Fix Dockhand auth for topology flow and accurate CDC change stats.

Pass DOCKHAND_API_TOKEN to container inventory calls so pipeline_active
and topology animation work after Authentik. Replace ring-buffer-only CDC
stats with minute rollups (no 1000 cap), add 15m/1h/6h/24h window selector
on the Live Changes tab, and poll recent events on an interval.

Co-authored-by: Cursor <cursoragent@cursor.com>
This commit is contained in:
root
2026-07-01 15:14:48 +02:00
parent cdc8aa4bf7
commit b46e7f01dd
10 changed files with 537 additions and 106 deletions
+24 -2
View File
@@ -17,6 +17,13 @@ from database_inventory import collect_database_inventory
from node_registry import NODE_REGISTRY
DOCKHAND_URL = os.getenv("DOCKHAND_URL", "http://10.0.21.45:8082")
DOCKHAND_API_TOKEN = os.getenv("DOCKHAND_API_TOKEN", "")
def _dockhand_headers() -> dict[str, str]:
if DOCKHAND_API_TOKEN:
return {"Authorization": f"Bearer {DOCKHAND_API_TOKEN}"}
return {}
HDFS_NN_URL = os.getenv("HDFS_NN_URL", "http://10.0.21.61:9870")
LAKEHOUSE_HOST = os.getenv("LAKEHOUSE_HOST", "10.0.21.50")
AIRFLOW_URL = os.getenv("AIRFLOW_URL", "http://10.0.21.55:8080")
@@ -110,7 +117,12 @@ async def dockhand_containers(
await _log(log, "cmd", "fetch", f"$ GET {url}")
t0 = time.monotonic()
try:
r = await client.get(f"{DOCKHAND_URL}/api/containers", params={"env": env_id}, timeout=8.0)
r = await client.get(
f"{DOCKHAND_URL}/api/containers",
params={"env": env_id},
headers=_dockhand_headers(),
timeout=8.0,
)
ms = int((time.monotonic() - t0) * 1000)
r.raise_for_status()
data = r.json()
@@ -750,11 +762,12 @@ async def collect_full_lab_context(
if gpu_data is None:
gpu_data = await collect_gpu_metrics(client, log)
docker_raw, db_raw, lake_raw, cc_raw, hdfs, etl, objectscale = await asyncio.gather(
docker_raw, db_raw, lake_raw, cc_raw, gpu_raw, hdfs, etl, objectscale = await asyncio.gather(
dockhand_containers(client, DOCKHAND_ENVS["docker01"], log),
dockhand_containers(client, DOCKHAND_ENVS["db02"], log),
dockhand_containers(client, DOCKHAND_ENVS["lakehouse"], log),
dockhand_containers(client, DOCKHAND_ENV_COMMAND_CENTER, log),
dockhand_containers(client, DOCKHAND_ENVS["gpu_dev"], log),
collect_hdfs(client, log),
collect_etl(client, log),
collect_objectscale(client, log),
@@ -771,6 +784,15 @@ async def collect_full_lab_context(
databases["inventory"] = {"error": str(exc)}
lakehouse = await collect_lakehouse(client, lake_raw, log)
command_center = await collect_command_center(client, cc_raw, log)
if gpu_data is not None and gpu_raw:
gpu_running = sum(1 for c in gpu_raw if c.get("state") == "running")
gpu_data = {
**gpu_data,
"dockhand_env": DOCKHAND_ENVS["gpu_dev"],
"dockhand_containers": _container_rows(gpu_raw, GPU_URL.replace("http://", "").split(":")[0]),
"dockhand_running": gpu_running,
"dockhand_total": len(gpu_raw),
}
try:
governance = collect_governance(log)