Record installed R19 recovery and onboard K1 acceptance

docs/audits/2026-09-08-node-runtime-and-system-monitor.md

@@ -162,3 +162,36 @@ refusal, and both successful/failing monitor setup across Node/K1 lifecycle
 states. They implement the actual Debian `%I` assertion rule, so the previous
 hyphenated name fails. Real root-installed service and archived samples remain
 an explicit post-install gate, never inferred from stub or fixture success.
+
+
+## R19 installed acceptance — 08.09.2026
+
+At 10:50:10 MSK the owner installer completed with code 0. Node 0.8.16 and
+K1 0.1.14 are both fully configured. K1 status reports available=true locally
+and through the paired Core API. The renamed 16/ndcmonitor cluster is online
+with the original custom data directory; no default cluster or database TCP
+listener exists, and the temporary postgresql-common guard is removed.
+Node/K1/collector/PostgreSQL services are active with NRestarts=0.
+
+Real continuous collection now exposes 201 metrics in ten resource groups.
+Core reports available/fresh=true, storage=ready, backlog=0 and growing durable
+history since 10:49:51 MSK. A sampled idle footprint was 17,281,024 bytes for the
+collector and 41,951,232 bytes for PostgreSQL. This is a point observation,
+not an upper resource bound or load qualification. Browser QA verifies the
+connected device, real CPU/GPU series, missing GPU counter explanations,
+six-hour rollups, and frozen time controls.
+
+The owner explicitly confirmed that the actual onboard Mac mini application
+shows the device, point cloud and camera after the repair. This is an owner
+laboratory acceptance of local K1 viewing, distinct from the earlier synthetic
+WebKit probe. No physical commands were issued by the agent. Long-duration,
+power-loss/network-outage and resource-pressure qualifications remain open.
+
+Node 0.8.16 artifact SHA-256:
+`aaf035c806850635fb1a7f0b732db6005e433716e4e72a4b89a9e2982ab87266`.
+Runtime source commit 14f8375; K1 0.1.14 artifact is unchanged from R18.
+R19 APT plan: one Node upgrade, zero added/removed packages, configuration of
+the two previously incomplete packages. The correction's 15 focused tests
+passed, including nine package failure/migration cases; canonical Node build
+passed. The large shared UI suite was not repeated for this packaging-only
+correction. Private evidence retains both the failed R18 and accepted R19.

tests/test_node_monitor_package.py

@@ -27,7 +27,8 @@ conf=root/'etc/postgresql/16'
 if name=='id':print(0)
 elif name=='install':pathlib.Path(args[-1]).mkdir(parents=True,exist_ok=True)
 elif name=='pg_conftool':
- print(os.environ.get('TEST_DATA_ROOT',str(root/'var/lib/mission-core-monitor-db')) if args[-1]=='data_directory' else '5433')
+ data=os.environ.get('TEST_DATA_ROOT',str(root/'var/lib/mission-core-monitor-db'))
+ print(data if args[-1]=='data_directory' else '5433')
 elif name=='pg_createcluster':
  (conf/args[1]).mkdir();(conf/args[1]/'postgresql.conf').write_text('new database')
 elif name=='pg_renamecluster':