wip(#3001): remove submit layer, rescue power ops into job_queue/power.rs

TREE IS RED ON PURPOSE — there is no compiling intermediate between
deleting submit and converting every caller. Checkpoint commit so the
work is durable; do not "fix" it by restoring submit.

Done:
- JobQueue::submit -> JobQueue::insert (no source/reason/container;
  returns the ids insert_job names).
- submit.rs deleted. Its 6 pure chain builders + 3 async *_many
  gatherers were NOT wrapper code and are rescued into
  job_queue/power.rs (templates.rs documents power ops as living
  outside it, because their shape needs a live is_running read).
- Converted: meta_inputs 1, topology 2, permissions 3, auto_update 2,
  actions 3, lifecycle_handlers 3.
- Dropped source/reason at every converted site: nothing ever read
  NodeKind::Dag's fields (only `{ .. }` matches exist), so they are
  write-only. Dead reason-only locals deleted; the boot sweep's summary
  became a tracing::info! rather than being lost.

Remaining: dashboard/lifecycle_ops 7, server.rs 7, and the test suite —
tests.rs has its own submit() helper whose u64 return is used as the
handle to navigate the inserted DAG, so those need a different way to
find nodes, not a mechanical port.
This commit is contained in:
atlas 2026-08-04 11:17:37 +02:00 committed by mara
commit 7c0d9d2379
9 changed files with 198 additions and 301 deletions

View file

@ -0,0 +1,277 @@
//! Power ops (`stop` / `start` / `restart`) — the DAG shapes whose per-agent
//! form depends on **live** container state, so they cannot be static
//! [`super::templates`] entries.
//!
//! The split is the purity line, not the subject matter: the `*_chain` /
//! `*_nodes` builders below are pure (they take `running` / `stale` as
//! parameters, which is what keeps them unit-testable without a container),
//! and only the `*_many` entry points do the async `lifecycle::is_running`
//! read that produces those parameters.
//!
//! Each entry point declares its group and inserts it. There is no metadata
//! parameter and no container node: attribution is not something every caller
//! has to invent, and a DAG is addressed by the nodes a template names.
use std::sync::Arc;
use hive_jobq::NodeId;
use super::JobBuilder;
use super::templates;
use crate::coordinator::Coordinator;
use crate::lifecycle;
// ---- dynamic power-op DAG assembly ----------------------------------------
//
// The pure per-agent chain builders below take `running` (and `stale`)
// explicitly so they stay pure + unit-testable without a live container;
// the async `*_many` fns read the real state via `lifecycle::is_running`
// then hand it in. Each chain declares into the shared job it is handed, and
// names the nodes it depends on — so there is nothing to rebase.
/// One agent's **stop** subgraph. `SetWanted(Off)` head + `Reconcile` tail
/// always; the graceful `Signal → Drain` quiesce only when the agent is
/// actually running (nothing to drain on a down container). The `Reconcile`
/// stays even for a down agent so a race-up between the state read and exec
/// is still stopped in-DAG.
fn stop_chain(builder: &JobBuilder, agent: &str, graceful: bool, running: bool) {
// `SetWanted` is the group root and owns the agent lease; the mechanical
// steps are its children (borrow the lease, run once it reaches `Finishing`,
// dep-ordered among themselves).
let a = || agent.to_owned();
let wanted = builder
.node(NodeKind::SetWanted {
agent: a(),
up: false,
})
.needs(Resource::Agent(a()));
// Declaration order is dependency order: the quiesce steps come first so
// the `Reconcile` that waits on them can name them.
if graceful && running {
// `SetWanted` is the brace here, so the quiesce pair borrows its grant
// rather than declaring the lease itself — same shape the rebuild
// template uses, one definition.
let drain = super::templates::quiesce(builder, agent, wanted);
let _ = builder
.node(NodeKind::Reconcile { agent: a() })
.needs(Resource::Agent(a()))
.part_of(wanted)
.after_ok(drain);
} else {
let _ = builder
.node(NodeKind::Reconcile { agent: a() })
.needs(Resource::Agent(a()))
.part_of(wanted);
}
}
/// One agent's **start** subgraph. `SetWanted(Up)` head; a down + stale-rev
/// agent gets the rebuild subgraph (its tail `Reconcile` starts it on
/// current derivations), otherwise a plain `Reconcile` (which starts a down
/// agent and noops an already-running one).
fn start_chain(builder: &JobBuilder, agent: &str, running: bool, stale: bool) {
let wanted = builder
.node(NodeKind::SetWanted {
agent: agent.to_owned(),
up: true,
})
.needs(Resource::Agent(agent.to_owned()));
if !running && stale {
// Rebuild subtree chained behind the `SetWanted` head. `MetaSync`,
// `Prebuild` + `Reconcile` are their own group roots (top-level, per
// `rebuild_nodes`).
rebuild_nodes(builder, agent, true, Some(wanted));
} else {
let _ = builder
.node(NodeKind::Reconcile {
agent: agent.to_owned(),
})
.needs(Resource::Agent(agent.to_owned()))
.part_of(wanted);
}
}
/// One agent's **restart** subgraph. Restart NEVER rewrites `wanted`
/// intent (no `SetWanted` head, unlike stop/start): it bounces the
/// container and lets the tail `Reconcile` converge to the agent's
/// EXISTING intent, so a deliberately-stopped (`wanted = Off`) agent is
/// not forced back up by a hive-wide restart. A running agent gets the
/// mechanical stop (`Signal → Drain` when graceful, then `StopForUpdate`)
/// before `Reconcile`; a down agent gets just `Reconcile`, which
/// converges to intent — a stopped (`wanted = Off`) agent stays stopped,
/// a crashed (`wanted = Up`) agent comes back up.
fn restart_chain(builder: &JobBuilder, agent: &str, graceful: bool, running: bool) {
let a = || agent.to_owned();
if !running {
// Nothing to bounce — a lone Reconcile converges to intent.
let _ = builder
.node(NodeKind::Reconcile { agent: a() })
.needs(Resource::Agent(a()));
return;
}
// Running: mechanical stop then Reconcile. The first stop node is the group
// ROOT (no SetWanted head) and owns the agent lease; the rest are its
// children (borrow the lease, dep-ordered), so the bounce holds one
// continuous lease and `Reconcile` cancel-cascades if a stop step fails.
//
// `Reconcile` gates on the last mechanical step. For a non-graceful bounce
// that step *is* the root, and the parent gate already orders it — a child
// must NOT dep on its own parent (dep-scope), so it takes no sibling edge.
//
// ⚠️ This is the one quiesce site that does NOT use `templates::quiesce`.
// The helper needs a brace holding the lease above the pair; here `Signal`
// *is* the holder, and the nesting under it is what keeps the grant
// continuous across the bounce. Giving this chain its own brace would
// unify all three sites — at the cost of one extra no-op node on every
// graceful restart, which an operator would see. Deliberately not done as
// a side effect of a rebuild-shape change.
if graceful {
let signal = builder
.node(NodeKind::Signal { agent: a() })
.needs(Resource::Agent(a()));
let drain = builder
.node(NodeKind::Drain { agent: a() })
.needs(Resource::Agent(a()))
.part_of(signal);
let stop = builder
.node(NodeKind::StopForUpdate { agent: a() })
.needs(Resource::Agent(a()))
.part_of(signal)
.after_ok(drain);
let _ = builder
.node(NodeKind::Reconcile { agent: a() })
.needs(Resource::Agent(a()))
.part_of(signal)
.after_ok(stop);
} else {
let stop = builder
.node(NodeKind::StopForUpdate { agent: a() })
.needs(Resource::Agent(a()));
let _ = builder
.node(NodeKind::Reconcile { agent: a() })
.needs(Resource::Agent(a()))
.part_of(stop);
}
}
// The `*_nodes` declarers below are the PURE core the async `*_many` fns call
// after reading live state — they take the per-agent running (and stale)
// flags explicitly, so unit tests exercise the online/offline shapes without
// a live container. `*_many` = gather state + declare + submit.
//
// A power op has no tail node: its effect is its nodes (`SetWanted` +
// `Reconcile`), with nothing left to do once they settle.
//
// There is no concatenation step either: every chain declares into the same
// builder and each keeps its own root, so the per-agent subgraphs are
// independent and run concurrently, each on its own lease. Rebasing one
// subgraph's indices onto another's used to be a function.
/// Declare the stop DAG from explicit `(agent, running)` targets.
pub(crate) fn stop_nodes(builder: &JobBuilder, targets: &[(String, bool)], graceful: bool) {
for (agent, running) in targets {
stop_chain(builder, agent, graceful, *running);
}
}
/// Assemble the start DAG from explicit `(agent, running, stale)` targets.
///
/// No DAG-level pill: each agent's dashboard label is derived from the node
/// running under its lease, so a down+stale agent that grew a rebuild subgraph
/// reports `rebuilding` during its swap and `starting` at its reconcile,
/// without the DAG having to guess one label covering every target.
pub(crate) fn start_nodes(builder: &JobBuilder, targets: &[(String, bool, bool)]) {
for (agent, running, stale) in targets {
start_chain(builder, agent, *running, *stale);
}
}
/// Declare the restart DAG from explicit `(agent, running)` targets.
pub(crate) fn restart_nodes(builder: &JobBuilder, targets: &[(String, bool)], graceful: bool) {
for (agent, running) in targets {
restart_chain(builder, agent, graceful, *running);
}
}
// ---- entry points ---------------------------------------------------------
//
// Each reads the live state its shape depends on, then declares + inserts.
/// Restart `agents` in a **single** DAG — one per-agent subgraph each, built
/// from live running state and run concurrently on their own leases. A running
/// agent gets the stop→reconcile chain (`graceful` prepends signal→drain); a
/// down agent gets a lone `Reconcile`. Restart never writes `wanted`, so the
/// tail `Reconcile` converges each agent to its EXISTING intent — a
/// deliberately-stopped agent stays down.
///
/// # Errors
/// Propagates a graph-insert error.
pub async fn restart_many(
coord: &Arc<Coordinator>,
agents: &[String],
graceful: bool,
) -> anyhow::Result<Vec<NodeId>> {
let mut targets = Vec::with_capacity(agents.len());
for agent in agents {
targets.push((agent.clone(), lifecycle::is_running(agent).await));
}
let ids = coord.job_queue.insert(|b| {
restart_nodes(b, &targets, graceful);
Vec::new()
})?;
coord.emit_rebuild_queue_snapshot();
Ok(ids)
}
/// Start `agents` in a **single** DAG. A down agent gets
/// `SetWanted(Up) → Reconcile` (or, rev stale, a rebuild-then-start so it comes
/// up on current derivations); an already-running agent gets the same shape
/// with the reconcile noop'ing.
///
/// # Errors
/// Propagates a graph-insert error.
pub async fn start_many(coord: &Arc<Coordinator>, agents: &[String]) -> anyhow::Result<Vec<NodeId>> {
let current = crate::auto_update::current_flake_rev(&coord.hyperhive_flake);
let mut targets = Vec::with_capacity(agents.len());
for agent in agents {
let running = lifecycle::is_running(agent).await;
let stored = std::fs::read_to_string(crate::paths::applied_rev_marker(agent)).ok();
let stale = current
.as_ref()
.is_some_and(|rev| stored.as_deref() != Some(rev.as_str()));
if !running && stale {
tracing::info!(%agent, "start: rev stale + agent down — rebuild-then-start");
}
targets.push((agent.clone(), running, stale));
}
let ids = coord.job_queue.insert(|b| {
start_nodes(b, &targets);
Vec::new()
})?;
coord.emit_rebuild_queue_snapshot();
Ok(ids)
}
/// Stop `agents` in a **single** DAG. A running agent gets
/// `SetWanted(Off) → [Signal → Drain →](graceful) Reconcile`; a down agent
/// skips the pointless quiesce but keeps the `Reconcile` as the race-up
/// backstop.
///
/// # Errors
/// Propagates a graph-insert error.
pub async fn stop_many(
coord: &Arc<Coordinator>,
agents: &[String],
graceful: bool,
) -> anyhow::Result<Vec<NodeId>> {
let mut targets = Vec::with_capacity(agents.len());
for agent in agents {
targets.push((agent.clone(), lifecycle::is_running(agent).await));
}
let ids = coord.job_queue.insert(|b| {
stop_nodes(b, &targets, graceful);
Vec::new()
})?;
coord.emit_rebuild_queue_snapshot();
Ok(ids)
}