;
602	id.clone()
603	}
604	None => {
605	let Some(name) = name.clone() else {
606	return Ok(fail(FailureCode::Invalid, "A new group needs a name."));
607	};
608	if groups.len() >= 50 {
609	return Ok(fail(FailureCode::Conflict, "A workspace has at most 50 runner groups."));
610	}
611	let id = new_id("rng", now_ms());
612	self.db
613	.prepare(
614	"INSERT INTO runner_groups (id, workspace, name, is_default, repositories, created_at, updated_at) VALUES (?, ?, ?, 0, ?, ?, ?)",
615	)
616	.bind(&[
617	id.as_str().into(),
618	place.workspace.as_str().into(),
619	name.into(),
620	serde_json::to_string(&repositories.clone().unwrap_or_default())?.into(),
621	at.as_str().into(),
622	at.as_str().into(),
623	])?
624	.run()
625	.await?;
626	id
627	}
628	};
629	let groups = self.group_rows(&place.workspace).await?;
630	let runners = self.runner_rows(&place.workspace).await?;
631	let Some(group) = groups.iter().find(\|g\| g.id == id) else {
632	return Ok(fail(FailureCode::NotFound, "There is no such runner group."));
633	};
634	self.audit(vec![entry(
635	AuditActor::of(&a.actor),
636	if a.id.is_some() { "update_runner_group" } else { "create_runner_group" },
637	&place.workspace,
638	None,
639	Some(group.name.clone()),
640	format!(
641	"Runner group {}: {}.",
642	group.name,
643	if group.repositories().is_empty() { "every repository".to_owned() } else { group.repositories().join(", ") }
644	),
645	)])
646	.await;
647	Ok(Outcome::Ok(group_view(group, &runners)))
648	}
649
650	/// `delete_runner_group`.
651	pub async fn delete_runner_group(&self, a: DeleteRunnerGroupArgs) -> Result<Outcome<bool>> {
652	let owner = RunnersOwner { repo: None, workspace: Some(a.workspace.clone()) };
653	let place = check!(self.runner_place(&a.actor, &owner, true).await?);
654	let default = self.default_group(&place.workspace).await?;
655	let Some(group) = self.group_rows(&place.workspace).await?.into_iter().find(\|g\| g.id == a.id) else {
656	return Ok(fail(FailureCode::NotFound, "There is no such runner group."));
657	};
658	if group.is_default != 0 {
659	return Ok(fail(FailureCode::Conflict, "The default group cannot be deleted."));
660	}
661	self.db
662	.batch(vec![
663	self.db.prepare("UPDATE runners SET group_id = ? WHERE group_id = ?").bind(&[default.id.as_str().into(), group.id.as_str().into()])?,
664	self.db.prepare("DELETE FROM runner_groups WHERE id = ?").bind(&[group.id.as_str().into()])?,
665	])
666	.await?;
667	self.audit(vec![entry(
668	AuditActor::of(&a.actor),
669	"delete_runner_group",
670	&place.workspace,
671	None,
672	Some(group.name.clone()),
673	format!("Deleted the runner group {}; its runners joined {}.", group.name, default.name),
674	)])
675	.await;
676	Ok(Outcome::Ok(true))
677	}
678
679	async fn settings_row(&self, owner: &str) -> Result<Option<SettingsRow>> {
680	self.db.prepare("SELECT * FROM runner_settings WHERE owner = ?").bind(&[owner.into()])?.first::<SettingsRow>(None).await
681	}
682
683	fn settings_of(row: Option<SettingsRow>, inherited: bool) -> RunnerSettings {
684	match row {
685	Some(row) => RunnerSettings {
686	agents_on_self_hosted: row.agents != 0,
687	agent_labels: serde_json::from_str(&row.agent_labels).unwrap_or_else(\|_\| vec![model::SELF_HOSTED.to_owned()]),
688	fork_pull_requests: row.fork_pulls != 0,
689	inherited,
690	},
691	None => RunnerSettings {
692	agent_labels: vec![model::SELF_HOSTED.to_owned()],
693	inherited,
694	..RunnerSettings::default()
695	},
696	}
697	}
698
699	/// The settings that apply in a workspace, or in one of its
700	/// repositories: the repository's own if it has them.
701	pub async fn effective_runner_settings(&self, workspace: &str, repo_id: Option<&str>) -> Result<RunnerSettings> {
702	if let Some(repo_id) = repo_id
703	&& let Some(own) = self.settings_row(repo_id).await?
704	{
705	return Ok(Self::settings_of(Some(own), false));
706	}
707	let inherited = repo_id.is_some();
708	Ok(Self::settings_of(self.settings_row(&workspace.to_lowercase()).await?, inherited))
709	}
710
711	/// `runner_settings`.
712	pub async fn runner_settings(&self, a: RunnerSettingsArgs) -> Result<Outcome<RunnerSettings>> {
713	let place = check!(self.runner_place(&a.actor, &a.owner, false).await?);
714	Ok(Outcome::Ok(self.effective_runner_settings(&place.workspace, place.repo.as_ref().map(\|r\| r.id.as_str())).await?))
715	}
716
717	/// `report_desktop`: what a person's desktop app says of its runner,
718	/// in a workspace they own. Kept per computer, replaced each time.
719	pub async fn report_desktop(&self, a: ReportDesktopArgs) -> Result<Outcome<bool>> {
720	let owner = RunnersOwner { repo: None, workspace: Some(a.workspace.clone()) };
721	let place = check!(self.runner_place(&a.actor, &owner, true).await?);
722	let machine = match valid_name(&a.machine) {
723	Ok(name) => name,
724	Err(problem) => return Ok(fail(FailureCode::Invalid, problem)),
725	};
726	if !matches!(a.backend.as_str(), "sandboxed" \| "docker" \| "host") {
727	return Ok(fail(FailureCode::Invalid, "A desktop runs jobs sandboxed, in Docker or on the host."));
728	}
729	let clip = \|text: &str, most: usize\| text.chars().take(most).collect::<String>();
730	let tidy = \|d: &DesktopDiagnosis\| DesktopDiagnosis {
731	code: clip(&d.code, 40),
732	tone: clip(&d.tone, 20),
733	title: clip(&d.title, 200),
734	cause: d.cause.as_deref().map(\|t\| clip(t, 1000)),
735	fix: d.fix.as_deref().map(\|t\| clip(t, 1000)),
736	capabilities: d
737	.capabilities
738	.iter()
739	.take(model::MAX_DESKTOP_CAPABILITIES)
740	.map(\|c\| DesktopCapability {
741	id: clip(&c.id, 20),
742	label: clip(&c.label, 60),
743	tone: clip(&c.tone, 20),
744	title: clip(&c.title, 200),
745	cause: c.cause.as_deref().map(\|t\| clip(t, 1000)),
746	fix: c.fix.as_deref().map(\|t\| clip(t, 1000)),
747	state: c.state.as_deref().map(\|t\| clip(t, 20)),
748	headline: c.headline.as_deref().map(\|t\| clip(t, 200)),
749	about: c.about.as_deref().map(\|t\| clip(t, 200)),
750	enabled: c.enabled,
751	job: c.job.as_ref().map(\|j\| DesktopJob { kind: clip(&j.kind, 20), name: clip(&j.name, 200), repo: clip(&j.repo, 200) }),
752	runner: c.runner.as_deref().map(\|t\| clip(t, 80)),
753	isolation: c.isolation.as_deref().map(\|t\| clip(t, 40)),
754	version: c.version.as_deref().map(\|t\| clip(t, 40)),
755	update: c.update.as_deref().map(\|t\| clip(t, 200)),
756	})
757	.collect(),
758	in_capabilities: d.in_capabilities && !d.capabilities.is_empty(),
759	update: d.update.as_ref().map(\|u\| DesktopUpdate {
760	checked_at: u.checked_at.as_deref().map(\|t\| clip(t, 40)),
761	ready: u.ready.as_deref().map(\|t\| clip(t, 40)),
762	note: u.note.as_deref().map(\|t\| clip(t, 200)),
763	}),
764	};
765	let at = now_ms();
766	let known = self
767	.db
768	.prepare("SELECT COUNT(*) AS n FROM desktop_runners WHERE workspace = ? AND lower(machine) <> lower(?)")
769	.bind(&[place.workspace.as_str().into(), machine.as_str().into()])?
770	.first::<Count>(None)
771	.await?
772	.map_or(0, \|c\| c.n);
773	if known >= model::MAX_DESKTOPS {
774	// The oldest report makes room: a computer long gone.
775	self.db
776	.prepare("DELETE FROM desktop_runners WHERE rowid IN (SELECT rowid FROM desktop_runners WHERE workspace = ? ORDER BY reported_at LIMIT 1)")
777	.bind(&[place.workspace.as_str().into()])?
778	.run()
779	.await?;
780	}
781	self.db
782	.prepare(
783	"INSERT INTO desktop_runners (workspace, machine, person, backend, version, diagnosis, closed, reported_at)
784	VALUES (?, ?, ?, ?, ?, ?, ?, ?)
785	ON CONFLICT (workspace, machine) DO UPDATE SET person = ?3, backend = ?4, version = ?5, diagnosis = ?6, closed = ?7, reported_at = ?8",
786	)
787	.bind(&[
788	place.workspace.as_str().into(),
789	machine.as_str().into(),
790	a.actor.username.as_str().into(),
791	a.backend.as_str().into(),
792	optional(a.version.as_deref().map(\|v\| clip(v, 40)).as_deref()),
793	serde_json::to_string(&tidy(&a.diagnosis))?.into(),
794	serde_json::to_string(&tidy(&a.closed))?.into(),
795	rfc3339(at).into(),
796	])?
797	.run()
798	.await?;
799	Ok(Outcome::Ok(true))
800	}
801
802	/// `desktop_runners`: the workspace's desktops, as each app last said.
803	pub async fn desktop_runners(&self, a: DesktopRunnersArgs) -> Result<Outcome<Vec<DesktopRunner>>> {
804	let owner = RunnersOwner { repo: None, workspace: Some(a.workspace.clone()) };
805	let place = check!(self.runner_place(&a.actor, &owner, false).await?);
806	#[derive(Deserialize)]
807	struct Row {
808	workspace: String,
809	machine: String,
810	person: String,
811	backend: String,
812	version: Option<String>,
813	diagnosis: String,
814	closed: String,
815	reported_at: String,
816	}
817	let rows = self
818	.db
819	.prepare("SELECT * FROM desktop_runners WHERE workspace = ? ORDER BY lower(machine)")
820	.bind(&[place.workspace.as_str().into()])?
821	.all()
822	.await?
823	.results::<Row>()?;
824	let open_since = rfc3339(now_ms().saturating_sub(model::DESKTOP_OPEN_WITHIN_MS));
825	Ok(Outcome::Ok(
826	rows.into_iter()
827	.map(\|row\| DesktopRunner {
828	open: row.reported_at.as_str() > open_since.as_str(),
829	diagnosis: serde_json::from_str(&row.diagnosis).unwrap_or_default(),
830	closed: serde_json::from_str(&row.closed).unwrap_or_default(),
831	workspace: row.workspace,
832	machine: row.machine,
833	person: row.person,
834	backend: row.backend,
835	version: row.version,
836	reported_at: row.reported_at,
837	})
838	.collect(),
839	))
840	}
841
842	/// `set_runner_settings`.
843	pub async fn set_runner_settings(&self, a: SetRunnerSettingsArgs) -> Result<Outcome<RunnerSettings>> {
844	let place = check!(self.runner_place(&a.actor, &a.owner, true).await?);
845	let (owner, scope) = match &place.repo {
846	Some(repo) => (repo.id.clone(), "repository"),
847	None => (place.workspace.clone(), "workspace"),
848	};
849	if a.inherit && place.repo.is_some() {
850	self.db.prepare("DELETE FROM runner_settings WHERE owner = ?").bind(&[owner.as_str().into()])?.run().await?;
851	} else {
852	let current = self.effective_runner_settings(&place.workspace, place.repo.as_ref().map(\|r\| r.id.as_str())).await?;
853	let labels = match &a.agent_labels {
854	Some(given) => match model::runner_labels(given, model::SELF_HOSTED, model::SELF_HOSTED) {
855	Ok(labels) => labels,
856	Err(problem) => return Ok(fail(FailureCode::Invalid, problem)),
857	},
858	None => current.agent_labels.clone(),
859	};
860	let agents = a.agents_on_self_hosted.unwrap_or(current.agents_on_self_hosted);
861	let forks = a.fork_pull_requests.unwrap_or(current.fork_pull_requests);
862	self.db
863	.prepare(
864	"INSERT INTO runner_settings (owner, scope, agents, agent_labels, fork_pulls, updated_at, updated_by)
865	VALUES (?1, ?2, ?3, ?4, ?5, ?6, ?7)
866	ON CONFLICT (owner) DO UPDATE SET agents = ?3, agent_labels = ?4, fork_pulls = ?5, updated_at = ?6, updated_by = ?7",
867	)
868	.bind(&[
869	owner.as_str().into(),
870	scope.into(),
871	u32::from(agents).into(),
872	serde_json::to_string(&labels)?.into(),
873	u32::from(forks).into(),
874	now().into(),
875	a.actor.username.as_str().into(),
876	])?
877	.run()
878	.await?;
879	}
880	let settings = self.effective_runner_settings(&place.workspace, place.repo.as_ref().map(\|r\| r.id.as_str())).await?;
881	self.audit(vec![entry(
882	AuditActor::of(&a.actor),
883	"update_runner_settings",
884	&place.workspace,
885	place.repo.as_ref().map(\|r\| format!("{}/{}", r.namespace, r.name)),
886	None,
887	format!(
888	"Agents on self-hosted runners: {}; labels {}; pull requests from forks: {}.",
889	if settings.agents_on_self_hosted { "on" } else { "off" },
890	settings.agent_labels.join(", "),
891	if settings.fork_pull_requests { "allowed" } else { "not allowed" }
892	),
893	)])
894	.await;
895	Ok(Outcome::Ok(settings))
896	}
897
898	// --- The runner's side ---------------------------------------------------
899
900	/// `runner_register`.
901	pub async fn runner_register(&self, a: RegisterArgs) -> Result<Outcome<Registered>> {
902	let refused = \|\| fail(FailureCode::Unauthenticated, "That registration token is not valid, or it expired. Make a new one under Settings, Runners.");
903	if !a.token.starts_with(model::REGISTRATION_PREFIX) {
904	return Ok(refused());
905	}
906	let registration = self
907	.db
908	.prepare("SELECT * FROM runner_registrations WHERE token_hash = ?")
909	.bind(&[hash(&a.token).into()])?
910	.first::<RegistrationRow>(None)
911	.await?;
912	let at = now_ms();
913	let Some(registration) = registration.filter(\|r\| r.expires_at.as_str() > rfc3339(at).as_str()) else {
914	return Ok(refused());
915	};
916	let name = match valid_name(&a.name) {
917	Ok(name) => name,
918	Err(problem) => return Ok(fail(FailureCode::Invalid, problem)),
919	};
920	let (Some(os), Some(arch)) = (model::os_label(&a.os), model::arch_label(&a.arch)) else {
921	return Ok(fail(FailureCode::Invalid, "A runner's os is linux, macos or windows, and its arch x64 or arm64."));
922	};
923	let labels = match model::runner_labels(&a.labels, os, arch) {
924	Ok(labels) => labels,
925	Err(problem) => return Ok(fail(FailureCode::Invalid, problem)),
926	};
927	let workspace = registration.workspace.clone();
928	// Abuse: so many machines, so fast, are not a team's.
929	let recent = self
930	.db
931	.prepare("SELECT COUNT(*) AS n FROM runners WHERE workspace = ? AND created_at > ?")
932	.bind(&[workspace.as_str().into(), rfc3339(at.saturating_sub(60 * 1000)).into()])?
933	.first::<Count>(None)
934	.await?
935	.map_or(0, \|c\| c.n);
936	if recent >= model::MAX_REGISTRATIONS_PER_MINUTE {
937	return Ok(fail(FailureCode::Conflict, "Too many runners registered in the last minute. Wait a moment and try again."));
938	}
939	let total = self
940	.db
941	.prepare("SELECT COUNT(*) AS n FROM runners WHERE workspace = ?")
942	.bind(&[workspace.as_str().into()])?
943	.first::<Count>(None)
944	.await?
945	.map_or(0, \|c\| c.n);
946	// A group: the one asked for, the token's, or the default.
947	let group = if registration.repo_id.is_some() {
948	None
949	} else {
950	match a.group.as_deref().map(str::trim).filter(\|g\| !g.is_empty()) {
951	Some(wanted) => match self.find_group(&workspace, wanted).await? {
952	Some(group) => Some(group.id),
953	None => return Ok(fail(FailureCode::NotFound, format!("There is no runner group called {wanted}."))),
954	},
955	None => match registration.group_id.clone() {
956	Some(id) => Some(id),
957	None => Some(self.default_group(&workspace).await?.id),
958	},
959	}
960	};
961	let existing = self
962	.db
963	.prepare("SELECT * FROM runners WHERE workspace = ? AND COALESCE(repo_id, '') = ? AND lower(name) = lower(?)")
964	.bind(&[workspace.as_str().into(), registration.repo_id.clone().unwrap_or_default().into(), name.as_str().into()])?
965	.first::<RunnerRow>(None)
966	.await?;
967	if let Some(existing) = &existing {
968	if !a.replace {
969	return Ok(fail(
970	FailureCode::Conflict,
971	format!("A runner called {name} is already registered here. Register with --replace to take its place, or choose another --name."),
972	));
973	}
974	self.forget_runner(existing, &format!("The runner {name} was replaced by a new registration.")).await?;
975	} else if total >= model::MAX_RUNNERS {
976	return Ok(fail(FailureCode::Conflict, format!("A workspace has at most {} self-hosted runners.", model::MAX_RUNNERS)));
977	}
978	// How it keeps work from its machine, if it said (older runners do not).
979	let isolation = a.isolation.as_deref().filter(\|level\| model::ISOLATIONS.contains(level));
980	let id = new_id("rnr", at);
981	let credential = format!("{}{}", model::CREDENTIAL_PREFIX, random_hex(32));
982	let created = rfc3339(at);
983	self.db
984	.prepare(
985	"INSERT INTO runners (id, workspace, repo_id, repo, group_id, name, labels, os, arch, version, ephemeral, credential_hash,
986	rotated_at, last_seen_at, created_at, created_by, isolation)
987	VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)",
988	)
989	.bind(&[
990	id.as_str().into(),
991	workspace.as_str().into(),
992	optional(registration.repo_id.as_deref()),
993	optional(registration.repo.as_deref()),
994	optional(group.as_deref()),
995	name.as_str().into(),
996	serde_json::to_string(&labels)?.into(),
997	os.into(),
998	arch.into(),
999	a.version.chars().take(40).collect::<String>().into(),
1000	u32::from(a.ephemeral).into(),
1001	hash(&credential).into(),
1002	created.as_str().into(),
1003	created.as_str().into(),
1004	created.as_str().into(),
1005	// The person who made the token: a desktop's runner is theirs.
1006	optional(registration.created_by.as_deref()),
1007	optional(isolation),
1008	])?
1009	.run()
1010	.await?;
1011	self.db
1012	.prepare("UPDATE runner_registrations SET used = used + 1 WHERE token_hash = ?")
1013	.bind(&[hash(&a.token).into()])?
1014	.run()
1015	.await?;
1016	let row = self.runner_by_id(&id).await?.ok_or_else(\|\| worker::Error::RustError("the runner was not kept".into()))?;
1017	self.audit(vec![entry(
1018	runner_actor(&row),
1019	"register_runner",
1020	&workspace,
1021	row.repo.clone(),
1022	Some(name.clone()),
1023	format!("Registered the self-hosted runner {name} ({os}, {arch}; labels {}).", labels.join(", ")),
1024	)])
1025	.await;
1026	let groups = self.group_rows(&workspace).await?;
1027	Ok(Outcome::Ok(Registered {
1028	runner: self.runner_view(&row, &groups, at).await?,
1029	credential,
1030	}))
1031	}
1032
1033	/// The runner a call is from, if its credential is that runner's (or
1034	/// the one before a rotation, until the new one is used).
1035	async fn authenticated(&self, auth: &RunnerAuth) -> Result<Option<(RunnerRow, bool)>> {
1036	if !auth.credential.starts_with(model::CREDENTIAL_PREFIX) {
1037	return Ok(None);
1038	}
1039	let Some(row) = self.runner_by_id(&auth.runner).await? else { return Ok(None) };
1040	let given = hash(&auth.credential);
1041	if g1t_secrets::same(&row.credential_hash, &given) {
1042	if row.previous_hash.is_some() {
1043	self.db.prepare("UPDATE runners SET previous_hash = NULL WHERE id = ?").bind(&[row.id.as_str().into()])?.run().await?;
1044	}
1045	return Ok(Some((row, true)));
1046	}
1047	if row.previous_hash.as_deref().is_some_and(\|previous\| g1t_secrets::same(previous, &given)) {
1048	return Ok(Some((row, false)));
1049	}
1050	Ok(None)
1051	}
1052
1053	fn unknown_runner<T>() -> Outcome<T> {
1054	fail(FailureCode::Unauthenticated, "This runner is not registered, or its credential is not valid. Register it again.")
1055	}
1056
1057	/// `runner_poll`.
1058	pub async fn runner_poll(&self, a: PollArgs) -> Result<Outcome<Poll>> {
1059	let Some((row, current)) = self.authenticated(&a.auth).await? else {
1060	return Ok(Self::unknown_runner());
1061	};
1062	let at = now_ms();
1063	let seen = rfc3339(at);
1064	// A new credential every day, handed over on a poll that used the
1065	// current one.
1066	let mut credential = None;
1067	let rotate_after = rfc3339(at.saturating_sub(model::ROTATE_AFTER_MS));
1068	if current && row.rotated_at.as_str() < rotate_after.as_str() {
1069	let fresh = format!("{}{}", model::CREDENTIAL_PREFIX, random_hex(32));
1070	self.db
1071	.prepare("UPDATE runners SET previous_hash = credential_hash, credential_hash = ?, rotated_at = ? WHERE id = ?")
1072	.bind(&[hash(&fresh).into(), seen.as_str().into(), row.id.as_str().into()])?
1073	.run()
1074	.await?;
1075	credential = Some(fresh);
1076	}
1077	// What it holds that it should stop, and the heartbeat for the rest.
1078	let mut cancel = Vec::new();
1079	let mut active: Option<(String, &'static str)> = None;
1080	for id in a.running.iter().take(10) {
1081	if let Some(job) = self.db.prepare("SELECT * FROM jobs WHERE id = ?").bind(&[id.as_str().into()])?.first::<JobRow>(None).await? {
1082	if job.status == "in_progress" && job.runner_id.as_deref() == Some(row.id.as_str()) {
1083	if seen_is_stale(job.seen_at.as_deref(), at) {
1084	self.db.prepare("UPDATE jobs SET seen_at = ? WHERE id = ?").bind(&[seen.as_str().into(), id.as_str().into()])?.run().await?;
1085	}
1086	active = Some((id.clone(), "workflow"));
1087	} else {
1088	cancel.push(id.clone());
1089	}
1090	continue;
1091	}
1092	match self.db.prepare("SELECT * FROM runner_tasks WHERE id = ?").bind(&[id.as_str().into()])?.first::<TaskRow>(None).await? {
1093	Some(task) if task.status == "in_progress" && task.runner_id.as_deref() == Some(row.id.as_str()) => {
1094	if seen_is_stale(task.seen_at.as_deref(), at) {
1095	self.db.prepare("UPDATE runner_tasks SET seen_at = ? WHERE id = ?").bind(&[seen.as_str().into(), id.as_str().into()])?.run().await?;
1096	}
1097	active = Some((id.clone(), "agent"));
1098	}
1099	_ => cancel.push(id.clone()),
1100	}
1101	}
1102	// Written when something changed, or once a minute to keep it online.
1103	let version: String = if a.version.is_empty() { row.version.clone() } else { a.version.chars().take(40).collect() };
1104	let work_id = active.as_ref().map(\|(id, _)\| id.as_str());
1105	let work_kind = active.as_ref().map(\|(_, kind)\| *kind);
1106	// How it is isolated, as it says now: a runner re-registered or
1107	// restarted with or without Docker says so on its next poll.
1108	let isolation = a
1109	.isolation
1110	.as_deref()
1111	.filter(\|level\| model::ISOLATIONS.contains(level))
1112	.map(str::to_owned)
1113	.or_else(\|\| row.isolation.clone());
1114	let changed = version != row.version
1115	\|\| work_id != row.work_id.as_deref()
1116	\|\| work_kind != row.work_kind.as_deref()
1117	\|\| isolation != row.isolation;
1118	if changed \|\| seen_is_stale(row.last_seen_at.as_deref(), at) {
1119	self.db
1120	.prepare("UPDATE runners SET last_seen_at = ?, version = ?, work_id = ?, work_kind = ?, isolation = ? WHERE id = ?")
1121	.bind(&[
1122	seen.as_str().into(),
1123	version.into(),
1124	optional(work_id),
1125	optional(work_kind),
1126	optional(isolation.as_deref()),
1127	row.id.as_str().into(),
1128	])?
1129	.run()
1130	.await?;
1131	}
1132	let row = RunnerRow { isolation, ..row };
1133	let mut poll = Poll { assignment: None, cancel, credential, removed: false };
1134	// Busy, or an ephemeral runner that has had its one job.
1135	if active.is_some() \|\| !a.running.is_empty() {
1136	return Ok(Outcome::Ok(poll));
1137	}
1138	if row.ephemeral != 0 && row.spent != 0 {
1139	self.forget_runner(&row, "The ephemeral runner finished its job.").await?;
1140	poll.removed = true;
1141	return Ok(Outcome::Ok(poll));
1142	}
1143	let groups = self.group_rows(&row.workspace).await?;
1144	let group = row.group_id.as_ref().and_then(\|id\| groups.iter().find(\|g\| &g.id == id)).cloned();
1145	let deadline = at + a.wait_ms.min(model::MAX_POLL_WAIT_MS);
1146	let mut looked = Looked::default();
1147	loop {
1148	if let Some(assignment) = self.claim(&row, group.as_ref(), &mut looked).await? {
1149	poll.assignment = Some(assignment);
1150	return Ok(Outcome::Ok(poll));
1151	}
1152	if now_ms() + POLL_EVERY_MS > deadline {
1153	return Ok(Outcome::Ok(poll));
1154	}
1155	worker::Delay::from(std::time::Duration::from_millis(POLL_EVERY_MS)).await;
1156	}
1157	}
1158
1159	/// Takes the oldest work this runner can do, if there is any.
1160	async fn claim(&self, runner: &RunnerRow, group: Option<&GroupRow>, looked: &mut Looked) -> Result<Option<Assignment>> {
1161	let labels = runner.labels();
1162	let group_name = group.map(\|g\| g.name.clone());
1163	// Workflow jobs.
1164	let mut sql = "SELECT jobs.*, runs.repo AS run_repo FROM jobs JOIN runs ON runs.id = jobs.run_id
1165	WHERE jobs.status = 'queued' AND jobs.labels IS NOT NULL AND lower(jobs.namespace) = ?"
1166	.to_owned();
1167	let mut binds: Vec<JsValue> = vec![runner.workspace.as_str().into()];
1168	if let Some(repo_id) = &runner.repo_id {
1169	sql.push_str(" AND jobs.repo_id = ?");
1170	binds.push(repo_id.as_str().into());
1171	}
1172	// One job of a concurrency group runs at a time.
1173	sql.push_str(
1174	" AND (jobs.concurrency_group IS NULL OR NOT EXISTS (SELECT 1 FROM jobs o WHERE o.repo_id = jobs.repo_id
1175	AND o.concurrency_group = jobs.concurrency_group AND o.status IN ('in_progress', 'calling')))",
1176	);
1177	sql.push_str(" ORDER BY jobs.rowid LIMIT 50");
1178	#[derive(Deserialize)]
1179	struct Queued {
1180	#[serde(flatten)]
1181	job: JobRow,
1182	run_repo: String,
1183	}
1184	let mut queued = self.db.prepare(sql).bind(&binds)?.all().await?.results::<Queued>()?;
1185	// A workspace without the plan runs FREE_SELF_HOSTED_JOBS of its jobs
1186	// on its own runners at once (limits.rs). Billing is asked only once
1187	// that many are going, so a runner's poll stays cheap.
1188	if !queued.is_empty() && self.limits.free_self_hosted_jobs > 0 {
1189	let running = self
1190	.db
1191	.prepare("SELECT COUNT(*) AS n FROM jobs WHERE status = 'in_progress' AND runner_id IS NOT NULL AND lower(namespace) = ?")
1192	.bind(&[runner.workspace.as_str().into()])?
1193	.first::<Count>(None)
1194	.await?
1195	.map_or(0, \|c\| c.n);
1196	if !crate::limits::self_hosted_room(running as u64, self.limits.free_self_hosted_jobs)
1197	&& self.standing(&runner.workspace).await.is_some_and(\|standing\| !standing.has_plan)
1198	{
1199	// Its workflow jobs wait for one to finish; anything else the
1200	// runner takes (below) is not held back.
1201	queued.clear();
1202	}
1203	}
1204	for Queued { job, run_repo } in queued {
1205	let stored: Vec<String> = job.labels.as_deref().and_then(\|l\| serde_json::from_str(l).ok()).unwrap_or_default();
1206	let wanted = Wanted::from_stored(&stored);
1207	if !wanted.matches(&labels, group_name.as_deref()) {
1208	continue;
1209	}
1210	if runner.repo_id.is_none() && !group.is_none_or(\|g\| g.allows(&run_repo)) {
1211	continue;
1212	}
1213	// A host-mode runner takes no public repository's job: it waits
1214	// for a sandboxed runner, and says so when none could take it.
1215	if runner.isolation.as_deref() == Some("host") {
1216	let visibility = self.visibility_once(looked, &job.repo_id).await;
1217	if !model::isolation_allows(runner.isolation.as_deref(), visibility.as_deref()) {
1218	if looked.noted.insert(job.id.clone()) {
1219	self.wait_for_sandbox(&job, &run_repo, &wanted).await?;
1220	}
1221	continue;
1222	}
1223	}
1224	if let Some(max) = job.max_parallel {
1225	let siblings = self
1226	.db
1227	.prepare("SELECT COUNT(*) AS n FROM jobs WHERE run_id = ? AND key = ? AND status = 'in_progress'")
1228	.bind(&[job.run_id.as_str().into(), job.key.as_str().into()])?
1229	.first::<Count>(None)
1230	.await?
1231	.map_or(0, \|c\| c.n);
1232	if siblings >= max {
1233	continue;
1234	}
1235	}
1236	let token = random_hex(24);
1237	let at = now();
1238	let claimed = self
1239	.db
1240	.prepare(
1241	"UPDATE jobs SET status = 'in_progress', token_hash = ?, first_token_hash = NULL, runner_id = ?, runner_name = ?, reason = NULL, started_at = ?, seen_at = ?
1242	WHERE id = ? AND status = 'queued' RETURNING id",
1243	)
1244	.bind(&[
1245	sha256_hex(&token).into(),
1246	runner.id.as_str().into(),
1247	runner.name.as_str().into(),
1248	at.as_str().into(),
1249	at.as_str().into(),
1250	job.id.as_str().into(),
1251	])?
1252	.first::<Value>(None)
1253	.await?;
1254	if claimed.is_none() {
1255	continue;
1256	}
1257	self.db
1258	.batch(vec![
1259	self.db
1260	.prepare("UPDATE runs SET status = 'in_progress', started_at = COALESCE(started_at, ?) WHERE id = ? AND status = 'queued'")
1261	.bind(&[at.as_str().into(), job.run_id.as_str().into()])?,
1262	self.db
1263	.prepare("UPDATE runners SET work_id = ?, work_kind = 'workflow', spent = ephemeral WHERE id = ?")
1264	.bind(&[job.id.as_str().into(), runner.id.as_str().into()])?,
1265	])
1266	.await?;
1267	self.audit(vec![entry(
1268	runner_actor(runner),
1269	"runner.take_job",
1270	&runner.workspace,
1271	Some(run_repo.clone()),
1272	Some(job.name.clone()),
1273	format!("The self-hosted runner {} took the job {}.", runner.name, job.name),
1274	)])
1275	.await;
1276	// A job that deploys: its run's deployment is under way.
1277	self.job_started(&job.id).await?;
1278	let image = self.job_image(&job).await?;
1279	let visibility = self.visibility(&job.repo_id).await;
1280	return Ok(Some(Assignment {
1281	kind: "workflow".into(),
1282	id: job.id.clone(),
1283	name: job.name.clone(),
1284	repo: run_repo,
1285	timeout_minutes: job.timeout_minutes,
1286	image,
1287	token: Some(token),
1288	env: None,
1289	visibility,
1290	}));
1291	}
1292
1293	// Agent work.
1294	let tasks = self
1295	.db
1296	.prepare("SELECT * FROM runner_tasks WHERE status = 'queued' AND workspace = ? ORDER BY created_at LIMIT 20")
1297	.bind(&[runner.workspace.as_str().into()])?
1298	.all()
1299	.await?
1300	.results::<TaskRow>()?;
1301	for task in tasks {
1302	let wanted = Wanted::of(&serde_json::from_str::<Vec<String>>(&task.labels).unwrap_or_default(), None);
1303	if !wanted.matches(&labels, None) {
1304	continue;
1305	}
1306	// Nor a public repository's agent work.
1307	if runner.isolation.as_deref() == Some("host")
1308	&& let Some(repo_id) = &task.repo_id
1309	{
1310	let visibility = self.visibility_once(looked, repo_id).await;
1311	if !model::isolation_allows(runner.isolation.as_deref(), visibility.as_deref()) {
1312	continue;
1313	}
1314	}
1315	if let Some(repo_id) = &runner.repo_id
1316	&& task.repo_id.as_deref() != Some(repo_id.as_str())
1317	{
1318	continue;
1319	}
1320	if runner.repo_id.is_none() && !group.is_none_or(\|g\| g.allows(&task.repo)) {
1321	continue;
1322	}
1323	let Some(sealed) = task.env.clone() else { continue };
1324	let at = now();
1325	let claimed = self
1326	.db
1327	.prepare(
1328	"UPDATE runner_tasks SET status = 'in_progress', runner_id = ?, runner_name = ?, env = NULL, started_at = ?, seen_at = ?
1329	WHERE id = ? AND status = 'queued' RETURNING id",
1330	)
1331	.bind(&[runner.id.as_str().into(), runner.name.as_str().into(), at.as_str().into(), at.as_str().into(), task.id.as_str().into()])?
1332	.first::<Value>(None)
1333	.await?;
1334	if claimed.is_none() {
1335	continue;
1336	}
1337	self.db
1338	.prepare("UPDATE runners SET work_id = ?, work_kind = 'agent', spent = ephemeral WHERE id = ?")
1339	.bind(&[task.id.as_str().into(), runner.id.as_str().into()])?
1340	.run()
1341	.await?;
1342	let env: Map<String, Value> = self
1343	.sealer
1344	.as_ref()
1345	.and_then(\|sealer\| sealer.open(&sealed, &task.id))
1346	.and_then(\|text\| serde_json::from_str(&text).ok())
1347	.unwrap_or_default();
1348	if env.is_empty() {
1349	self.end_task(&TaskRow { status: "in_progress".into(), ..task.clone() }, 1, "The work's environment could not be opened.").await?;
1350	continue;
1351	}
1352	self.audit(vec![entry(
1353	runner_actor(runner),
1354	"runner.take_task",
1355	&runner.workspace,
1356	Some(task.repo.clone()),
1357	Some(task.kind.clone()),
1358	format!("The self-hosted runner {} took agent work: {}.", runner.name, task.title),
1359	)])
1360	.await;
1361	let visibility = match &task.repo_id {
1362	Some(repo_id) => self.visibility(repo_id).await,
1363	None => None,
1364	};
1365	return Ok(Some(Assignment {
1366	kind: "agent".into(),
1367	id: task.id.clone(),
1368	name: task.title.clone(),
1369	repo: task.repo.clone(),
1370	timeout_minutes: task.timeout_minutes,
1371	image: None,
1372	token: None,
1373	env: Some(env),
1374	visibility,
1375	}));
1376	}
1377	Ok(None)
1378	}
1379
1380	/// [`Self::visibility`], asked once per repository in a poll.
1381	async fn visibility_once(&self, looked: &mut Looked, repo_id: &str) -> Option<String> {
1382	if let Some(known) = looked.visibilities.get(repo_id) {
1383	return known.clone();
1384	}
1385	let found = self.visibility(repo_id).await;
1386	looked.visibilities.insert(repo_id.to_owned(), found.clone());
1387	found
1388	}
1389
1390	/// A public repository's job that a host-mode runner passed over: when
1391	/// no online runner that may take it could (its labels, group,
1392	/// repository, and isolation), its page says it waits for a sandboxed
1393	/// one. Written only when that changes what it says.
1394	async fn wait_for_sandbox(&self, job: &JobRow, run_repo: &str, wanted: &Wanted) -> Result<()> {
1395	let at = now_ms();
1396	let groups = self.group_rows(&job.namespace.to_lowercase()).await?;
1397	let runners = self.runner_rows(&job.namespace.to_lowercase()).await?;
1398	let sandboxed = runners.iter().any(\|other\| takes_public_job(other, &groups, &job.repo_id, run_repo, wanted, at));
1399	if sandboxed {
1400	return Ok(());
1401	}
1402	self.db
1403	.prepare("UPDATE jobs SET reason = ? WHERE id = ? AND status = 'queued' AND reason IS NOT ?")
1404	.bind(&[model::WAITING_FOR_SANDBOX.into(), job.id.as_str().into(), model::WAITING_FOR_SANDBOX.into()])?
1405	.run()
1406	.await?;
1407	Ok(())
1408	}
1409
1410	/// A repository's visibility for a runner, `public` or `private`; none
1411	/// when it cannot be read, and the runner decides without it.
1412	async fn visibility(&self, repo_id: &str) -> Option<String> {
1413	match self.repo_by_id(repo_id).await {
1414	Ok(Some((repo, _))) => Some(if repo.is_private { "private" } else { "public" }.to_owned()),
1415	_ => None,
1416	}
1417	}
1418
1419	/// The image a job's `container:` names, when it names one plainly.
1420	async fn job_image(&self, job: &JobRow) -> Result<Option<String>> {
1421	let Some(run) = self.run_row(&job.run_id).await? else { return Ok(None) };
1422	let raw = match job.callee() {
1423	Some((_, spec, _)) => spec.raw,
1424	None => match g1t_actions::workflow::parse(&run.source) {
1425	Ok(workflow) => match workflow.jobs.into_iter().find(\|j\| j.id == job.key) {
1426	Some(spec) => spec.raw,
1427	None => return Ok(None),
1428	},
1429	Err(_) => return Ok(None),
1430	},
1431	};
1432	let image = match raw.get("container") {
1433	Some(Value::String(image)) => Some(image.clone()),
1434	Some(Value::Object(container)) => container.get("image").and_then(Value::as_str).map(str::to_owned),
1435	_ => None,
1436	};
1437	Ok(image.filter(\|image\| !image.contains("${{") && !image.trim().is_empty()))
1438	}
1439
1440	/// `runner_finished`.
1441	pub async fn runner_finished(&self, a: FinishedArgs) -> Result<Outcome<bool>> {
1442	let Some((row, _)) = self.authenticated(&a.auth).await? else {
1443	return Ok(Self::unknown_runner());
1444	};
1445	let reason = a.reason.clone().filter(\|r\| !r.trim().is_empty()).map(\|r\| r.chars().take(2000).collect::<String>());
1446	let mut changed = false;
1447	if let Some(job) = self.db.prepare("SELECT * FROM jobs WHERE id = ?").bind(&[a.id.as_str().into()])?.first::<JobRow>(None).await? {
1448	if job.runner_id.as_deref() == Some(row.id.as_str()) && job.status == "in_progress" {
1449	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));
1450	self.finish_job(&job.id, "failure", Some(&why), None).await?;
1451	changed = true;
1452	}
1453	} else if let Some(task) = self.db.prepare("SELECT * FROM runner_tasks WHERE id = ?").bind(&[a.id.as_str().into()])?.first::<TaskRow>(None).await?
1454	&& task.runner_id.as_deref() == Some(row.id.as_str())
1455	&& task.status == "in_progress"
1456	{
1457	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) });
1458	self.end_task(&task, a.exit_code, &why).await?;
1459	changed = true;
1460	}
1461	self.db
1462	.prepare("UPDATE runners SET work_id = NULL, work_kind = NULL, last_seen_at = ? WHERE id = ? AND work_id = ?")
1463	.bind(&[now().into(), row.id.as_str().into(), a.id.as_str().into()])?
1464	.run()
1465	.await?;
1466	Ok(Outcome::Ok(changed))
1467	}
1468
1469	/// `runner_remove_self`.
1470	pub async fn runner_remove_self(&self, a: RemoveSelfArgs) -> Result<Outcome<bool>> {
1471	let Some((row, _)) = self.authenticated(&a.auth).await? else {
1472	return Ok(Self::unknown_runner());
1473	};
1474	self.forget_runner(&row, &format!("The runner {} was removed from its machine.", row.name)).await?;
1475	self.audit(vec![entry(
1476	runner_actor(&row),
1477	"remove_runner",
1478	&row.workspace,
1479	row.repo.clone(),
1480	Some(row.name.clone()),
1481	format!("The self-hosted runner {} removed itself.", row.name),
1482	)])
1483	.await;
1484	Ok(Outcome::Ok(true))
1485	}
1486
1487	/// A self-hosted job finished, however it did: its runner is free, and
1488	/// its time goes on the workspace's usage as self-hosted, at $0.
1489	pub async fn released(&self, job: &JobRow) -> Result<()> {
1490	let Some(runner_id) = &job.runner_id else { return Ok(()) };
1491	self.db
1492	.prepare("UPDATE runners SET work_id = NULL, work_kind = NULL WHERE id = ? AND work_id = ?")
1493	.bind(&[runner_id.as_str().into(), job.id.as_str().into()])?
1494	.run()
1495	.await?;
1496	let (Some(started), Some(finished)) = (job.started_at.as_deref(), job.finished_at.as_deref().map(str::to_owned).or_else(\|\| Some(now()))) else {
1497	return Ok(());
1498	};
1499	let seconds = match (g1t_contracts::time::parse_rfc3339(started), g1t_contracts::time::parse_rfc3339(&finished)) {
1500	(Some(a), Some(b)) if b > a => ((b - a) / 1000).max(1),
1501	_ => 1,
1502	};
1503	let run = self.run_row(&job.run_id).await?;
1504	let repo = run.as_ref().map(\|r\| r.repo.clone());
1505	let recorded: Result<Outcome<bool>> = g1t_kit::call(
1506	&self.billing,
1507	"record_sandbox",
1508	&RecordSandboxArgs {
1509	workspace: job.namespace.to_lowercase(),
1510	seconds: u32::try_from(seconds).unwrap_or(u32::MAX),
1511	description: format!(
1512	"{} in {} on the self-hosted runner {}",
1513	job.name,
1514	repo.clone().unwrap_or_default(),
1515	job.runner_name.clone().unwrap_or_default()
1516	),
1517	repo,
1518	reference: format!("selfhosted/{}/{}", job.id, started),
1519	kind: Some(ComputeKind::Workflow),
1520	cpu_seconds: None,
1521	// A workflow job on a self-hosted runner: no agent ran it and nobody asked an agent for it.
1522	agent: None,
1523	asked_by: None,
1524	reservation_id: None,
1525	self_hosted: true,
1526	instance: None,
1527	},
1528	)
1529	.await;
1530	if let Err(error) = recorded {
1531	worker::console_error!("actions: self-hosted time not recorded for {}: {error}", job.id);
1532	}
1533	Ok(())
1534	}
1535
1536	/// `runner` and `RUNNER_*` for a job on a self-hosted runner.
1537	pub async fn runner_context_for(&self, runner_id: &str, variables: &mut Map<String, Value>) -> Result<Value> {
1538	let mut context = g1t_actions::events::runner_context();
1539	let Some(runner) = self.runner_by_id(runner_id).await? else { return Ok(context) };
1540	let os = match runner.os.as_str() {
1541	"macos" => "macOS",
1542	"windows" => "Windows",
1543	_ => "Linux",
1544	};
1545	let arch = if runner.arch == "arm64" { "ARM64" } else { "X64" };
1546	context["name"] = json!(runner.name);
1547	context["os"] = json!(os);
1548	context["arch"] = json!(arch);
1549	context["environment"] = json!("self-hosted");
1550	{
1551	let vars = variables;
1552	vars.insert("RUNNER_NAME".into(), json!(runner.name));
1553	vars.insert("RUNNER_OS".into(), json!(os));
1554	vars.insert("RUNNER_ARCH".into(), json!(arch));
1555	vars.insert("RUNNER_ENVIRONMENT".into(), json!("self-hosted"));
1556	}
1557	Ok(context)
1558	}
1559
1560	// --- Agent work, for the runner service ----------------------------------
1561
1562	/// `runner_route`: the labels g1t's own work in `repo` runs on, when
1563	/// its workspace (or the repository) sends it to self-hosted runners.
1564	pub async fn runner_route(&self, a: RouteArgs) -> Result<Option<Vec<String>>> {
1565	let repo_id = match self.repo_by_path(&a.repo).await? {
1566	Some(repo) => Some(repo.id),
1567	None => None,
1568	};
1569	let settings = self.effective_runner_settings(&a.workspace, repo_id.as_deref()).await?;
1570	Ok(settings.agents_on_self_hosted.then_some(settings.agent_labels))
1571	}
1572
1573	async fn repo_by_path(&self, path: &RepoPath) -> Result<Option<Repo>> {
1574	let Some(actor) = self.workspace_actor(&path.namespace).await? else { return Ok(None) };
1575	self.visible_repo(path, &Some(actor)).await
1576	}
1577
1578	/// `enqueue_task`.
1579	pub async fn enqueue_task(&self, a: EnqueueTaskArgs) -> Result<Outcome<String>> {
1580	let Some(sealer) = &self.sealer else {
1581	return Ok(fail(FailureCode::Conflict, "Agent work cannot be handed to self-hosted runners yet: g1t's key for it is not set."));
1582	};
1583	let id = new_id("rtk", now_ms());
1584	let sealed = sealer.seal(&serde_json::to_string(&a.env)?, &id);
1585	let repo_id = self.repo_by_path(&a.repo).await?.map(\|r\| r.id);
1586	let labels = match model::runner_labels(&a.labels, model::SELF_HOSTED, model::SELF_HOSTED) {
1587	Ok(labels) => labels,
1588	Err(problem) => return Ok(fail(FailureCode::Invalid, problem)),
1589	};
1590	let at = now();
1591	let inserted = self
1592	.db
1593	.prepare(
1594	"INSERT OR IGNORE INTO runner_tasks (id, sandbox, workspace, repo_id, repo, kind, title, labels, env, timeout_minutes, status, created_at, run_id)
1595	VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, 'queued', ?, ?) RETURNING id",
1596	)
1597	.bind(&[
1598	id.as_str().into(),
1599	a.sandbox.as_str().into(),
1600	a.workspace.to_lowercase().into(),
1601	optional(repo_id.as_deref()),
1602	format!("{}/{}", a.repo.namespace, a.repo.name).into(),
1603	a.kind.as_str().into(),
1604	a.title.chars().take(300).collect::<String>().into(),
1605	serde_json::to_string(&labels)?.into(),
1606	sealed.into(),
1607	a.timeout_minutes.max(1).into(),
1608	at.into(),
1609	optional(a.run_id.as_deref().filter(\|id\| !id.is_empty())),
1610	])?
1611	.first::<Value>(None)
1612	.await?;
1613	if inserted.is_none() {
1614	return Ok(fail(FailureCode::Conflict, "That sandbox already handed work to a self-hosted runner."));
1615	}
1616	Ok(Outcome::Ok(id))
1617	}
1618
1619	/// `cancel_task`: the sandbox stopped it. Its runner hears on its next
1620	/// poll; the sandbox does its own ending.
1621	pub async fn cancel_task(&self, a: CancelTaskArgs) -> Result<Outcome<bool>> {
1622	let done = self
1623	.db
1624	.prepare(
1625	"UPDATE runner_tasks SET status = 'completed', exit_code = -1, reason = ?, env = NULL, finished_at = ?
1626	WHERE sandbox = ? AND status != 'completed' RETURNING id",
1627	)
1628	.bind(&[optional(a.reason.as_deref()), now().into(), a.sandbox.as_str().into()])?
1629	.first::<Value>(None)
1630	.await?;
1631	Ok(Outcome::Ok(done.is_some()))
1632	}
1633
1634	/// Ends a task and tells the sandbox waiting on it, once.
1635	async fn end_task(&self, task: &TaskRow, exit_code: i32, reason: &str) -> Result<()> {
1636	let ended = self
1637	.db
1638	.prepare(
1639	"UPDATE runner_tasks SET status = 'completed', exit_code = ?, reason = ?, env = NULL, finished_at = ?
1640	WHERE id = ? AND status != 'completed' RETURNING id",
1641	)
1642	.bind(&[exit_code.into(), optional(Some(reason).filter(\|r\| !r.is_empty())), now().into(), task.id.as_str().into()])?
1643	.first::<Value>(None)
1644	.await?;
1645	if ended.is_none() {
1646	return Ok(());
1647	}
1648	if let Some(runner_id) = &task.runner_id {
1649	self.db
1650	.prepare("UPDATE runners SET work_id = NULL, work_kind = NULL WHERE id = ? AND work_id = ?")
1651	.bind(&[runner_id.as_str().into(), task.id.as_str().into()])?
1652	.run()
1653	.await?;
1654	}
1655	let told: Result<Value> = g1t_kit::call(
1656	&self.runner,
1657	"task_ended",
1658	&json!({
1659	"sandbox": task.sandbox,
1660	"exitCode": exit_code,
1661	"reason": if reason.is_empty() { Value::Null } else { json!(reason) },
1662	"runner": task.runner_name,
1663	}),
1664	)
1665	.await;
1666	if let Err(error) = told {
1667	worker::console_error!("actions: the sandbox waiting on task {} was not told: {error}", task.id);
1668	}
1669	Ok(())
1670	}
1671
1672	/// `runner_activity`: what runs for the workspace now, in g1t's
1673	/// sandboxes and handed to its own runners. Its owners only, as its
1674	/// runners are.
1675	pub async fn runner_activity(&self, a: RunnerActivityArgs) -> Result<Outcome<RunnerActivity>> {
1676	let owner = RunnersOwner { repo: None, workspace: Some(a.workspace.clone()) };
1677	let place = check!(self.runner_place(&a.actor, &owner, false).await?);
1678	let workspace = place.workspace;
1679	let cloud_jobs = self
1680	.db
1681	.prepare(
1682	"SELECT jobs.id, jobs.name, jobs.run_id, runs.repo, jobs.started_at FROM jobs JOIN runs ON runs.id = jobs.run_id
1683	WHERE lower(jobs.namespace) = ? AND jobs.status = 'in_progress' AND jobs.labels IS NULL
1684	ORDER BY jobs.started_at DESC LIMIT ?",
1685	)
1686	.bind(&[workspace.as_str().into(), ACTIVITY_LIMIT.into()])?
1687	.all()
1688	.await?
1689	.results::<CloudJob>()?;
1690	let cloud_jobs_queued = self.queued_jobs(&workspace, false).await?;
1691	let self_hosted_jobs_queued = self.queued_jobs(&workspace, true).await?;
1692	let handed_over = self
1693	.db
1694	.prepare(
1695	"SELECT id, run_id, kind, title, repo, status, runner_name, created_at, started_at FROM runner_tasks
1696	WHERE status IN ('queued', 'in_progress') AND workspace = ?
1697	ORDER BY created_at DESC LIMIT ?",
1698	)
1699	.bind(&[workspace.as_str().into(), ACTIVITY_LIMIT.into()])?
1700	.all()
1701	.await?
1702	.results::<HandedOverTask>()?;
1703	Ok(Outcome::Ok(RunnerActivity { cloud_jobs, cloud_jobs_queued, handed_over, self_hosted_jobs_queued }))
1704	}
1705
1706	/// The workspace's workflow jobs queued for its own runners
1707	/// (`labelled`), or for g1t's sandboxes.
1708	async fn queued_jobs(&self, workspace: &str, labelled: bool) -> Result<u32> {
1709	let sql = if labelled {
1710	"SELECT COUNT(*) AS n FROM jobs WHERE lower(namespace) = ? AND status = 'queued' AND labels IS NOT NULL"
1711	} else {
1712	"SELECT COUNT(*) AS n FROM jobs WHERE lower(namespace) = ? AND status = 'queued' AND labels IS NULL"
1713	};
1714	Ok(self.db.prepare(sql).bind(&[workspace.into()])?.first::<Count>(None).await?.map_or(0, \|c\| c.n))
1715	}
1716
1717	/// `stuck_jobs`: jobs of the viewer's workspaces that have waited ten
1718	/// minutes or more for a self-hosted runner while none that could take
1719	/// them is online, for Mission control's Needs you.
1720	pub async fn stuck_jobs(&self, a: StuckJobsArgs) -> Result<Vec<StuckJob>> {
1721	let Some(viewer) = a.viewer else { return Ok(Vec::new()) };
1722	let at = now_ms();
1723	let mut out = Vec::new();
1724	for membership in viewer.workspaces.iter().take(20) {
1725	let workspace = membership.slug.to_lowercase();
1726	#[derive(Deserialize)]
1727	struct Waiting {
1728	id: String,
1729	name: String,
1730	run_id: String,
1731	labels: String,
1732	queued_at: Option<String>,
1733	repo: String,
1734	repo_id: String,
1735	}
1736	let waiting = self
1737	.db
1738	.prepare(
1739	"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
1740	WHERE jobs.status = 'queued' AND jobs.labels IS NOT NULL AND lower(jobs.namespace) = ? AND jobs.queued_at < ?
1741	ORDER BY jobs.rowid LIMIT 20",
1742	)
1743	.bind(&[workspace.as_str().into(), rfc3339(at.saturating_sub(10 * 60 * 1000)).into()])?
1744	.all()
1745	.await?
1746	.results::<Waiting>()?;
1747	if waiting.is_empty() {
1748	continue;
1749	}
1750	let groups = self.group_rows(&workspace).await?;
1751	let online: Vec<RunnerRow> = self.runner_rows(&workspace).await?.into_iter().filter(\|r\| r.online(at)).collect();
1752	// A repository's visibility is asked only when a host-mode runner
1753	// is online, since only those pass over a public repository's jobs.
1754	let any_host = online.iter().any(\|runner\| runner.isolation.as_deref() == Some("host"));
1755	let mut looked = Looked::default();
1756	for job in waiting {
1757	let wanted = Wanted::from_stored(&serde_json::from_str::<Vec<String>>(&job.labels).unwrap_or_default());
1758	let visibility = if any_host { self.visibility_once(&mut looked, &job.repo_id).await } else { None };
1759	let served = online.iter().any(\|runner\| {
1760	let group = runner.group_id.as_ref().and_then(\|id\| groups.iter().find(\|g\| &g.id == id));
1761	wanted.matches(&runner.labels(), group.map(\|g\| g.name.as_str()))
1762	&& model::isolation_allows(runner.isolation.as_deref(), visibility.as_deref())
1763	});
1764	if !served {
1765	out.push(StuckJob {
1766	id: job.id,
1767	name: job.name,
1768	run_id: job.run_id,
1769	repo: job.repo,
1770	labels: wanted.describe(),
1771	queued_at: job.queued_at.unwrap_or_default(),
1772	});
1773	}
1774	}
1775	}
1776	Ok(out)
1777	}
1778
1779	// --- Every minute ----------------------------------------------------------
1780
1781	/// Jobs and tasks that waited too long, tasks whose runner went quiet,
1782	/// runners offline for weeks, and registration tokens long expired.
1783	pub async fn sweep_runners(&self, at: u64) -> Result<()> {
1784	let before = \|ms: u64\| rfc3339(at.saturating_sub(ms));
1785	let waited = model::MAX_WAIT_HOURS * 60 * 60 * 1000;
1786	let stale = self
1787	.db
1788	.prepare("SELECT * FROM jobs WHERE status = 'queued' AND labels IS NOT NULL AND queued_at < ? LIMIT 50")
1789	.bind(&[before(waited).into()])?
1790	.all()
1791	.await?
1792	.results::<JobRow>()?;
1793	for job in stale {
1794	let stored: Vec<String> = job.labels.as_deref().and_then(\|l\| serde_json::from_str(l).ok()).unwrap_or_default();
1795	let why = format!(
1796	"No self-hosted runner with labels {} took it within {} hours.",
1797	Wanted::from_stored(&stored).describe(),
1798	model::MAX_WAIT_HOURS
1799	);
1800	self.finish_job(&job.id, "failure", Some(&why), None).await?;
1801	}
1802	let tasks = self
1803	.db
1804	.prepare(
1805	"SELECT * FROM runner_tasks WHERE (status = 'queued' AND created_at < ?) OR (status = 'in_progress' AND seen_at < ?) LIMIT 50",
1806	)
1807	.bind(&[before(waited).into(), before(SILENT_MS).into()])?
1808	.all()
1809	.await?
1810	.results::<TaskRow>()?;
1811	for task in tasks {
1812	let why = if task.status == "queued" {
1813	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)
1814	} else {
1815	format!("The self-hosted runner {} stopped answering.", task.runner_name.clone().unwrap_or_default())
1816	};
1817	self.end_task(&task, 1, &why).await?;
1818	}
1819	let forgotten = self
1820	.db
1821	.prepare("SELECT * FROM runners WHERE COALESCE(last_seen_at, created_at) < ? LIMIT 50")
1822	.bind(&[before(FORGET_OFFLINE_MS).into()])?
1823	.all()
1824	.await?
1825	.results::<RunnerRow>()?;
1826	for row in forgotten {
1827	self.forget_runner(&row, "The runner was offline for 14 days.").await?;
1828	}
1829	// An ephemeral runner that took its job and then went quiet.
1830	let spent = self
1831	.db
1832	.prepare("SELECT * FROM runners WHERE ephemeral = 1 AND spent = 1 AND work_id IS NULL AND COALESCE(last_seen_at, created_at) < ? LIMIT 50")
1833	.bind(&[before(60 * 60 * 1000).into()])?
1834	.all()
1835	.await?
1836	.results::<RunnerRow>()?;
1837	for row in spent {
1838	self.forget_runner(&row, "The ephemeral runner finished its job.").await?;
1839	}
1840	self.db
1841	.prepare("DELETE FROM runner_registrations WHERE expires_at < ?")
1842	.bind(&[before(24 * 60 * 60 * 1000).into()])?
1843	.run()
1844	.await?;
1845	Ok(())
1846	}
1847	}
1848
1849	#[derive(Deserialize)]
1850	struct IdRow {
1851	id: String,
1852	}
1853
1854	/// Whether `runner` is online and could take a public repository's job:
1855	/// not host-mode, and its repository, group and labels let it.
1856	fn takes_public_job(runner: &RunnerRow, groups: &[GroupRow], repo_id: &str, run_repo: &str, wanted: &Wanted, at: u64) -> bool {
1857	let group = runner.group_id.as_ref().and_then(\|id\| groups.iter().find(\|g\| &g.id == id));
1858	runner.online(at)
1859	&& model::isolation_allows(runner.isolation.as_deref(), Some("public"))
1860	&& runner.repo_id.as_deref().is_none_or(\|repo\| repo == repo_id)
1861	&& (runner.repo_id.is_some() \|\| group.is_none_or(\|g\| g.allows(run_repo)))
1862	&& wanted.matches(&runner.labels(), group.map(\|g\| g.name.as_str()))
1863	}
1864
1865	fn group_view(group: &GroupRow, runners: &[RunnerRow]) -> RunnerGroup {
1866	RunnerGroup {
1867	id: group.id.clone(),
1868	name: group.name.clone(),
1869	default: group.is_default != 0,
1870	repositories: group.repositories(),
1871	runners: runners.iter().filter(\|r\| r.group_id.as_deref() == Some(group.id.as_str())).count() as u32,
1872	updated_at: group.updated_at.clone(),
1873	}
1874	}
1875
1876	#[cfg(test)]
1877	mod tests {
1878	use super::*;
1879
1880	fn group(repositories: &[&str]) -> GroupRow {
1881	GroupRow {
1882	id: "rng_1".into(),
1883	name: "Default".into(),
1884	is_default: 1,
1885	repositories: serde_json::to_string(&repositories).unwrap(),
1886	updated_at: "2026-10-06T00:00:00Z".into(),
1887	}
1888	}
1889
1890	#[test]
1891	fn being_seen_is_written_once_a_minute() {
1892	let at = 2_000_000_000_000;
1893	assert!(seen_is_stale(None, at));
1894	assert!(!seen_is_stale(Some(&rfc3339(at - 59_000)), at));
1895	assert!(seen_is_stale(Some(&rfc3339(at - 61_000)), at));
1896	// A runner written down at the last moment is still online until the
1897	// next write, a poll later.
1898	assert!(SEEN_RESOLUTION_MS + model::MAX_POLL_WAIT_MS < model::ONLINE_WITHIN_MS);
1899	assert!(SEEN_RESOLUTION_MS < crate::SILENT_MS);
1900	}
1901
1902	#[test]
1903	fn a_group_lets_its_repositories_use_it() {
1904	assert!(group(&[]).allows("acme/web"));
1905	assert!(group(&["web", "api"]).allows("acme/WEB"));
1906	assert!(group(&["web"]).allows("web"));
1907	assert!(!group(&["web"]).allows("acme/docs"));
1908	}
1909
1910	#[test]
1911	fn names_are_checked() {
1912	assert_eq!(valid_name(" build-01 ").unwrap(), "build-01");
1913	assert!(valid_name("").is_err());
1914	assert!(valid_name("has space").is_err());
1915	assert!(valid_name(&"x".repeat(65)).is_err());
1916	assert!(valid_group_name("GPU machines").is_ok());
1917	assert!(valid_group_name("").is_err());
1918	}
1919
1920	#[test]
1921	fn only_a_sandboxed_runner_that_could_take_it_serves_a_public_job() {
1922	let at = 1_000_000_000;
1923	let runner = \|isolation: Option<&str>\| RunnerRow {
1924	id: "rnr_1".into(),
1925	workspace: "acme".into(),
1926	repo_id: None,
1927	repo: None,
1928	group_id: None,
1929	name: "a".into(),
1930	labels: r#"["self-hosted","linux","x64"]"#.into(),
1931	os: "linux".into(),
1932	arch: "x64".into(),
1933	version: String::new(),
1934	ephemeral: 0,
1935	credential_hash: String::new(),
1936	previous_hash: None,
1937	rotated_at: String::new(),
1938	work_id: None,
1939	work_kind: None,
1940	spent: 0,
1941	last_seen_at: Some(rfc3339(at)),
1942	created_at: String::new(),
1943	created_by: None,
1944	isolation: isolation.map(str::to_owned),
1945	};
1946	let wanted = Wanted::of(&["self-hosted".to_owned(), "linux".to_owned()], None);
1947	let takes = \|row: &RunnerRow\| takes_public_job(row, &[], "rep_1", "acme/web", &wanted, at);
1948	assert!(!takes(&runner(Some("host"))));
1949	assert!(takes(&runner(Some("container"))));
1950	assert!(takes(&runner(Some("nested-container"))));
1951	assert!(takes(&runner(None)), "an older runner, which refuses for itself");
1952	assert!(!takes(&RunnerRow { repo_id: Some("rep_2".into()), ..runner(None) }), "another repository's");
1953	assert!(!takes(&RunnerRow { labels: r#"["self-hosted","macos"]"#.into(), ..runner(None) }));
1954	assert!(!takes(&RunnerRow { last_seen_at: None, ..runner(None) }), "offline");
1955	}
1956
1957	#[test]
1958	fn online_means_seen_in_the_last_ninety_seconds() {
1959	let row = RunnerRow {
1960	id: "rnr_1".into(),
1961	workspace: "acme".into(),
1962	repo_id: None,
1963	repo: None,
1964	group_id: None,
1965	name: "a".into(),
1966	labels: "[]".into(),
1967	os: "linux".into(),
1968	arch: "x64".into(),
1969	version: String::new(),
1970	ephemeral: 0,
1971	credential_hash: String::new(),
1972	previous_hash: None,
1973	rotated_at: String::new(),
1974	work_id: None,
1975	work_kind: None,
1976	spent: 0,
1977	last_seen_at: Some(rfc3339(1_000_000_000)),
1978	created_at: String::new(),
1979	created_by: None,
1980	isolation: None,
1981	};
1982	assert!(row.online(1_000_000_000 + 60_000));
1983	assert!(!row.online(1_000_000_000 + 120_000));
1984	assert_eq!(row.workspace, "acme");
1985	}
1986	}