[#37] server install 增加重试自检

This commit is contained in:
2025-11-04 11:37:27 +08:00
parent c8279997a4
commit 9858f4471e
4 changed files with 77 additions and 13 deletions

View File

@@ -40,6 +40,8 @@ prepare_data_dirs() {
# still ensure basic directories exist (no chown)
mkdir -p \
"$PKG_ROOT/private/argus/etc" \
"$PKG_ROOT/private/argus/log/elasticsearch" \
"$PKG_ROOT/private/argus/log/kibana" \
"$PKG_ROOT/private/argus/metric/prometheus" \
"$PKG_ROOT/private/argus/metric/prometheus/data" \
"$PKG_ROOT/private/argus/metric/prometheus/rules" \
@@ -153,6 +155,37 @@ YAML
(cd "$PKG_ROOT/compose" && "${COMPOSE[@]}" -p "$PROJECT_NAME" -f docker-compose.yml -f $(basename "$ov") up -d "${services[@]}")
}
# Post bootstrap container-side fixes that do not require sudo on host.
post_bootstrap_fixes() {
# Kibana: ensure /usr/share/kibana/data is a symlink into mounted path to avoid EACCES
if docker ps --format '{{.Names}}' | grep -q '^argus-kibana-sys$'; then
docker exec argus-kibana-sys bash -lc '
set -e
mkdir -p /private/argus/log/kibana && chmod 777 /private/argus/log/kibana || true
if [ -d /usr/share/kibana/data ] && [ ! -L /usr/share/kibana/data ]; then rm -rf /usr/share/kibana/data; fi
if [ ! -e /usr/share/kibana/data ]; then ln -s /private/argus/log/kibana /usr/share/kibana/data; fi
' >/dev/null 2>&1 || true
fi
# Elasticsearch: ensure data path points to mounted path and is writable
if docker ps --format '{{.Names}}' | grep -q '^argus-es-sys$'; then
docker exec argus-es-sys bash -lc '
set -e
mkdir -p /private/argus/log/elasticsearch && chmod 777 /private/argus/log/elasticsearch || true
if [ -d /usr/share/elasticsearch/data ] && [ ! -L /usr/share/elasticsearch/data ]; then rm -rf /usr/share/elasticsearch/data; fi
if [ ! -e /usr/share/elasticsearch/data ]; then ln -s /private/argus/log/elasticsearch /usr/share/elasticsearch/data; fi
' >/dev/null 2>&1 || true
fi
# Bind9: ensure rndc.key exists
if docker ps --format '{{.Names}}' | grep -q '^argus-bind-sys$'; then
docker exec argus-bind-sys bash -lc '
set -e
mkdir -p /etc/bind
if [ ! -f /etc/bind/rndc.key ]; then rndc-confgen -a -c /etc/bind/rndc.key; fi
chmod 644 /etc/bind/rndc.key || true
' >/dev/null 2>&1 || true
fi
}
dns_bootstrap() {
log "DNS bootstrap: initializing shared dns.conf and container resolv.conf"
local etc_dir="$PKG_ROOT/private/argus/etc"
@@ -206,8 +239,31 @@ dns_bootstrap() {
}
selfcheck() {
log "running selfcheck"
bash "$PKG_ROOT/scripts/server-selfcheck.sh" || { err "selfcheck failed"; exit 1; }
# Initial selfcheck with retries to absorb cold starts
local max_retries="${SELF_CHECK_RETRIES:-5}" # 重试次数(不含首次),默认 5
local wait_seconds="${SELF_CHECK_WAIT_SECONDS:-30}" # 每次重试前等待秒数,默认 30s
local attempt=0
while :; do
attempt=$((attempt+1))
if (( attempt == 1 )); then
log "running selfcheck (attempt ${attempt})"
else
log "running selfcheck (attempt ${attempt}/${max_retries}+1)"
fi
if bash "$PKG_ROOT/scripts/server-selfcheck.sh"; then
return 0
fi
# failed
if (( attempt > max_retries )); then
err "selfcheck failed after ${attempt} attempt(s)"
exit 1
fi
log "selfcheck not ready yet; retrying in ${wait_seconds}s..."
sleep "$wait_seconds"
done
}
main() {
@@ -216,6 +272,7 @@ main() {
prepare_data_dirs
load_images
bring_up
post_bootstrap_fixes
dns_bootstrap
selfcheck
log "install completed. See logs in $PKG_ROOT/logs/"