Interactive demo

Batch processing under failure

A 100,000-record policy job with ten steps. Inject a failure partway through, then see where the job stopped, what was rolled back, and how the restart avoids processing anything twice.

Job: 100,000 policy records, 10 steps, chunks of 1,000 records. Each chunk commits in one transaction and records a checkpoint.

Job status

Not started

Last checkpoint

None yet

Records committed

0

Rolled back (never committed)

0

Quarantined

0

Repeat writes prevented

0

Steps

  1. 01Read policy extractpending
  2. 02Validate fieldspending
  3. 03Normalize dates and currencypending
  4. 04Write policy rowspending
  5. 05Calculate premiumspending
  6. 06Write premium rowspending
  7. 07Apply endorsementspending
  8. 08Reconcile totalspending
  9. 09Write audit trailpending
  10. 10Publish summarypending

Execution log

  1. No runs yet.

This is a browser simulation of how a chunked, checkpointed job behaves. It does not run Spring Batch, does not connect to a database, and does not use any client's data or code. The record counts and the failure are illustrative.