The hive applies one `agentCpuQuota` / `agentMemoryMax` to every
container. That's the right default and the wrong ceiling: a build-heavy
agent needs headroom the other twelve don't, and raising the hive-wide
value to suit it hands that headroom to everyone.
Adds a per-agent override, persisted host-side and resolved per-field
against the hive defaults.
Follows the existing `meta/*.json` pattern (`capabilities.json`,
`tool-groups.json`): a host-side map read by `hive-c0re`, staged and
committed in the meta repo so every change lands in the audit trail.
```json
{ "sock": { "cpu_quota": "400%", "memory_max": "8G" } }
```
Fallback is **per field**, not per agent: an entry with only
`memory_max` leaves that agent on the hive-wide CPU quota. Absent file,
absent agent and absent field all resolve to the hive default, so the
feature is inert until someone opts an agent in.
Unlike the other meta files this one is **not** injected into the
container — a limit is something done *to* an agent, not something it
reads about itself.
```
hivectl agents set-limits sock --cpu-quota 400% --memory-max 8G
hivectl agents set-limits sock --reset
```
Values are validated before they're persisted: they go into a systemd
drop-in verbatim, and a typo there makes the unit fail to *start* —
turning a fat-fingered quota into a container that won't come back.
The command is declarative: each call replaces the agent's whole entry.
That makes a forgotten flag a silent revert, so a bare `set-limits
<name>` is rejected at the clap layer and clearing needs an explicit
`--reset`.
`ContainerView` gains `cpu_quota` / `memory_max`, both always populated:
there's no "unset" state to render, only "same as everyone else". They
reflect what the drop-in *says* — what the next start will enforce — not
a live cgroup reading.
The write goes through `meta::commit_resource_limits` rather than the
bare setter, so it's staged and committed under `META_LOCK`. Writing
without committing would leave the meta working tree dirty for the next
`prepare_deploy` to trip over.
Docs: `persistence.md` (the new meta file, and why it isn't injected),
`tools/hivectl.md` (the prose guide), `tools/hivectl-cli.md`
(regenerated clap dump).
Closes: internal/requests issue 25
A paused agent keeps its container, its claude session and its
dashboard/todo servers up, but stops driving turns. Messages queue
unacked and are drained on resume.
The whole protocol is a single marker file, `<harness>/paused`. That
directory is already a bind-mount shared between host and container, so
both sides just stat the same path: the harness reads it to decide
whether to drive a turn, hive-c0re reads it to render the badge and
writes/removes it for `hivectl pause|resume`. No new wire protocol, no
container round-trip, and it is sticky across restarts by construction.
Not calling `recv_next` while paused *is* the queueing semantic, so
there is no fencing to get wrong: reminders buffer in their unbounded
channel, the todo `Notify` permit coalesces, and a `request_next_turn`
that raced the pause survives because the gate sits above
`self_continue.take()`.
Graceful stop is handled host-side rather than in the harness: a paused
agent provably has no turn in flight, so `run_signal` skips the fence
entirely instead of eating the full `GRACEFUL_STOP_TIMEOUT` waiting for
a checkpoint turn that will never run.
`paused` is reported on `ContainerView` / `AgentStatusRow` for the
dashboard, orthogonal to `running` and reported for stopped containers
too.
Closes: hyperhive/hyperhive issue 2271
- server::await_dags: a DAG is settled when gone from the snapshot (fully
Done) or present with all nodes terminal; pending only with a non-terminal
node (DagView no longer carries a rolled-up state).
- DagView::rollup_state() added to hive-sh4re — the shared node-set roll-up
derivation every Rust consumer uses.
- JobQueue::build_log_id_of(node_id) — the node_id -> build_logs lookup the
query endpoint will use; tests assert log-id via it now.
- tests: derive roll-up state; drop the off-wire step/build_log_id wire asserts.
With the meta-update cascade (#2476) and startup sweep (#2450) folded
into single DAGs that grow per-agent subgraphs via append_subgraph,
nothing links parent/child DAGs anymore — parent_id is dead.
hive-c0re: drop parent_id from Dag/DagSpec (+ the DagView copy); delete
append_children and cancel_children (no callers); simplify trim_history
(no more terminal-parent-with-live-children guard — a one-big-DAG is
terminal only when its whole graph settles); drop the rebuild() parent_id
param; QueueDag returns just the polled DAG (no fan-out children to
gather). hive-sh4re: drop the DagView.parent_id wire field.
frontend: a multi-step op is one DAG now, so renderRebuildQueue drops the
childrenOf/orphans cross-DAG grouping and renders each entry flat; its
per-agent subgraphs render as nodes within the one row (split by deps).
Removed the dead rqe-child style + isChild plumbing.
Docs + the child-DAG queue tests updated/removed to match.
Hive-wide `stop` / `start` / `restart` emit ONE DAG with a per-agent
subgraph each (concurrent on their own leases) instead of N DAGs — and each
subgraph is now built dynamically from the agent's live running state rather
than a fixed template shape:
- online agent: the full stop→reconcile (restart: stop-for-update→reconcile)
chain; `graceful` prepends signal→drain.
- offline agent: just `SetWanted → Reconcile` (nothing to quiesce/stop; a
restart of a down agent is really a start).
The head `SetWanted` (intent) and tail `Reconcile` (convergence guarantee)
are always present; only the mechanical `Signal`/`Drain`/`StopForUpdate`
nodes are state-conditional. Keeping `Reconcile` in every shape closes the
TOCTOU window — a race-up between the `is_running` read and node exec is
still converged in-DAG (with `StopForUpdate`-noop as the backstop) — with no
reliance on an external reconcile sweep.
The state-aware assembly needs an async `is_running` read, so it moves out
of the pure/sync `templates.rs` into `submit.rs`, layered as pure
`*_chain(running)` → pure `*_spec(targets)` (the unit-test seam) → async
`*_many` (reads live state + submits). `templates.rs` keeps only the shared
pure primitives (`node`/`after_ok`/`rebuild_nodes`).
Callers await the now-async submit fns (server, dashboard, socket_server).
Tests exercise both the online and offline shapes via the pure `*_spec`
seam. docs/coordinator.md shapes updated.
A hive-wide restart was N separate single-agent DAGs (one submit::restart
per agent). Now that agent is per-node (#2445), make it ONE DAG with a
per-agent restart subgraph each.
- templates::restart takes an agent list: each agent gets an independent
subgraph (a head SetWanted(Up) root, then its restart chain), so the N
subgraphs run concurrently on their own leases. One agent = the ordinary
single-agent restart; unifies the old restart + graceful_restart fns.
- submit::restart / graceful_restart stay as single-agent wrappers over
the new submit::restart_many(agents, graceful).
- server.rs handle_restart_all + handle_restart_scoped submit one
restart_many call instead of looping per agent. Infra containers
unchanged (no lease/DAG, synchronous).
Scope: restart + restart-all only. Broad stop+start is the same pattern
(stop/start templates take agent lists) — a follow-up increment.
argus caught it: the function-level /// comment still described the
old submit-await-submit graceful approach after the code moved to
one atomic GracefulRestart DAG.
mara's review on #2436: no submit-await-submit composition, even
server-side. Adds Template::GracefulRestart (Signal -> Drain ->
StopForUpdate -> Reconcile, wanted=Up) mirroring how Restart already
does StopForUpdate -> Reconcile, plus submit::graceful_restart and
templates::graceful_restart. handle_restart_scoped now submits exactly
one DAG per agent up front for both the graceful and non-graceful
case -- no await_dags in the loop anymore.
hivectl restart --agent NAME previously composed stop() then start() as
two separate client-side daemon calls glued by CLI-process control flow
— not one DAG, and a dropped hivectl connection mid-restart (ssh drop,
Ctrl-C) left the agent stopped with no automatic follow-up. mara flagged
this as the first target for the 'dagify hivectl commands' issue.
New HostRequest::RestartScoped{scope, graceful} handles it server-side:
each targeted agent now rides exactly one atomic Restart-template DAG
(same one hivectl agents restart / restart-all already use) in the
common non-graceful case. --graceful has no single-DAG template yet, so
it submits the graceful-stop DAGs, awaits them server-side, then submits
the start DAGs — still one daemon call end to end, just not yet a single
DAG (noted as a follow-up). Infra containers restart synchronously as
before (no lease/DAG concept for them).
CLI-side restart() now just makes the one call + waits, same output
shape as before via render_lifecycle.
Split the priv-socket wire types (PrivRequest/PrivResponse/PrivEvent and
friends) out of hive-sh4re into their own hive-priv-sock crate, mirroring
the existing hive-host-sock split. hive-priv — the root-privileged
helper — now depends on just this narrow protocol crate instead of the
much larger daemon-shared crate, shrinking its dependency surface and
making the privsep boundary easier to audit. No server/client
implementation lives here, only the wire contract; hive-c0re still
depends on hive-sh4re directly for everything else.
dispatch was 101 lines (1 over limit) due to the AgentStatus arm.
Extract it to a dedicated handle_agent_status helper to bring dispatch
under the 100-line lint limit without the allow attribute.
Per mara review comment on PR #2379.
- Backtick-quote `pull_request` in doc comments (4x doc_markdown)
- Add #[allow(clippy::too_many_lines)] to server::dispatch (101/100;
+1 line from submit_kind fetched_sha param in 5dd0a36f)
Two hardening items from argus's review of #2374:
1. PR state check at submission:
- Add `pr_is_open(repo, pr)` to forge/pr_merge.rs using
`repo_get_pull_request` + `StateType` — early error if the PR is
already closed or merged instead of queuing a card that fails later
- Call it in `submit_merge_config_pr` before fetching the head sha
2. Atomic fetched_sha INSERT:
- Add `fetched_sha: Option<&str>` to `Approvals::submit_kind` so
the sha can be included in the INSERT rather than a follow-up UPDATE
- MergeConfigPr already knows the sha before inserting the row
(pr_head_sha runs first) → pass `Some(&sha)`, drop the separate
`set_fetched_sha` call → truly atomic
- ApplyCommit still needs two writes (sha resolved by git_fetch_to_tag
after the row exists) → pass `None`, `set_fetched_sha` unchanged
- All other callers (InitConfig, Spawn, UpdateMetaInputs,
SchedulePrompt) pass `None` — no behavioural change
- Add `fetched_sha_in_insert_is_readable_via_get` test covering the
MergeConfigPr path
Collapse the scattered ensure_agent_runtime_dir calls into the lifecycle
functions themselves so callers have a single responsibility:
- lifecycle::spawn: calls ensure_agent_runtime_dir before write_dropins.
Callers (handle_spawn, ensure_root_agent) no longer need a separate
preamble step.
- lifecycle::rebuild_no_meta spawn path: calls ensure_agent_runtime_dir
before write_dropins. apply_commit / merge_config_pr flows no longer
need a manual ensure_agent_runtime_dir.
- run_create (job-queue): drops ensure_agent_runtime_dir + register_agent.
The tail Reconcile's converge_start_preamble handles the runtime dir
and mcp_sockets::spawn_poll handles the listener. Create stays purely
'provision + create', not 'create + start'.
- handle_spawn (server.rs): drops manual preamble; lifecycle::spawn owns it.
Drops unneeded unregister_agent on failure (supervisor handles listener).
- ensure_root_agent (auto_update.rs): drops manual ensure_agent_runtime_dir.
- actions.rs apply_commit / merge_config_pr: drop manual
ensure_agent_runtime_dir; rebuild_no_meta's spawn path handles it.
Result: ensure_agent_runtime_dir lives in exactly two places —
lifecycle::spawn (direct spawn) and converge_start_preamble (start/reconcile
path). All other callers are clean call sites.
- lifecycle::ensure_agent_runtime_dir(name): pure filesystem op, no
Coordinator dep. Creates /run/hyperhive/agents/<name> without touching
the MCP listener map.
- workers/mcp_sockets::spawn_poll(coord): 10 s reconcile loop (same shape
as agent_sockets::spawn_poll). Converges 'agent running => MCP listener
bound'. First tick is immediate so hive-c0re restarts re-register all
running agents without waiting a full interval. Fixes the dead-listener-
after-daemon-restart gap.
- All ensure_runtime() call sites updated:
- Prebuild/Swap/WriteDropin: Coordinator::agent_dir() (pure, no IO)
- Reconcile-Start: ensure_agent_runtime_dir + agent_dir (dir may be
missing after reboot; listener deferred to supervisor)
- run_create / handle_spawn: ensure_agent_runtime_dir + register_agent
(eager on first spawn so socket ready before harness first turn)
- apply_commit / merge_config_pr: ensure_agent_runtime_dir + agent_dir
- Manager (auto_update): ensure_agent_runtime_dir + agent_dir
(manager has no MCP listener; socket_server::start_manager owns it)
- ensure_runtime() retained in Coordinator with updated doc pointing at
the preferred split form. No callers remain outside tests.
Root cause of the boot outage: container@h-* units try to start before
hive-c0re reaches ensure_runtime, so bind-mount source dirs are missing.
Fix: hive-c0re (via hive-priv, which runs as root) writes
/etc/tmpfiles.d/hyperhive-agents.conf whenever the agent set changes.
systemd-tmpfiles-setup.service (sysinit.target) reads it at every boot
BEFORE any container units start, pre-creating:
/run/hyperhive/agents/<name> — MCP socket dir (bind -> /run/hive)
/run/hive-agent/<name> — web socket dir (bind -> /run/hive-agent)
This alone removes the outage class: even if hive-c0re is slow to start,
the bind-mount sources exist and container units can activate.
Added:
- PrivRequest::SyncAgentTmpfiles { agents } in hive-sh4re
- sync_agent_tmpfiles() in hive-priv: generates content, writes atomically,
calls systemd-tmpfiles --create to apply immediately
- priv_client::sync_agent_tmpfiles() wrapper
- lifecycle::sync_tmpfiles() best-effort helper (list + priv call)
- Call sites: hive-c0re startup, handle_spawn success, destroy success
- subvol upgrade waits for the queued stop DAG before migrating (was
snapshotting + swapping state under a live bind mount) and for the
restart job after
- history trim gets a 5-min grace for fresh terminals so broad
stop/start waits can't miss a failed DAG evicted by the per-template
cap (cap still applies past the grace)
- restart-all returns its DAG ids so hivectl actually waits
- hard stops await their agent DAGs (bounded) before infra goes down,
restoring the agents-before-infra invariant
- hivectl wait uses node-level terminality so the after-any recovery
reconcile is watched to completion; infra render errors no longer
skip watching already-queued agent DAGs
- fold hive-bash-mcp's last local now_unix into wire_time
every agent lifecycle verb on the admin socket (rebuild / restart /
restart-all / kill / stop / start) now submits job-queue DAGs and
returns their ids; hivectl polls the new HostRequest::QueueDag and
prints a live node-chain progress line per DAG (fan-out children
included), exiting non-zero on failure — --no-wait opts out. DagView
and the queue wire enums move to hive_sh4re::jobs (wire types live in
the shared crate); the last fused rebuild path (lifecycle::rebuild)
is gone. tracker: #2166
- deploy-window gate (meta::exclusive) + path-limited meta commits:
a perm/lock/topology commit can no longer sweep an ApprovalDeploy's
staged flake.lock and neuter abort_deploy (regression test included)
- cancel surfaces now buffer terminal roll-ups the scheduler drains,
so a queued approval DAG cancelled by the operator resolves its
approval instead of dangling, and cancelled power ops revert their
wanted flip to the observed state
- hivectl restart / restart-all ride the queue (lease serialization,
transient guard) and restart sets wanted=Up like the old kill+start
- exactly one Rebuilt event per rebuild DAG, emitted at terminal
- StopForUpdate pre-seeds a missing agent_power row from the pre-stop
observation so a rebuild can't strand an unknown agent offline
- history trim keeps terminal fan-out parents with live children
- audit_log back on db::open; swarm.js badge for reconcile DAGs
jobs are now DAGs of primitive nodes (prebuild, stop-for-update, swap,
reconcile, signal, drain, ...) driven by one scheduler with N build
slots + per-agent lifecycle leases. per-agent power intent (wanted
up/offline) is durable in agent_power.sqlite; Reconcile nodes converge
observed state to it. kills the graceful-stop watcher thread, the
deferred-start follow-up, and the cascade pre-enqueue (fan-out on
MetaLock completion instead). tracker: #2166
Per argus review: std::env::var(..).ok() yields Some("") for an empty
env value, so require_hive_domain would 'succeed' with an empty domain
and emit invalid nix (swarm.peers."" = …). Filter empty so it resolves
to None → the CLI's clear 'domain unset' error fires instead.
Per operator review: hivectl shouldn't make the operator retype this
hive's own domain. Add a HostRequest::HiveDomain admin-socket query
(c0re answers from HYPERHIVE_HIVE_DOMAIN, already in its service env) and
a domain field on HostResponse.
hivectl peer-config now resolves the domain as: --domain override (kept
for offline/scripted use), else the daemon query; errors with a clear
hint when neither resolves. wg init resolves it best-effort too, so it
prints the hand-over peer-config block without --domain (skipped, not
failed, when the daemon is unreachable).
Regenerated docs/tools/hivectl-cli.md.
Replace the stringly-typed infra-control path with an InfraContainer enum
(Ci/Forge/Gateway/Matrix). The variants are the allowlist: serde rejects any
unknown or unsafe name (hive-c0re has no variant) at the wire boundary, so
hive-priv no longer needs a root-side SIBLING_CONTAINERS.contains() check on
ControlInfraContainer — the type enforces it, and 'the daemon can't stop
itself' is a compile-time guarantee.
- priv_proto: InfraContainer enum; manual Serialize/Deserialize + FromStr +
unit_name() all key off one mapping, so the wire form ('hive-ci', …) is
unchanged and there's no drift. ControlInfraContainer.container: String ->
InfraContainer.
- hive-priv / priv_client / server.rs: thread the enum; scoped_infra returns
Vec<InfraContainer>; the control handler uses unit_name().
- agent_server: the infra_admin restart gate parses the name via FromStr
instead of a slice .contains().
- SIBLING_CONTAINERS stays (validate_container_name/_system_name still use it
for journals / general container validation); a test keeps the enum and the
slice in lockstep.
Per review: c0re expands the LifecycleScope to explicit container-name
lists (scoped_agents / scoped_infra) in the dispatch arm, then hands those
lists to handle_stop / handle_start. The 'all agents' flag no longer flows
past the resolution boundary, so downstream consumers (incl. the future
graceful-stop queue) always operate on concrete names. CLI --agents flag
unchanged.
