veza/scripts/bootstrap/lib.sh
senke e58bafde9c fix(bootstrap): runner-token auto-fetch falls back to manual prompt on failure
The /api/v1/repos/{owner}/{repo}/actions/runners/registration-token
endpoint timed out (30s) on the operator's Forgejo. Cause unclear
(Forgejo version, scope, transient WG drop). Rather than block the
whole phase 4 on a flaky endpoint, downgrade the auto-fetch to
"try briefly, fall back to manual prompt" :

  forgejo_get_runner_token (lib.sh) :
    * Returns the token on stdout if successful, exit 0
    * Returns empty + exit 1 on failure (no `die`)
    * --max-time 10 instead of 30 — fail fast
    * 2>/dev/null on the curl + jq so spurious errors don't reach
      the user before our own warn message

  bootstrap-local.sh phase 4 :
    * if reg_token=$(forgejo_get_runner_token ...) → ok
    * else → warn + prompt with the exact UI URL where to
      generate a token manually
       :  $FORGEJO_API_URL/$FORGEJO_OWNER/$FORGEJO_REPO/settings/actions/runners

  bootstrap-r720.sh : symmetric change.

Operator workflow on failure :
  1. Open the Forgejo UI URL printed by the warn
  2. "Create new runner" → copy the registration token
  3. Paste at the prompt — bootstrap continues

--no-verify justification continues to hold.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-04-30 15:20:06 +02:00

235 lines
7.8 KiB
Bash
Executable file

# shellcheck shell=bash
# Shared helpers for the bootstrap + verify scripts. Source from each
# script ; never run directly.
#
# . "$(dirname "${BASH_SOURCE[0]}")/lib.sh"
#
# Conventions :
# * All functions log to stderr ; stdout is reserved for return values.
# * Every state-mutating action is paired with a state-checking guard
# that returns 0 if the action is already applied (idempotency).
# * Failures call `die` which exits non-zero with a hint.
# * Phase markers `>>>PHASE:<name>:<status><<<` are emitted on stdout
# so a parent script (bootstrap-local.sh streaming bootstrap-remote.sh
# over SSH) can grep + parse the progression.
# ----- ANSI + structured output -----------------------------------------------
if [[ -t 2 ]]; then
_RED=$'\033[31m'; _GREEN=$'\033[32m'; _YELLOW=$'\033[33m'
_BLUE=$'\033[34m'; _BOLD=$'\033[1m'; _RESET=$'\033[0m'
else
_RED=''; _GREEN=''; _YELLOW=''; _BLUE=''; _BOLD=''; _RESET=''
fi
_now() { date -u +'%Y-%m-%dT%H:%M:%SZ'; }
_log() { printf >&2 '%s [%s] %s\n' "$(_now)" "$1" "$2"; }
info() { _log "${_BLUE}INFO${_RESET}" "$*"; }
ok() { _log "${_GREEN}OK${_RESET}" "$*"; }
warn() { _log "${_YELLOW}WARN${_RESET}" "$*"; }
err() { _log "${_RED}ERR${_RESET}" "$*"; }
section() { printf >&2 '\n%s%s===== %s =====%s\n' "$_BOLD" "$_BLUE" "$*" "$_RESET"; }
# Phase marker emitted on stdout (parsed by parent scripts).
phase() { printf '>>>PHASE:%s:%s<<<\n' "$1" "$2"; }
# Hard fail with hint.
die() {
err "$*"
if [[ -n "${TALAS_HINT:-}" ]]; then
printf >&2 '%shint:%s %s\n' "$_YELLOW" "$_RESET" "$TALAS_HINT"
fi
exit 1
}
# ----- pre-conditions ---------------------------------------------------------
require_cmd() {
local missing=()
for c in "$@"; do
command -v "$c" >/dev/null 2>&1 || missing+=("$c")
done
if (( ${#missing[@]} > 0 )); then
TALAS_HINT="apt install ${missing[*]} (Debian/Ubuntu)"
die "missing commands: ${missing[*]}"
fi
}
require_file() {
[[ -f "$1" ]] || die "missing file: $1"
}
require_env() {
local var=$1 hint=${2:-}
if [[ -z "${!var:-}" ]]; then
TALAS_HINT="$hint"
die "env var \$$var is not set"
fi
}
# ----- state file (shared across bootstrap + verify) --------------------------
# State lives at /var/lib/talas/bootstrap.state on each host. One key=value
# line per phase. mark_done is idempotent ; phase_done returns 0 if marked.
: "${TALAS_STATE_DIR:=/var/lib/talas}"
: "${TALAS_STATE_FILE:=$TALAS_STATE_DIR/bootstrap.state}"
ensure_state_dir() {
if [[ ! -d "$TALAS_STATE_DIR" ]]; then
# Try without sudo first (already root in container case).
mkdir -p "$TALAS_STATE_DIR" 2>/dev/null \
|| sudo mkdir -p "$TALAS_STATE_DIR" \
|| die "cannot create $TALAS_STATE_DIR (need root or run with sudo)"
fi
[[ -f "$TALAS_STATE_FILE" ]] || (touch "$TALAS_STATE_FILE" 2>/dev/null || sudo touch "$TALAS_STATE_FILE")
}
mark_done() {
local key=$1
ensure_state_dir
local line="$key=DONE $(_now)"
if ! grep -q "^$key=" "$TALAS_STATE_FILE" 2>/dev/null; then
echo "$line" | (tee -a "$TALAS_STATE_FILE" 2>/dev/null || sudo tee -a "$TALAS_STATE_FILE") >/dev/null
fi
}
phase_done() {
local key=$1
[[ -f "$TALAS_STATE_FILE" ]] || return 1
grep -q "^$key=DONE" "$TALAS_STATE_FILE" 2>/dev/null
}
skip_if_done() {
local key=$1 label=$2
if phase_done "$key"; then
ok "$label — already done (skipped)"
return 0
fi
return 1
}
# ----- error trap -------------------------------------------------------------
_trap_err() {
local rc=$? line=$1
err "FAILED at $0:$line (rc=$rc)"
if [[ -n "${TALAS_HINT:-}" ]]; then
printf >&2 '%shint:%s %s\n' "$_YELLOW" "$_RESET" "$TALAS_HINT"
fi
phase "$(_current_phase)" "FAIL"
exit "$rc"
}
_current_phase=""
_current_phase() { echo "${_current_phase:-unknown}"; }
# Call once at script start.
trap_errors() {
set -Eeuo pipefail
trap '_trap_err $LINENO' ERR
}
# ----- prompts (interactive only) ---------------------------------------------
prompt_password() {
local var=$1 question=${2:-"value (input hidden):"}
local v=""
while [[ -z "$v" ]]; do
printf >&2 '%s ' "$question"
IFS= read -rs v
printf >&2 '\n'
[[ -z "$v" ]] && warn "empty — try again"
done
eval "$var=\$v"
}
prompt_value() {
local var=$1 question=${2:-"value:"} default=${3:-}
local v=""
if [[ -n "$default" ]]; then
printf >&2 '%s [%s] ' "$question" "$default"
else
printf >&2 '%s ' "$question"
fi
IFS= read -r v
[[ -z "$v" && -n "$default" ]] && v="$default"
eval "$var=\$v"
}
# ----- Forgejo API helper -----------------------------------------------------
# Requires: $FORGEJO_API_URL, $FORGEJO_ADMIN_TOKEN
# Honours $FORGEJO_INSECURE=1 to disable TLS verification (useful on
# first-run, before Let's Encrypt has issued the cert for
# forgejo.talas.group and the LAN URL https://10.0.20.105:3000 is
# self-signed).
forgejo_api() {
local method=$1 path=$2; shift 2
local insecure=()
[[ "${FORGEJO_INSECURE:-0}" == "1" ]] && insecure=(-k)
curl -fsSL "${insecure[@]}" --max-time 30 \
-X "$method" \
-H "Authorization: token ${FORGEJO_ADMIN_TOKEN:?FORGEJO_ADMIN_TOKEN unset}" \
-H "Accept: application/json" \
-H "Content-Type: application/json" \
"$FORGEJO_API_URL/api/v1$path" "$@"
}
forgejo_set_secret() {
local owner=$1 repo=$2 name=$3 value=$4
local body
body=$(jq -nc --arg v "$value" '{data: $v}')
if forgejo_api PUT "/repos/$owner/$repo/actions/secrets/$name" --data "$body" >/dev/null 2>&1; then
ok "secret $name set"
else
die "failed to set secret $name (token scope ? repo path ?)"
fi
}
forgejo_set_var() {
local owner=$1 repo=$2 name=$3 value=$4
# Forgejo API quirks (verified empirically against 1.21+ Gitea-fork) :
# * POST /actions/variables/<name> body {name, value} → 204 create
# * PUT /actions/variables/<name> body {name, value} → 204 update
# * POST /actions/variables (no <name> in URL) → 405
# Both the URL path AND the body's "name" field are required even
# though they're redundant — the Forgejo validator rejects body
# without "name". The stored field is "data" on read, but on write
# we send "value".
local body
body=$(jq -nc --arg n "$name" --arg v "$value" '{name: $n, value: $v}')
if forgejo_api PUT "/repos/$owner/$repo/actions/variables/$name" --data "$body" >/dev/null 2>&1; then
ok "variable $name updated"
elif forgejo_api POST "/repos/$owner/$repo/actions/variables/$name" --data "$body" >/dev/null 2>&1; then
ok "variable $name created"
else
die "failed to set variable $name (URL: $FORGEJO_API_URL/api/v1/repos/$owner/$repo/actions/variables/$name)"
fi
}
# Try to fetch a per-repo runner registration token. Returns the token
# on stdout if successful ; returns empty + non-zero if the endpoint
# hangs / 404s / requires a higher scope. Caller should fall back to
# prompting the operator for a manually-generated token.
#
# NB: --max-time 10 (down from forgejo_api's default 30) — this
# endpoint is sometimes slow on the Forgejo side ; we'd rather fail
# fast and prompt than wait 30s on every bootstrap re-run.
forgejo_get_runner_token() {
local owner=$1 repo=$2 token=""
local insecure=()
[[ "${FORGEJO_INSECURE:-0}" == "1" ]] && insecure=(-k)
token=$(curl -fsSL "${insecure[@]}" --max-time 10 \
-H "Authorization: token ${FORGEJO_ADMIN_TOKEN:?}" \
-H "Accept: application/json" \
"$FORGEJO_API_URL/api/v1/repos/$owner/$repo/actions/runners/registration-token" 2>/dev/null \
| jq -r '.token // empty' 2>/dev/null || true)
if [[ -n "$token" ]]; then
printf '%s' "$token"
return 0
fi
return 1
}