fix(bootstrap): 关闭 BuildKit 默认 attestation，无改动重跑不再重启 login/lobby；同版本重跑同时重启 registry gate

deploy/bootstrap.sh

@@ -400,6 +400,13 @@ ZYPPER_BACKGROUND_SERVICES=(
 PATH="/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin"
 export PATH
 
+# BuildKit attaches a provenance attestation to every build by default, and it records the
+# build's start time. On Docker's containerd image store the image id is the digest of the
+# index that carries it, so an unchanged rebuild would get a new id each run: the login and
+# lobby pods would restart on every rerun, and each run would push another versioned tag.
+# Without it the id is the manifest digest, which an all-cached build reproduces.
+export BUILDX_NO_DEFAULT_ATTESTATIONS=1
+
 # ---------------------------------------------------------------------------
 # Logging
 # ---------------------------------------------------------------------------
@@ -3254,12 +3261,13 @@ EOF
 }
 
 deploy_bundle() {
-  local prev_api prev_operator
+  local prev_api prev_operator prev_gate
   export KUBECONFIG=/etc/rancher/k3s/k3s.yaml
   write_felis_toml "${STATE_DIR}/felis.pod.toml" "${NODE_IP}"
 
   prev_api="$(deployment_image felis-api api)"
   prev_operator="$(deployment_image felis-operator operator)"
+  prev_gate="$(deployment_image registry registry-gate)"
   if [ -n "$prev_api" ] && [ "$prev_api" != "$FELIS_IMAGE" ]; then
     PREVIOUS_FELIS_IMAGE="$prev_api"
     printf '%s\n' "$prev_api" > "${STATE_DIR}/previous-felis-image"
@@ -3336,7 +3344,7 @@ deploy_bundle() {
     if [ -n "$size" ]; then manifest_args+=(--backup-storage "$size"); fi
   fi
   "$HOST_BIN" manifests "${manifest_args[@]}" | kube apply -f -
-  restart_existing_control_plane "$prev_api" "$prev_operator"
+  restart_existing_control_plane "$prev_api" "$prev_operator" "$prev_gate"
 
   log "waiting for control-plane rollouts"
   local d
@@ -3378,8 +3386,11 @@ deployment_image() {
 # version, or a FELIS_IMAGE the operator reuses). A Deployment whose image changed is rolling
 # from the apply already, and must not be restarted on top: the restart is a second template
 # change, so `rollout undo` would step back to the new image instead of the previous release.
+#
+# The registry pod runs the same binary in its registry-gate and registry-gc containers, so it
+# follows the same rule; the rollout wait below covers it before anything is pushed.
 restart_existing_control_plane() {
-  local prev_api="$1" prev_operator="$2"
+  local prev_api="$1" prev_operator="$2" prev_gate="${3:-}"
   # `if`, not `[ test ] && cmd`: as the LAST command of the function the and-list returns 1
   # when the test is false, which becomes the function's exit status and kills the whole
   # install under `set -Eeuo pipefail` — right after the bundle is applied and before the
@@ -3392,6 +3403,10 @@ restart_existing_control_plane() {
     log "restarting felis-operator onto the rebuilt ${FELIS_IMAGE}"
     kube -n "$CONTROL_NS" rollout restart deployment/felis-operator
   fi
+  if [ "$prev_gate" = "$FELIS_IMAGE" ]; then
+    log "restarting the registry's gate onto the rebuilt ${FELIS_IMAGE}"
+    kube -n "$CONTROL_NS" rollout restart deployment/registry
+  fi
 }
 
 # push_image_to_registry <ref> re-tags a locally built image for the node's

deploy/bootstrap_test.sh

@@ -1670,25 +1670,26 @@ expect "an explicit FELIS_IMAGE is used as given" "reg.example/felis:mine" \
 
 rsblock="$(awk '/^restart_existing_control_plane\(\) \{/,/^}/' "$BS")"
 [ -n "$rsblock" ] || { echo "FAIL: no restart_existing_control_plane found in $BS"; exit 1; }
-run_restart() { # prev-api prev-operator
+run_restart() { # prev-api prev-operator [prev-gate]
   FELIS_IMAGE=reg/felis/felis:v2 CONTROL_NS=felis bash -c '
     set -Eeuo pipefail
     log() { :; }
     kube() { printf "KUBE %s\n" "$*"; }
     '"$rsblock"'
-    restart_existing_control_plane "$1" "$2"
+    restart_existing_control_plane "$1" "$2" "$3"
-    echo DONE' _ "$1" "$2"
+    echo DONE' _ "$1" "$2" "${3:-}"
 }
 
-out="$(run_restart reg/felis/felis:v1 reg/felis/felis:v1)"
+out="$(run_restart reg/felis/felis:v1 reg/felis/felis:v1 reg/felis/felis:v1)"
 case "$out" in
   *"rollout restart"*) echo "FAIL an upgrade restarted the control plane on top of the apply's roll"; fails=$((fails + 1)) ;;
   *DONE*) echo "PASS an upgrade leaves the roll to the apply" ;;
   *) echo "FAIL restart_existing_control_plane died on an upgrade: $out"; fails=$((fails + 1)) ;;
 esac
-out="$(run_restart reg/felis/felis:v2 reg/felis/felis:v2)"
+out="$(run_restart reg/felis/felis:v2 reg/felis/felis:v2 reg/felis/felis:v2)"
 expect "a rerun of the same tag restarts felis-api onto the rebuilt image" "KUBE -n felis rollout restart deployment/felis-api" "$out"
 expect "a rerun of the same tag restarts felis-operator too" "KUBE -n felis rollout restart deployment/felis-operator" "$out"
+expect "a rerun of the same tag restarts the registry's gate too" "KUBE -n felis rollout restart deployment/registry" "$out"
 out="$(run_restart '' '')"
 case "$out" in
   *"rollout restart"*) echo "FAIL a first install restarted Deployments that did not exist"; fails=$((fails + 1)) ;;
@@ -1902,6 +1903,16 @@ expect "a v6 node address gets a v6 rule" "tcp dport 5432 ip6 saddr 2001:db8::7
 rm -rf "$fwdir"
 
 
+# --- reproducible image ids ---------------------------------------------------------------
+# restart_existing_system_servers compares image ids across runs; a default BuildKit
+# provenance attestation (it carries a timestamp) would make every rebuild look new.
+attest_line="$(grep -n '^export BUILDX_NO_DEFAULT_ATTESTATIONS=1$' "$BS" | cut -d: -f1 | head -1)"
+build_line="$(grep -n '^  docker build ' "$BS" | cut -d: -f1 | head -1)"
+if [ -n "$attest_line" ] && [ -n "$build_line" ] && [ "$attest_line" -lt "$build_line" ]; then
+  echo "PASS default build attestations are off before the first docker build"
+else
+  echo "FAIL BUILDX_NO_DEFAULT_ATTESTATIONS=1 must be exported before the first docker build"; fails=$((fails + 1))
+fi
 
 
 # ---------------------------------------------------------------------------------------

docs/troubleshooting.md

@@ -1366,9 +1366,9 @@ runs changed:
 | Component | Restarted when |
 |---|---|
 | `felis-velocity` (the proxy) | its unit, the JRE, `velocity.jar`, `velocity.toml`, the forwarding secret, the felis-link settings or a plugin jar changed, or it was not running. The fingerprint lives in `/etc/felis/velocity.fingerprint`; delete it to force a restart. |
-| login and lobby pods | the rebuilt limbo or lobby image has a new image ID (`/etc/felis/system-server-images`). Each restarts on its own. |
+| login and lobby pods | the rebuilt limbo or lobby image has a new image ID (`/etc/felis/system-server-images`). Each restarts on its own. The installer turns off BuildKit's default provenance attestation (`BUILDX_NO_DEFAULT_ATTESTATIONS=1`): it records the build time, which would give every rebuild a new ID. |
 | PostgreSQL | first install only (`listen_addresses` needs a restart). A rerun reloads the configuration, which keeps connections open. |
-| felis-api, felis-operator | the image tag changed (an upgrade), or a same-version rerun rebuilt it. |
+| felis-api, felis-operator, the registry pod (its gate and GC containers run the felis binary) | the image tag changed (an upgrade), or a same-version rerun rebuilt it. |
 
 **PostgreSQL across reruns.** On hosts without firewalld the installer loads an
 nftables table, `inet felis_postgres`, from `felis-postgres-firewall.service`: