ezer-mishpati/legal-ai
1
0
Fork 0
Code Issues Pull Requests 1 Actions Packages Projects Releases Wiki Activity

feat(ops): host_sync.sh — auto-sync host tree to origin/main (P-#160)
All checks were successful
G12 Leak-Guard / leak-guard (pull_request) Successful in 4s
Details
Lint — undefined names / undefined-names (pull_request) Successful in 10s
Details

Browse Source 
The writer, learning panels, and legal_mcp tools run from the HOST tree
(Paperclip agents claude -p with cwd=~/legal-ai; pm2 services import from
mcp-server/src) — not the container. A merge + Coolify deploy only refreshes the
FastAPI container; host-side code stayed on the last manual `git pull`. On
2026-06-28 the host was 16 commits behind, so a day of learning-loop fixes sat
dormant until pulled by hand.

host_sync.sh (run every minute via cron, like the existing data/training pull)
fast-forwards the host tree to origin/main when it falls behind. Fast-forward
ONLY — never forces, so the perpetually-dirty .taskmaster/tasks/tasks.json is
never clobbered (clean ff leaves it; an upstream conflict logs + bails for manual
handling). Agent runs spawn fresh and pick up new code with no restart; only the
long-running pm2 services that import app code (chat_service, court_fetch_service,
reaper) are restarted, and only when their own files changed (precise, minimal
churn). Logs to data/logs/host_sync.log.

Cron entry (added on the host, flock-guarded):
  * * * * * flock -n /tmp/host_sync.lock bash ~/legal-ai/scripts/host_sync.sh

SCRIPTS.md updated.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
This commit is contained in:
Chaim
2026-06-28 22:20:13 +00:00
parent 9455da7567
commit 80a5daa3e3
2 changed files with 51 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

50
scripts/host_sync.sh Executable file
View File

@@ -0,0 +1,50 @@
#!/usr/bin/env bash
# host_sync.sh — keep the HOST working tree (~/legal-ai) in sync with origin/main.
#
# Why this exists: the writer (block_writer), the learning panels
# (style_lesson_panel), and the legal_mcp MCP tools run from the host tree — the
# Paperclip agents `claude -p` with cwd=/home/chaim/legal-ai, and the long-running
# pm2 services import legal_mcp from mcp-server/src. A merge to main + Coolify
# deploy only refreshes the CONTAINER (FastAPI endpoints); host-side code stays on
# whatever was last `git pull`ed. On 2026-06-28 the host was 16 commits behind, so a
# full day of learning-loop fixes sat dormant until a manual pull. This closes that
# gap (TaskMaster #160).
#
# Safety: fast-forward ONLY (never force, never lose local work). The host tree
# carries a perpetually-dirty .taskmaster/tasks/tasks.json (uncommitted by policy) —
# a clean ff leaves it untouched; if upstream ever touches the same file the ff
# fails and we log + bail rather than clobber it. Agent runs spawn fresh, so they
# pick up new code on the next wakeup with no restart; only the long-running pm2
# services that IMPORT app code are restarted, and only when their own files change.
set -uo pipefail
REPO=/home/chaim/legal-ai
LOG="$REPO/data/logs/host_sync.log"
mkdir -p "$REPO/data/logs"
log() { echo "$(date -u +%Y-%m-%dT%H:%M:%SZ) $*" >> "$LOG"; }
cd "$REPO" || { log "ERROR cd $REPO failed"; exit 1; }
# Only sync when on main (don't disturb a manual branch checkout).
branch=$(git rev-parse --abbrev-ref HEAD 2>/dev/null)
[ "$branch" = "main" ] || { log "skip: on branch '$branch', not main"; exit 0; }
git fetch origin main --quiet 2>>"$LOG" || { log "ERROR git fetch failed"; exit 1; }
local_head=$(git rev-parse HEAD)
remote_head=$(git rev-parse origin/main)
[ "$local_head" = "$remote_head" ] && exit 0 # up to date — quiet no-op
changed=$(git diff --name-only "$local_head" "$remote_head")
if ! git pull --ff-only origin main --quiet 2>>"$LOG"; then
log "ERROR ff-only pull failed ($local_head → $remote_head) — manual intervention needed (likely local tasks.json vs upstream); NOT forcing"
exit 1
fi
log "synced $local_head → $(git rev-parse --short HEAD) ($(echo "$changed" | wc -l) files)"
# Restart only the long-running services whose OWN code changed (precise, minimal
# churn). The agent path (writer/panels via fresh `claude -p`) needs no restart.
restart() { pm2 restart "$1" --update-env >/dev/null 2>>"$LOG" && log "restarted $1"; }
echo "$changed" | grep -q '^mcp-server/src/legal_mcp/chat_service/' && restart legal-chat-service
echo "$changed" | grep -qE '^mcp-server/src/legal_mcp/court_fetch_service/' && restart legal-court-fetch-service
echo "$changed" | grep -q '^scripts/reap_orphan_procs.py' && restart legal-reaper
exit 0