Audit
The Audit page of a model generates discovery and data lineage reports for
auditing. Both are pdf documents that are created as drafts and can then be
signed.
| Report type | What it documents | Requirement |
|---|---|---|
| Discover | The labels found in the tap and who confirmed them | A discover of the tap |
| Data Lineage | Where each column was written and how it was transformed | None. It shows the data lineage recorded by the jobs of the model |
Creating a report
Click Create report. The Create Report dialog asks you to Select report type:
┌───────────────────────────────────────────────┐
│ Create Report [X] │
├───────────────────────────────────────────────┤
│ Select report type │
│ [ Discover ] [ Data Lineage ] │
│ │
│ Include destination types │
│ [x] Dataset │
│ [x] Sink │
│ [x] Original Tap │
├───────────────────────────────────────────────┤
│ [Cancel] [Create] │
└───────────────────────────────────────────────┘- Discover creates a discovery report. You have to run a discover on the tap before you can create one.
- Data Lineage shows Include destination types, with Dataset, Sink and Original Tap checked by default. Uncheck the ones you do not want in the report. Create is disabled until at least one is checked.
Reports table
| Column | Description |
|---|---|
| When | Date the report was created |
| Report Type | Discover (blue tag) or Data Lineage (purple tag) |
| Schema | Schema version of the discovery report. - for lineage reports |
| Coverage | % of entities confirmed. - for lineage reports |
| Status | Draft or Signed |
| Who | User who created the report |
| Actions | View / Download, and Discard while the report is not signed |
Open a report with View / Download. Its page has the Download Report and Sign buttons.
Discovery report
The discovery report describes the current situation of the tap. It includes information about:
-
When and who has started a new discover.
-
Labels have been added, modified or deleted
-
The person responsible for each of the changes made in the discover.
-
The % of entities confirmed at the time of generating the draft report
These reports, of a informative nature, can be generated at any time, although it is highly recommended to generate them when all entities are confirmed.
Once the report is generated, it will be created in draft status where it will
be possible to view it, discard it or sign it.
A signed report cannot be deleted or modified and will reflect the exact situation of the tap at a given time.
Data lineage report
The data lineage report is a PDF version of the Data Lineage view of the model. It does not need a discover.
The report contains:
- A header Data Lineage Report:
<tap>/<model>. - A table with Name, Created By, Created At, Tap, Model,
Destination Types (
Allwhen no type was stored), Signed-off By, Signed-off At and Comments (the last three are filled in when the report is signed). - A summary line with the number of destinations and columns, and the colour legend of the data path (Tap, Entity, Rule, Dataset, Sink).
- One section per destination, first the datasets, then the sinks and then the
original tap, each one starting on a new page. The title is
<type>: <name>, followed by the number of columns and of entities with transformations.
Each section lists only the entities that have transformed fields. Their fields are grouped by data path, with the path drawn as coloured badges, and shown in a table:
| Column | Description |
|---|---|
| Field | Column name |
| Type | Column type |
| Transformation | Transformation applied (see Data Lineage) |
| Details | Faker function used, or name of the saved custom function |
| Rule | Rule that applied the transformation |
| Last Updated | Date the column lineage was last written |
Draft reports carry a D R A F T watermark.
A lineage report keeps a copy of the lineage it shows, taken when the report is created. The PDF is always built from that copy, so jobs run after the report, before or after it is signed, do not change it. Reports created before this copy existed show the current lineage until they are signed, and keep the lineage of the moment they were signed from then on.