Run once each morning · —Progress saves automatically for today onlyLast run: 2026-10-12 00:06 UTC
0 / 16
Pipeline MonitorConfirmed
Azure Synapse Analytics
Pipelines run in Azure Synapse Analytics (not Fabric). Workspace: aca-synapse-workspace · Resource group: ACA-PBI-RG.
Open pipeline runs monitor →
All pipeline runs in the last 24h succeeded
Checked 2026-10-12 · az synapse pipeline-run query-by-workspace, window 2026-10-11 00:00 – 2026-10-12 00:00 UTC — verified live
No pipeline runs stuck "In progress" beyond expected duration
Checked 2026-10-12 · az synapse pipeline-run query-by-workspace, elapsed time per active run — verified live
No missed or failed trigger fires
Checked 2026-10-12 · az synapse trigger-run query-by-workspace, window 2026-10-11 00:00 – 2026-10-12 00:00 UTC — verified live
Cloud Connections CheckConfirmed
No on-prem gateway in use
All 6 semantic models share one Personal Cloud Connection (not a gateway) — a known single point of failure, tracked separately as a one-time migration fix rather than a daily item. This panel's daily check is below.
Alert: no refresh failed due to a broken cloud connection
Checked 2026-10-12 · Detected via each model's refresh-history error type (DataSourceAccessError / Mashup / OAuth-token errors = connection failure, vs. capacity or timeout errors which aren't) — verified live
Data Flow ChecksConfirmed
Azure Synapse Analytics
This workspace defines no Data Flow activities at all — data-flow list is empty and none of the 3 pipelines contain an ExecuteDataFlow activity. Data movement is done by Copy activities, so the items below are assessed against those. Note: the allowlisted CLI form az synapse activity-run query-by-pipeline-run needs --name, not the documented --pipeline-name flag. 2026-10-12: drilled into the 14:01 UTC Nightly chain: Full Overnight Pipeline (40cde00d, 2 ExecutePipeline wrappers) → Dynamic Pipeline (5f83e76e, 98 activity-run entries returned, zero Failed/InProgress, zero non-empty errorCode values) → nested Static Pipelines (269dece1, 5 named Copy activities). Also sampled the standalone 00:03 UTC hourly Static Pipelines run (cbd0ae80). All Copy activities across every run checked Succeeded, zero error codes. The nightly-vs-hourly Copy duration ratios remain healthy for a third check running: City nightly 140.9s vs. 131.9s hourly (1.07x), ESWQueryActivityEntryTime nightly 140.9s vs. 131.8s hourly (1.07x) — see df-4. Separately, City's row count broke its multi-week exact freeze by exactly +1 row (3,873,931 → 3,873,932) since the 2026-09-28 check — still Fail, see df-3.
Open pipeline runs monitor →
Every Data Flow activity in today's runs succeeded
Checked 2026-10-12 · az synapse activity-run query-by-pipeline-run --name — verified live
No schema or validation errors in the activity output
Checked 2026-10-12 · Column mismatches, type errors, missing source fields — verified live
Row counts are in the expected range vs. prior day
Checked 2026-10-12 · Sink row count shown on the Data Flow activity's output tab — verified live
Root cause: Unknown pending investigation -- the City copy activity itself succeeds every run with no error, but the row count sourced from SELECT * FROM City moved by only 1 row over the last two weeks after being bit-for-bit frozen at 3,873,931 for multiple prior checks, which is inconsistent with an actively maintained table and consistent with either a near-static reference table or a stuck/cached upstream source.
Impact: Any downstream report joining on "City" is safe from a data-freshness standpoint only if the table is genuinely meant to barely change; if it isn't, those reports have been silently serving stale geography data for well over a month with the pipeline reporting "Succeeded" the whole time.
Next step: Confirm with the Synapse/source-database owner whether "City" is expected to be a near-static reference table; if not, check the source SQL Server table directly for recent inserts/updates that the Copy activity isn't picking up. Still open as of 2026-10-12 -- no owner response yet.
Run duration is within normal bounds
Checked 2026-10-12 · No timeout warnings or unusually long Data Flow activity runs — verified live
Model Refresh CheckLive data
6 semantic models
2026‑10‑12: model list re-confirmed present via Fabric MCP (list_items, all 6 names/IDs match, no drift) across both workspaces. Refresh history pulled live for all 6 models via az rest --method get --url .../refreshes. model-3 shows its latest refresh completing cleanly (2026-10-11 23:15 UTC) after two Resource Governance (out-of-memory) retries earlier in that run. model-4's latest refresh (2026-10-11 23:11 UTC) completed cleanly on the first attempt, with one earlier Resource Governance retry that day (21:10 UTC) that self-recovered. model-1 and model-2 remain stuck on their old "Disabled" schedule, unchanged from prior checks and now well over a year stale in both cases.
Root cause: The scheduled refresh for this semantic model was disabled around 2025-09-07, likely after repeated processing failures on the "Day Buckets" table a few days earlier, and nobody re-enabled it since.
Impact: Any report or dashboard built on "Acacia Data Lake Landing Zone Report" has been showing data frozen as of 2025-09-06 for roughly 401 days (well over a full year), with no automated alerting on this until this check surfaced it.
Next step: Confirm with the model owner whether this dataset is still in active use; if so, re-enable the scheduled refresh in the Power BI service and run a manual refresh to validate it completes cleanly, watching for a recurrence of the "Day Buckets" processing error. Still open as of 2026-10-12.
Root cause: The scheduled refresh for this dataset was disabled on 2025-03-15 with no preceding failures, suggesting an intentional deprecation (possibly in favor of "Report Usage Metrics Model") that was never cleaned up or documented.
Impact: If this model still backs any live report, that report has been showing data no newer than 2025-03-14 for about 577 days; if it's genuinely superseded, it's just dead weight but still worth confirming rather than assuming.
Next step: Check whether any report or dashboard still references this semantic model; if none do, archive/delete it, otherwise re-enable its scheduled refresh. Still open as of 2026-10-12 -- no change since the prior check.