716546d0d5
- prep.go TrackWorkspace mirrors into queue + concurrency store groups
(previously only registry); hydrateWorkspaces reaps filesystem ghosts
(dead PID → failed, persisted back to status.json) so cmdStatus and
out-of-process consumers see coherent state (RFC §15.3)
- sync.go queue read/write goes through go-store first per RFC §16.5
("Queue persists across restarts in db.duckdb"), file remains fallback
for graceful degradation
- statestore.go stateStoreGet helper for go-store-first reads
- tests/cli/restart — new CLI test for RFC §15.7 "dispatch → kill →
restart → no ghost agents" dead-PID reap flow
- 4 new statestore tests: queue group mirror, concurrency refresh,
sync queue persistence, fs ghost reap with disk write-back
Co-Authored-By: Virgil <virgil@lethean.io>
71 lines
2.9 KiB
YAML
71 lines
2.9 KiB
YAML
version: "3"
|
|
|
|
# RFC §15.7 — `tests/cli/restart/` validates that dispatch state survives a
|
|
# kill+restart cycle without leaving ghost agents in the registry. The
|
|
# Taskfile path mirrors the surface under test (`core-agent restart` is not a
|
|
# command — restart is the lifecycle, validated by simulating a stale
|
|
# workspace and confirming `status` reports it as failed/queued, not running).
|
|
tasks:
|
|
test:
|
|
cmds:
|
|
- |
|
|
bash <<'EOF'
|
|
set -euo pipefail
|
|
source ../_lib/run.sh
|
|
|
|
go build -trimpath -ldflags="-s -w" -o bin/core-agent ../../../cmd/core-agent
|
|
|
|
# Use an isolated CORE_WORKSPACE so the test cannot disturb the
|
|
# operator's own .core/ directory.
|
|
workspace="$(mktemp -d)"
|
|
export CORE_WORKSPACE="$workspace"
|
|
export CORE_HOME="$workspace"
|
|
export DIR_HOME="$workspace"
|
|
|
|
# Pre-seed a workspace with status.json showing a process that is
|
|
# not alive — simulating the "dispatch -> kill -> restart" cycle.
|
|
ghost_dir="$workspace/workspace/core/go-io/task-restart"
|
|
mkdir -p "$ghost_dir"
|
|
cat >"$ghost_dir/status.json" <<JSON
|
|
{
|
|
"status": "running",
|
|
"agent": "codex:gpt-5.4",
|
|
"repo": "go-io",
|
|
"org": "core",
|
|
"task": "restart-survival",
|
|
"branch": "agent/restart-survival",
|
|
"pid": 99999,
|
|
"started_at": "2026-04-14T12:00:00Z",
|
|
"updated_at": "2026-04-14T12:00:00Z",
|
|
"runs": 1
|
|
}
|
|
JSON
|
|
|
|
# Run status — the new agent process must NOT report the ghost as
|
|
# "running"; it must mark the workspace failed (RFC §15.3 ghost
|
|
# detection on hydrate).
|
|
output="$(mktemp)"
|
|
run_capture_all 0 "$output" ./bin/core-agent status
|
|
assert_contains "core/go-io/task-restart" "$output"
|
|
|
|
# Verify the ghost did not survive as `running`. The status command
|
|
# surfaces both the workspace and its current state — accepting any
|
|
# non-running label (failed, completed, queued, blocked) confirms
|
|
# the ghost was reaped.
|
|
if grep -E 'core/go-io/task-restart.*\brunning\b' "$output"; then
|
|
printf 'expected ghost agent reaped on restart, still reports running\n' >&2
|
|
cat "$output" >&2
|
|
exit 1
|
|
fi
|
|
assert_contains "failed" "$output"
|
|
|
|
# The reaped status should be persisted back to disk per RFC §15.3 —
|
|
# cross-process consumers (other tooling reading status.json) must
|
|
# see the same coherent state.
|
|
status_file="$ghost_dir/status.json"
|
|
if grep -q '"status":[[:space:]]*"running"' "$status_file"; then
|
|
printf 'expected reaped status persisted to %s\n' "$status_file" >&2
|
|
cat "$status_file" >&2
|
|
exit 1
|
|
fi
|
|
EOF
|