the page picked the top-level hyperhive attr, which stopped existing when the
agent namespace moved under services.hyperhive.agent, so it rendered no
options. the pick now uses the new root and asserts model, user.privateUsers
and nixBuilds are present. the acp preset submodule options get the
descriptions the now-rendered page requires.
Closes#6136
The docs sections moved the crate reference index to 4-implementation/crates,
but the derivation still wrote the per-crate pages to crates/ at the top of the
tree, so the website showed a separate top-level crates group. They now land
beside the index, and the links back into docs/ get one more ../.
a fresh host's hive-priv made /var/lib/hyperhive/agents root-owned, so
hive-c0re could not create agent dirs (nightly run #7113). the rule
names no mode, so existing hosts are unchanged.
Closes#6143
boundary.md keeps the model, the design rule, why network isolation
matters and host admin socket access; the hive-priv socket activation and
the per-agent socket directory move to 4-implementation/trust-boundary.
security.md keeps the agent trust model and the path-name rule; the nix
builds mode with its threat model, and the hive-c0re privilege separation,
move to their own pages there. Headings and text move as they were, with
one level shaved off the heading depth. Links, code and nix pointers, and
the docs index follow the moves.
Every docs directory moves under 1-getting-started, 2-handbook,
4-implementation or 5-project; the options reference (section 3) comes from
the website and has no directory here. trust-boundary stays whole under the
handbook for now. web-ui, swarm and tools split between handbook and
implementation, and the command-line references move to
2-handbook/reference. Links, path references in code and nix, and the docs
index follow the moves; no page content changes.
a late store root dependency-fails forgejo-swarm-controller-account on
first boot and the refresh never started it, so the controller got no
forge token (vm-nightly #7084). list it as a trust consumer.
Closes#6109
Only "No value found at" and a 403 mean the token is not minted yet and
exit 0. Any other failure exits 1 so the restart policy and store-retry
apply. The token file stays untouched either way.
Closes#6089
tuwunel 1.9.4 exports no metrics. The host collector gets an httpcheck
receiver on the gateway's /_matrix/client/versions (new
services.hyperhive.otel.httpChecks, declared by hive-matrix), and a timer in
the hive-matrix container pushes the database directory's size through the
forwarder's OTLP port, as the attic cache-size unit does. The size step
dereferences the state directory's symlink and pushes nothing for a result
under 4096 bytes.
The builder pushes its build metrics (builds by kind and outcome, build
duration, queue depth and wait, reply publish failures) to its forwarder's
OTLP receiver, which relays them to the swarm collector.
It also registers the standard store client-certificate gauges
(bao.client_cert.*, cert=builder) on the same exporter. They are the
builder's liveness signal: exported every minute whether or not a build
runs. The cert-metrics table in docs/scheduler/observability.md gains the
builder row.
The per-agent /stats page read these from its sqlite; the swarm board had
none of them. hive-agent now counts the tools a turn called, the commands
bash tasks ran (from the bash daemon's own table, newer rows only),
reminders scheduled and delivered, and records the first turn of a fresh
session; the turns tab shows them, with fresh sessions per agent. Claude's
own export has no tool-call counter, so the docs no longer say it does.
The ST4TS page (/stats.html), GET /api/stats-hive, hive-c0re's hive_stats
rollup and the modelPrices option that only fed its cost estimate are
gone. Each agent's own /stats page and the turn-stats database it reads
are unchanged. Setting modelPrices now fails evaluation with a message
saying so.
The per-agent /stats page shows what the swarm agents board lacked about
turns. The new tab reads the turn count and duration the harness already
records: average duration, cache hit rate, tokens per turn, turns per hour
by agent and by model, duration percentiles, and the wake source and result
mixes. The duration histogram used the SDK default buckets, which stop at
10 s while a turn averages 24 s, so every p95 read a flat 10 s; it now has
explicit bounds up to an hour.
U+2692 is text-presentation by default, so it drew as a small
monochrome glyph next to the colour icons of the other links. The
hammer is emoji-presentation by default.
