Editorial content · Investigation
Recovering content relationships after a migration
Missing relationships could not be repaired by matching an identifier alone. I worked through source evidence and publication context, then used previews, bounded batches and readback to support controlled repairs. Uncertain cases stayed separate for follow-up.
An identifier alone is not enough
Matching identifier
Matching source relationship
Source context identifies the appropriate target.
- Preview
- Apply a small batch
- Read back and verify
A list of unresolved content relationships mixed active problems with records that no longer needed attention. Some source identifiers also appeared in more than one publication. A plausible identifier match could therefore point to the wrong story.
I questioned the scope and counts, distinguished publication state from whether a relationship could be resolved, and asked for changes to be considered together at article level. The investigation used source-system relationship evidence and the active article content to identify appropriate targets.
With that basis established, the repair could be reviewed before applying it. The runner previewed proposed changes, limited the batch size and saved a checkpoint for each item. Unrelated parts of the article had to remain intact.
A failed request did not always establish that a write had failed. Reading the article back helped determine whether a change had already happened before deciding what to repeat. I also checked the visible result, rather than relying on a successful response alone.
The runner was used in successive production batches. Checkpoints supported review and resumption, while uncertain targets and unsuitable content states remained separate. The result was a bounded repair, with outstanding cases held for further work.