my prefect server setup
prefect-metrics.waow.tech
python
orchestration
fix: finished_job_ttl=300s + show dbt compile stdout on failure
- set finished_job_ttl=300 on kubernetes-pool so completed job pods
auto-delete after 5 min, preventing overlayfs snapshot accumulation
- include both stdout and stderr in dbt compile error message so the
actual dbt error is visible in prefect logs
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
+4
-1
+1
-1
flows/enrich.py
+1
-1
flows/enrich.py
···
27
27
capture_output=True, text=True,
28
28
)
29
29
if result.returncode != 0:
30
-
raise RuntimeError(f"dbt compile failed:\n{result.stderr}")
30
+
raise RuntimeError(f"dbt compile failed:\nSTDOUT:\n{result.stdout}\nSTDERR:\n{result.stderr}")
31
31
logger.info("dbt compile OK")
32
32
33
33
settings = PrefectDbtSettings(
+3
scripts/patch_work_pool.py
+3
scripts/patch_work_pool.py
···
49
49
# this makes deps available to the prefect engine without any sys.path hacks
50
50
envs["UV_PROJECT_ENVIRONMENT"] = "/usr/local"
51
51
52
+
# clean up finished job pods after 5 minutes — prevents overlayfs snapshot accumulation
53
+
props.setdefault("finished_job_ttl", {})["default"] = 300
54
+
52
55
await c.update_work_pool(
53
56
"kubernetes-pool",
54
57
WorkPoolUpdate(base_job_template=t),