Files
compose-files/deploy/git-guard.sh
T
Bot 8fa085ebde feat(git-guard): add -e/--emergency (try alternate Gitea hosts) and -f/--force (skip sync check)
Adds two opt-in flags to break the circular dependency discovered during
the 2026-09-12 traefik/keepalived incident: Traefik down -> VIP/hostname
unreachable -> git-guard can't fetch -> stack-deploy.sh traefik blocked ->
can't redeploy traefik to fix Traefik.

-e/--emergency: probes git.bryanmail.net, then the VIP (192.168.4.30), then
each node's direct IP (.31/.32/.33) on Gitea's direct ingress port 3000
(bypasses Traefik/VIP entirely), switches origin to the first reachable one,
then runs the NORMAL sync logic against it (behind/ahead/diverged handling
unchanged — this only changes which host is used, never skips the safety
checks). Prints a reminder to restore the real origin URL afterward; never
persists the swap anywhere.

-f/--force: skips the sync check entirely, for the genuine last-resort case
where Gitea itself (not just routing) is unreachable. Loud warning banner.

Verified git ls-remote succeeds unauthenticated over plain http against
each candidate before writing this, so no credential-smuggling concern.
2026-09-12 00:03:59 -07:00

446 lines
20 KiB
Bash

#!/usr/bin/env bash
# git-guard.sh — Ensures the compose-files working tree is in sync with Gitea
# before any deploy proceeds. Called automatically by stack-deploy.sh.
#
# Behavior:
# - Clean + up to date -> pass silently
# - Clean + behind (ff-only) -> auto `git pull --ff-only`, then pass
# - Clean + ahead only -> interactive: offer to push; non-interactive: BLOCK
# (unpushed)
# - Unresolved merge conflict -> REFUSE immediately. Never auto-commits over
# markers present conflict markers. Prints remediation options.
# - Dirty + local in sync -> offer to commit + push right now
# with origin (auto in non-interactive/CI runs, after a
# secret-pattern scan of the staged diff)
# - Dirty + local STALE/ -> NEVER commit on top of a stale base. Stash
# diverged vs origin the dirty changes first, resync main with
# origin using the same behind/ahead/diverged
# rules as the clean-tree case, then reapply
# the stash and re-run. On any failure the
# stash is preserved and remediation options
# (with exact commands) are printed.
# - Diverged (local AND -> REFUSE. Never auto-resolves. Prints the
# remote both moved) backup/stash/reset recovery steps and exits.
#
# Exit codes: 0 = safe to deploy, 1 = blocked, needs human intervention
#
# ── Flags (added 2026-09-12, circular-dependency bootstrap incident) ────────
#
# -e, --emergency
# Before the normal fetch, tries each candidate Gitea endpoint in order
# (git.bryanmail.net -> 192.168.4.30 VIP -> .31 -> .32 -> .33 node IPs,
# each over plain http on port 3000, which is Gitea's direct ingress
# port — bypasses Traefik/VIP entirely) and switches `origin` to the
# first one that responds to `git ls-remote` within a short timeout.
# This solves the bootstrap circular dependency where Traefik is down,
# so HTTPS access to git.bryanmail.net is unreachable, so git-guard
# can't fetch, so stack-deploy.sh can't redeploy traefik to fix itself.
# Once a reachable endpoint is found, NORMAL sync logic still runs
# (behind/ahead/diverged handling is unchanged) — this flag only changes
# *which host* is used, never skips the safety checks themselves.
# Prints a loud reminder to restore the real origin URL afterward; never
# commits the swapped URL anywhere.
#
# -f, --force
# Skips the sync check ENTIRELY — no fetch, no comparison, no commit/push
# offer. Deploys whatever is on disk right now, as-is. This is the last
# resort for a genuine emergency where NONE of the candidate hosts in
# --emergency are reachable (e.g. Gitea itself is down, not just
# routing). Prints a loud warning banner. Never use this for routine
# work — it defeats the entire purpose of this script and is the exact
# failure mode (deploying a stale/unreviewed tree) git-guard exists to
# prevent.
#
# Both flags are passed through from stack-deploy.sh's own -e/-f flags;
# see that script's header for the calling convention.
# ─────────────────────────────────────────────────────────────────────────
set -euo pipefail
DIR="/volume1/docker/compose-files"
cd "$DIR"
# ---- Flag parsing ----
EMERGENCY=0
FORCE=0
for arg in "$@"; do
case "$arg" in
-e|--emergency) EMERGENCY=1 ;;
-f|--force) FORCE=1 ;;
esac
done
# ---- Force mode: skip everything ----
if [ "$FORCE" -eq 1 ]; then
echo "!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!"
echo "!! FORCE MODE (-f/--force): git-guard sync check SKIPPED ENTIRELY."
echo "!! Deploying whatever is on disk right now, as-is. No fetch, no"
echo "!! comparison with origin/main was performed. This is a LAST RESORT"
echo "!! for emergencies where origin is completely unreachable — verify"
echo "!! independently that the local tree is what you intend to deploy."
echo "!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!"
exit 0
fi
# Non-interactive detection (Woodpecker/cron have no TTY on stdin)
INTERACTIVE=0
[ -t 0 ] && INTERACTIVE=1
echo "==> git-guard: checking repo sync state"
# ---- Emergency mode: find a reachable Gitea endpoint before fetching ----
# Candidate order: public hostname (normal path) -> VIP -> each node's direct
# IP. Each is tried over plain http on port 3000 (Gitea's direct ingress
# port, published outside Traefik — see traefik/traefik.yaml git service
# port mapping), since the whole point is to bypass Traefik/VIP when THOSE
# are what's broken. A short `git ls-remote` timeout keeps an unreachable
# candidate from stalling the whole check for long.
if [ "$EMERGENCY" -eq 1 ]; then
echo "!! EMERGENCY MODE (-e/--emergency): probing candidate Gitea endpoints"
echo "!! (bypassing the normal https://git.bryanmail.net path if needed)..."
ORIGINAL_URL="$(git remote get-url origin)"
CANDIDATES=(
"https://git.bryanmail.net/homelab/compose-files.git"
"http://192.168.4.30:3000/homelab/compose-files.git"
"http://192.168.4.31:3000/homelab/compose-files.git"
"http://192.168.4.32:3000/homelab/compose-files.git"
"http://192.168.4.33:3000/homelab/compose-files.git"
)
FOUND=""
for candidate in "${CANDIDATES[@]}"; do
echo -n " trying $candidate ... "
if timeout 5 git ls-remote "$candidate" HEAD >/dev/null 2>&1; then
echo "OK"
FOUND="$candidate"
break
else
echo "unreachable"
fi
done
if [ -z "$FOUND" ]; then
echo "!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!"
echo "!! EMERGENCY MODE: none of the candidate endpoints responded."
echo "!! Gitea itself may be down (not just routing) — this is beyond what"
echo "!! an alternate host path can fix. Options:"
echo "!! A) Diagnose Gitea directly: check the git_gitea-server service"
echo "!! and container on docker-1."
echo "!! B) If you are certain the on-disk tree is correct and Gitea is"
echo "!! genuinely unreachable, re-run with -f/--force instead — but"
echo "!! read that flag's warning carefully first."
echo "!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!"
exit 1
fi
if [ "$FOUND" != "$ORIGINAL_URL" ]; then
git remote set-url origin "$FOUND"
echo "==> origin temporarily switched to: $FOUND"
echo "!! REMINDER: once the normal path (git.bryanmail.net / Traefik) is"
echo "!! confirmed healthy again, restore the real origin URL:"
echo "!! cd $DIR && git remote set-url origin \"$ORIGINAL_URL\""
echo "!! This swap is never committed anywhere and only affects this"
echo "!! local checkout's git config."
else
echo "==> Normal origin URL ($FOUND) is reachable — no swap needed."
fi
echo
fi
git fetch origin --quiet
LOCAL="$(git rev-parse main)"
REMOTE="$(git rev-parse origin/main)"
BASE="$(git merge-base main origin/main)"
DIRTY=0
git status --porcelain | grep -q . && DIRTY=1
SECRET_PATTERN='(-----BEGIN [A-Z]+ PRIVATE KEY-----|AKIA[0-9A-Z]{16}|xox[baprs]-[0-9a-zA-Z-]+|password[[:space:]]*[:=][[:space:]]*[^$ ]|api[_-]?key[[:space:]]*[:=][[:space:]]*[^$ ])'
# resync_with_origin <local_sha> <remote_sha> <base_sha>
#
# Handles the behind/ahead/diverged cases against a CLEAN working tree.
# Shared by both the "tree was already clean" path and the new
# "dirty tree turned out to be stale, so we stashed first" path, so the
# two paths can never drift out of sync with each other.
#
# Returns 0 if it's now safe to deploy, 1 if it could not safely resolve
# (guidance already printed to stdout in that case).
resync_with_origin() {
local local_sha="$1" remote_sha="$2" base_sha="$3"
# ---- Case: fully in sync ----
if [ "$local_sha" = "$remote_sha" ]; then
echo "==> In sync with origin/main ($local_sha). OK to deploy."
return 0
fi
# ---- Case: behind only (fast-forwardable) ----
if [ "$local_sha" = "$base_sha" ]; then
echo "!! Local main is behind origin/main."
if [ "$INTERACTIVE" -eq 1 ]; then
read -rp "Fast-forward pull now? [y/N] " ans
else
ans="y"
echo "(non-interactive session — auto fast-forwarding)"
fi
if [[ "$ans" =~ ^[Yy]$ ]]; then
git pull --ff-only origin main
echo "==> Fast-forwarded to $(git rev-parse --short main). OK to deploy."
return 0
else
echo "Aborting - pull manually, then retry:"
echo " cd $DIR && git pull --ff-only origin main"
return 1
fi
fi
# ---- Case: ahead only (local commits not yet pushed) ----
if [ "$remote_sha" = "$base_sha" ]; then
echo "!! Local main is AHEAD of origin/main (unpushed commits):"
git log --oneline "origin/main..main"
echo
if [ "$INTERACTIVE" -eq 1 ]; then
read -rp "Push local commits to origin/main now? [y/N] " ans
else
ans="n"
echo "(non-interactive session — will NOT auto-push ahead commits; needs human review)"
fi
if [[ "$ans" =~ ^[Yy]$ ]]; then
git push origin main
echo "==> Pushed. OK to deploy."
return 0
else
echo "Aborting. Review with:"
echo " cd $DIR && git log origin/main..main"
echo "Then push manually when ready:"
echo " git push origin main"
return 1
fi
fi
# ---- Case: true divergence (both ahead and behind) — NEVER auto-fix ----
echo "!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!"
echo "!! DIVERGED: local main and origin/main have both moved independently."
echo "!!"
echo "!! Local-only commits:"
git log --oneline "$base_sha..main" | sed 's/^/!! /'
echo "!!"
echo "!! Remote-only commits:"
git log --oneline "$base_sha..origin/main" | sed 's/^/!! /'
echo "!!"
echo "!! This requires a human decision - git-guard will NOT auto-resolve this."
echo "!!"
echo "!! Before stashing anything new, check whether a git-guard safety stash"
echo "!! ALREADY exists from this same run (avoids confusing duplicate stashes):"
echo "!! git stash list"
echo "!!"
echo "!! Recommended recovery:"
echo "!! 1. tar backup: tar czf /volume1/docker/compose-files-backup-\$(date +%Y%m%d-%H%M%S).tar.gz -C /volume1/docker compose-files"
echo "!! 2. name the branch: git branch backup/pre-reset-\$(date +%Y%m%d)"
echo "!! 3. stash any NEW uncommitted state only if 'git stash list' above"
echo "!! didn't already show one for this run:"
echo "!! git stash push -u -m 'pre-reset-snapshot'"
echo "!! 4. reset to origin: git reset --hard origin/main"
echo "!! 5. selectively restore needed files from the stash/backup branch:"
echo "!! git stash list"
echo "!! git stash show -p stash@{N}"
echo "!! git stash apply stash@{N} # 'apply' keeps the stash as a backup; use 'pop' to also drop it"
echo "!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!"
return 1
}
# ---- Case: unresolved merge conflict already present ----
# Can happen if a PRIOR git-guard run's `git stash pop` conflicted and the
# resulting conflict markers were never resolved before the next deploy
# attempt. Must be checked BEFORE the dirty-tree commit flow below, because
# an unmerged path shows up as "dirty" too, and `git add -A` would silently
# stage the literal <<<<<<< / ======= / >>>>>>> markers into a real commit.
if git ls-files -u | grep -q .; then
echo "ERROR: unresolved merge conflict markers present in the working tree."
echo "Refusing to auto-commit over a conflict — this would push literal"
echo "<<<<<<< / ======= / >>>>>>> markers to origin/main."
echo
echo "Conflicted paths:"
git diff --name-only --diff-filter=U | sed 's/^/ /'
echo
echo "Remediation options:"
echo " A) Resolve the conflict by hand, then commit and push:"
echo " cd $DIR"
echo " git status # see conflicted paths"
echo " git diff # inspect the conflict markers"
echo " \$EDITOR <conflicted-file> # remove markers, keep correct content"
echo " git add <conflicted-file>"
echo " git commit -m 'resolve git-guard stash-pop conflict'"
echo " git push origin main"
echo " B) Discard the conflicted merge attempt entirely and start clean from"
echo " origin/main, then decide separately whether to re-apply anything"
echo " from a prior safety stash:"
echo " cd $DIR"
echo " git checkout -- ."
echo " git reset --hard origin/main"
echo " git stash list # look for a git-guard-safety-stash-* entry"
echo " git stash show -p stash@{N} # inspect before deciding"
echo " C) Once resolved (via A or B) and no longer needed, clean up the stash:"
echo " git stash drop stash@{N}"
exit 1
fi
# ---- Case: dirty tracked changes ----
if [ "$DIRTY" -eq 1 ]; then
echo "!! WORKING TREE DIRTY — uncommitted changes detected:"
git status --short
echo
# If local is ALSO stale/diverged from origin, committing right now would
# create a doomed commit on top of a base that's about to be rejected on
# push (this is exactly what caused a real incident: a stray on-disk edit
# sat on a checkout that was 4 commits behind, git-guard auto-committed
# anyway, then the push bounced). Stash first, resync safely using the
# same rules as the clean-tree path, then reapply.
if [ "$LOCAL" != "$REMOTE" ]; then
echo "!! Local main is ALSO stale/diverged from origin/main."
echo " Refusing to commit on top of a stale base — stashing the dirty"
echo " changes safely first, then resyncing with origin."
echo
STASH_MSG="git-guard-safety-stash-$(date -u +%Y%m%dT%H%M%SZ)"
if ! git stash push -u -m "$STASH_MSG"; then
echo "ERROR: 'git stash push' itself failed (disk full, permissions, or"
echo " some other git error). Your changes are still on disk,"
echo " uncommitted — nothing has been lost, but git-guard cannot"
echo " proceed safely until this is resolved."
echo
echo "Remediation options:"
echo " A) Check disk space and permissions, then retry the deploy:"
echo " df -h $DIR"
echo " ls -la $DIR"
echo " B) Identify and manually move aside whatever is blocking the stash,"
echo " then retry:"
echo " cd $DIR"
echo " git status --short # find the offending path(s)"
echo " mv <path> <path>.bak-\$(date +%s)"
echo " C) Inspect the raw git error above for specifics before proceeding."
exit 1
fi
echo "==> Stashed as: $STASH_MSG"
if resync_with_origin "$LOCAL" "$REMOTE" "$BASE"; then
echo "==> Resync succeeded. Reapplying stashed changes..."
if git stash pop; then
echo "==> Stash reapplied cleanly. Re-checking sync state..."
exec bash "$0" "$@"
else
echo "ERROR: 'git stash pop' did not complete successfully."
echo
if git ls-files -u | grep -q .; then
echo "This is a MERGE CONFLICT — your stashed changes were partially"
echo "applied and conflict markers (<<<<<<< / ======= / >>>>>>>) are now"
echo "in the working tree. The stash itself is still preserved as a backup."
echo
echo "Remediation options:"
echo " A) Resolve the conflict by hand, then commit and push:"
echo " cd $DIR"
echo " git status # see conflicted paths"
echo " git diff # inspect the markers"
echo " \$EDITOR <conflicted-file> # remove markers, keep correct content"
echo " git add <conflicted-file>"
echo " git commit -m 'resolve git-guard stash-pop conflict'"
echo " git push origin main"
echo " git stash list # confirm which entry is: $STASH_MSG"
echo " git stash drop stash@{N} # once confirmed no longer needed"
echo " B) Abandon this merge attempt and fall back to a clean, resynced"
echo " tree, then re-apply the change manually with full visibility:"
echo " cd $DIR"
echo " git checkout -- ."
echo " git reset --hard origin/main # now matches origin, no conflict"
echo " git stash list # find: $STASH_MSG"
echo " git stash show -p stash@{N} # review the content"
echo " git stash apply stash@{N} # 'apply' keeps the backup; use 'pop' to also drop it"
else
echo "This looks like an UNTRACKED-FILE COLLISION, not a merge conflict"
echo "(a file added upstream shares a path with an untracked file in your"
echo "stash). No conflict markers were written; the stash was NOT applied"
echo "and remains fully intact."
echo
echo "Remediation options:"
echo " A) Move the colliding upstream file aside, pop, then reconcile:"
echo " cd $DIR"
echo " git status --short # identify the colliding path"
echo " mv <path> <path>.upstream-\$(date +%s)"
echo " git stash pop"
echo " diff <path> <path>.upstream-* # reconcile manually, then remove the .upstream-* backup"
echo " B) Inspect the stash without applying, and hand-merge the needed"
echo " pieces instead:"
echo " git stash list # find N"
echo " git stash show -p stash@{N}"
fi
echo
echo "Your stash reference for this run: $STASH_MSG"
exit 1
fi
else
echo "ERROR: could not safely resync with origin/main."
echo "Your uncommitted changes are preserved in the stash: $STASH_MSG"
echo
echo "Remediation options:"
echo " A) Follow the manual recovery steps printed above (from the"
echo " behind/ahead/diverged case), THEN reapply your change:"
echo " cd $DIR"
echo " git stash list # find: $STASH_MSG"
echo " git stash apply stash@{N} # or 'pop' to also drop it once resynced"
echo " B) If the stashed change is no longer needed (e.g. it's already"
echo " represented in a since-merged PR), verify then drop it:"
echo " git stash show -p stash@{N}"
echo " git stash drop stash@{N}"
exit 1
fi
fi
if [ "$INTERACTIVE" -eq 1 ]; then
read -rp "Commit and push these changes to origin/main now? [y/N] " ans
else
ans="y"
echo "(non-interactive session — auto-committing and pushing)"
fi
if [[ "$ans" =~ ^[Yy]$ ]]; then
git add -A
if git diff --cached | grep -Eiq "$SECRET_PATTERN"; then
echo "ERROR: possible secret detected in staged changes. Refusing to auto-commit."
echo "Review manually:"
echo " cd $DIR && git diff --cached"
git reset
exit 1
fi
git commit -m "chore(auto): git-guard autofix - commit local changes before deploy $(date -u +%Y-%m-%dT%H:%M:%SZ)"
if git push origin main; then
echo "==> Pushed. Re-checking sync state..."
exec bash "$0" "$@"
else
echo "ERROR: push failed (likely diverged from origin). Aborting deploy."
echo "Run:"
echo " cd $DIR && git status"
exit 1
fi
else
echo "Aborting deploy - commit or stash changes manually, then retry:"
echo " cd $DIR"
echo " git add -A && git commit -m 'your message' && git push origin main"
echo " # or: git stash push -u -m 'manual-stash'"
exit 1
fi
fi
# ---- Clean tree: resync with origin using the shared logic above ----
if resync_with_origin "$LOCAL" "$REMOTE" "$BASE"; then
exit 0
else
exit 1
fi