fix: reconcile healthy terminal runners

package/gitea-runner-controller/controller.sh

@@ -194,6 +194,7 @@ gcr_alloc_deferred() {
         reused="$(gcr_record_get "$job_id" "$attempt")"
         vm_id="$(gcr_record_field "$reused" vm_id)"
         if gcr_vm_runner_service "$vm_id" start \
+            && gcr_vm_runner_service "$vm_id" health \
             && gcr_gitea_runner_disabled "$repo" "$(gcr_record_field "$reused" vm_name)" false; then
             reused="$(gcr_record_get "$job_id" "$attempt")"
             reused="$(printf '%s' "$reused" | jq -c '.bootstrapped = true | del(.reused_vm)')"
@@ -410,6 +411,7 @@ gcr_bootstrap_pending() {
 
         if [ "$(gcr_record_field "$rec" reused_vm)" = "true" ]; then
             if gcr_vm_runner_service "$vm_id" start \
+                && gcr_vm_runner_service "$vm_id" health \
                 && gcr_gitea_runner_disabled "$repo" "$runner_name" false; then
                 rec="$(printf '%s' "$rec" | jq -c '.bootstrapped = true | del(.reused_vm)')"
                 gcr_record_put "$job_id" "$attempt" "$rec"
@@ -506,58 +508,14 @@ gcr_reap_finished_jobs() {
                     pending_vm|vm_active) ;;
                     *) gcr_lock_release "$key"; continue ;;
                 esac
-                vm_id="$(gcr_record_field "$rec" vm_id)"
+                finish_status=0
-                if [ "$state" = "completed:success" ] \
+                gcr_vm_finish_terminal "$job_id" "$attempt" "$rec" "$state" reconcile \
-                    && idle_rec="$(gcr_record_idle_json "$rec")"; then
+                    || finish_status="$?"
-                    if ! gcr_lock_acquire idle-pool; then
-                        gcr_lock_release "$key"
-                        continue
-                    fi
-                    runner_name="$(gcr_record_field "$rec" vm_name)"
-                    if ! gcr_gitea_runner_disabled "$repo" "$runner_name" true \
-                        || ! gcr_vm_runner_service "$vm_id" stop; then
-                        gcr_lock_release idle-pool
-                        if gcr_vm_cleanup_start "$job_id" "$attempt" "$rec" \
-                            idle-stop-failed false; then
-                            gcr_event "vm-destroyed" "$job_id" \
-                                "{\"vm_id\":$vm_id,\"reason\":\"idle-stop-failed\",\"via\":\"reconcile\"}"
-                        else
-                            gcr_event "vm-cleanup-pending" "$job_id" \
-                                "{\"vm_id\":$vm_id,\"reason\":\"idle-stop-failed\",\"via\":\"reconcile\"}"
-                        fi
-                        gcr_lock_release "$key"
-                        continue
-                    fi
-                    gcr_record_put "$job_id" "$attempt" "$idle_rec"
-                    idle_expires="$(gcr_record_field "$idle_rec" idle_expires_at)"
-                    gcr_lock_release idle-pool
-                    gcr_lock_release "$key"
-                    gcr_event "vm-idle" "$job_id" "{\"vm_id\":$vm_id,\"expires_at\":$idle_expires,\"via\":\"reconcile\"}"
-                    gcr_log info --ns=sweep "job=$job_id succeeded, retaining vm=$vm_id until $idle_expires"
-                    continue
-                fi
-
-                gcr_log info --ns=sweep "job=$job_id terminal ($state), destroying vm=$vm_id"
-                if [ -n "$vm_id" ] && [ "$vm_id" != "0" ] && [ "$vm_id" != "null" ]; then
-                    case "$state" in
-                        completed:success|completed:cancelled|completed:skipped) ;;
-                        *)
-                            ip="$(gcr_vm_public_ip "$vm_id" || true)"
-                            gcr_vm_collect_diagnostics "$vm_id" "$ip" "$job_id" "$state" || true
-                            ;;
-                    esac
-                    if gcr_vm_cleanup_start "$job_id" "$attempt" "$rec" \
-                        job-completed false; then
-                        gcr_event "vm-destroyed" "$job_id" \
-                            "{\"vm_id\":$vm_id,\"reason\":\"job-completed\",\"state\":\"$state\"}"
-                    else
-                        gcr_event "vm-cleanup-pending" "$job_id" \
-                            "{\"vm_id\":$vm_id,\"reason\":\"job-completed\",\"state\":\"$state\"}"
-                    fi
-                else
-                    gcr_record_del "$job_id" "$attempt"
-                fi
                 gcr_lock_release "$key"
+                case "$finish_status" in
+                    0|2) ;;
+                    *) return "$finish_status" ;;
+                esac
                 ;;
         esac
     done

test/package/gitea-runner-controller/test/reap-diagnostics.sh

@@ -16,6 +16,7 @@ gcr_gitea_job_state() {
   case "$2" in
     101) printf 'completed:success' ;;
     102) printf 'completed:failure' ;;
+    105) printf 'completed:skipped' ;;
     *) return 1 ;;
   esac
 }
@@ -36,7 +37,9 @@ gcr_vm_runner_service() {
   printf 'runner %s vm=%s\n' "$2" "$1" >> "$calls"
 }
 
-gcr_gitea_runner_disabled() { :; }
+gcr_gitea_runner_disabled() {
+  printf 'disabled runner=%s value=%s\n' "$2" "$3" >> "$calls"
+}
 
 record_success='{"job_id":"101","run_attempt":"1","repo":"hinterland/hearth","label":"gross-nix-x86-perf","created_at":"0","ttl_min":480,"vm_id":41,"vm_name":"gcr-101-1","bootstrapped":true,"status":"vm_active"}'
 record_failure='{"job_id":"102","run_attempt":"1","repo":"hinterland/hearth","label":"gross-nix-x86-perf","created_at":"1","ttl_min":480,"vm_id":42,"vm_name":"gcr-102-1","bootstrapped":true,"status":"vm_active"}'
@@ -45,8 +48,14 @@ gcr_record_put 101 1 "$record_success"
 gcr_record_put 102 1 "$record_failure"
 gcr_reap_finished_jobs
 
-grep -q 'destroy vm=42' "$calls"
 grep -q 'diag vm=42 ip=192.0.2.42 job=102 reason=completed:failure' "$calls"
+grep -q 'runner health vm=42' "$calls"
+grep -q 'disabled runner=gcr-102-1 value=true' "$calls"
+grep -q 'runner stop vm=42' "$calls"
+test "$(grep -E '^(diag vm=42|runner health vm=42|disabled runner=gcr-102-1|runner stop vm=42)' "$calls")" = 'diag vm=42 ip=192.0.2.42 job=102 reason=completed:failure
+runner health vm=42
+disabled runner=gcr-102-1 value=true
+runner stop vm=42'
 if grep -q 'diag vm=41' "$calls"; then
   printf 'success job should not collect diagnostics\n' >&2
   exit 1
@@ -56,12 +65,28 @@ if grep -q 'destroy vm=41' "$calls"; then
   printf 'successful job VM should remain idle until billing boundary\n' >&2
   exit 1
 fi
+if grep -q 'destroy vm=42' "$calls"; then
+  printf 'failed bootstrapped reconciled VM should remain idle until billing boundary\n' >&2
+  exit 1
+fi
 jq -e 'select(.status == "idle_vm" and .idle_expires_at == 3600)' \
   "$(gcr_record_path 101 1)" >/dev/null
+jq -e 'select(.status == "idle_vm" and .idle_expires_at == 3601)' \
+  "$(gcr_record_path 102 1)" >/dev/null
+gcr_idle_record_usable "$(gcr_record_get 102 1)"
 idle_once="$(gcr_record_get 101 1)"
 gcr_reap_finished_jobs
 test "$(gcr_record_get 101 1)" = "$idle_once"
-test ! -e "$(gcr_record_path 102 1)"
+
+# Skipped terminal jobs follow same healthy retention policy without diagnostics.
+record_skipped='{"job_id":"105","run_attempt":"1","repo":"hinterland/hearth","label":"nix","created_at":"1","ttl_min":480,"vm_id":45,"vm_name":"gcr-105-1","bootstrapped":true,"status":"vm_active"}'
+gcr_record_put 105 1 "$record_skipped"
+gcr_reap_finished_jobs
+test "$(gcr_record_field "$(gcr_record_get 105 1)" status)" = idle_vm
+if grep -q 'diag vm=45' "$calls" || grep -q 'destroy vm=45' "$calls"; then
+  printf 'healthy skipped VM must be retained without failure diagnostics\n' >&2
+  exit 1
+fi
 
 # Reaper keeps cleanup ownership after DELETE failure and retries next sweep.
 record_delete_fail='{"job_id":"104","run_attempt":"1","repo":"hinterland/hearth","label":"nix","created_at":"1","ttl_min":480,"vm_id":44,"vm_name":"gcr-104-1","bootstrapped":true,"status":"vm_active"}'
@@ -76,6 +101,10 @@ gcr_vm_destroy() {
   printf 'destroy-failed vm=%s\n' "$1" >> "$calls"
   return 1
 }
+gcr_vm_runner_service() {
+  printf 'runner %s vm=%s\n' "$2" "$1" >> "$calls"
+  [ "$1" != 44 ] || [ "$2" != stop ]
+}
 gcr_reap_finished_jobs
 test "$(gcr_record_field "$(gcr_record_get 104 1)" status)" = cleanup_pending
 test "$(gcr_count_active)" = 1
@@ -101,6 +130,10 @@ gcr_gitea_job_state() {
 gcr_vm_public_ip() {
   return 1
 }
+gcr_vm_runner_service() {
+  printf 'runner %s vm=%s\n' "$2" "$1" >> "$calls"
+  return 1
+}
 
 gcr_vm_destroy() {
   printf 'destroy vm=%s\n' "$1" >> "$calls"
@@ -109,5 +142,6 @@ gcr_vm_destroy() {
 gcr_reap_finished_jobs
 
 grep -q 'diag vm=43 ip= job=103 reason=completed:failure' "$calls_ip_fail"
+grep -q 'runner health vm=43' "$calls_ip_fail"
 grep -q 'destroy vm=43' "$calls_ip_fail"
 test ! -e "$(gcr_record_path 103 1)"