Skip to content

Field and Metric Reference

Precise definitions of the columns, figures and terms used throughout the Logging Report.

Concepts

Term Meaning
Entity load The loading of one entity into one layer during one run. This is what almost every line and every status icon in the report stands for.
Process One complete load run, triggered by a schedule or by hand. A process covers all layers and all entities in the Load Planner Groups it was started for.
Process Id The identifier that ties one process together. Use it to follow a single run from the Landing Zone through Bronze into Silver.
Layer The medallion layer the activity belongs to: Landing Zone, Bronze or Silver.
Type What produced the row: a copy activity, a notebook, or the pipeline that orchestrated them. Bronze and Silver are always notebooks; the Landing Zone uses either, depending on the data source.

Status values

Icon Status Meaning
Succeeded The activity reported a successful end
Started The activity reported a start but no end. Normally in progress; if it is older than the load usually takes, it is more likely a run that ended without being able to report
Failed The activity reported a failure
(blank) No activity Nothing was recorded in the selected period

Where one icon summarises several activities, the most severe status wins: Failed over Started over Succeeded.

Started is not proof of progress

A status only changes when the load reports a result. A run that breaks off abruptly, or one that is cancelled from the portal, never reports one and therefore keeps showing Started for the rest of the retention window. See What Started really tells you.

Time columns

Column Meaning
Start Time When the activity started. On the Processes Run table and in the Process Start Time filter it is the moment the whole process started, which is why it identifies a run.
End Time When the activity finished. Empty while the activity is still running.
Duration (m) / Duration (Min) Run time in minutes. On the Processes Run table it is measured from the process start, so it includes queueing and the time between activities.
Last event time The start time of the most recent activity known to the report. Your freshness indicator.

Row counts

Which counts are available depends on what produced the row.

Column Meaning Copy activity Notebook
Rows Processed Rows retrieved from the source or offered to the layer
Rows New Rows inserted
Rows Updated Rows that changed
Rows Deleted Rows marked as deleted
Rows Unchanged Rows identical to what was already stored

Files show 0 rows

When a Landing Zone entity is a file that is stored as it is — a CSV, a Parquet file, a document — there is nothing to count line by line and Rows Processed shows 0. That is normal and is not a sign of an empty or failed load.

Why a copy activity has no change counts

A copy activity moves source data as-is and does not compare it against what is already stored, so it can only report what it read and wrote. Notebooks do compare, which is why the new, updated, deleted and unchanged counts come from them.

The Landingzone page leaves the change counts out of its table even for notebook-loaded entities, because that layer mixes both kinds of activity.

Filters

Filter Filters on Notes
Load Groups The Load Planner Group that triggered the load An entity can belong to more than one group
Domain Name The domain of the workspace Labelled Domain on the Entity Load History page
Datasource Name The data source the entity comes from Labelled DataSource on the Entity Load History page. Not present on the layer pages
Source Schema The schema, folder or container of the source object
Source Name The table, file or object name
Entities With Status Started, Succeeded or Failed Entity Load History page only. Keeps every entity where that status occurred at least once in the period, not only in the most recent load
Period 24H, 48H, 7D, 30D or YTD A rolling window ending now, not a calendar period
Process Start Time One specific process run Only lists runs inside the selected period
Environment Your environments, for example DVLM and PROD Sits in the header, not in the filter bar
Reset to 24H Default Not a filter but a button Layer pages only. Clears the filter bar, sets Period to 24H and returns the environment to PROD

Filters only offer values that occur in the retained history, and apply to the page you are on.

Column Where it takes you
Monitoring URL on a copy activity row The pipeline run in the Fabric monitoring hub
Monitoring URL on a notebook row The parallel main notebook run in the Fabric Spark monitor
Monitoring URL on a Processes Run row The pipeline run for the whole process

These links are where error messages, logs and full run detail live. Opening them requires access to the underlying Fabric workspace.

Fabric keeps that run detail for 30 days. A link to an older run no longer resolves, even though the report still shows the load itself.

Retention and freshness

Property Value
History in the Logging Report Approximately 90 days
Run detail in Fabric, behind a Monitoring URL 30 days
Furthest you can look back 30D, or YTD since 1 January — whichever reaches further, capped by the retained history
Data delay Typically under 15 minutes between an activity finishing and its row appearing, but it can take up to an hour
Refresh Continuous — reopen or refresh the page to see the latest results

The two retention windows differ, so a load older than 30 days still appears in the report while its run detail in Fabric has already gone. See How far back does it go? for what that means in practice.

Next steps