honeymoose/hapi-fhir
1
0
Fork 0
mirror of https://github.com/hapifhir/hapi-fhir.git synced 2025-02-13 00:14:57 +00:00
Code Issues Packages Projects Releases Wiki Activity

updating documentation

Browse Source
This commit is contained in:
leif stawnyczy 2024-03-08 09:38:50 -05:00
parent 4b22697af9
commit cb39d27dd0
1 changed files with 31 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

34
hapi-fhir-storage-batch2/src/main/java/ca/uhn/fhir/batch2/package-info.java
View File

@ -48,11 +48,39 @@
*
* Job and chunk processing follow state machines described {@link hapi-fhir-docs/src/main/resources/ca/uhn/hapi/fhir/docs/server_jpa_batch/batch2_states.md}
* Chunks have a simple {@link ca.uhn.fhir.batch2.model.WorkChunkStatusEnum state system} with states
* QUEUED, IN_PROGRESS, ERRORED, FAILED, COMPLETED.
* The initial state is QUEUED, and the final states are FAILED, and COMPLETED:
* READY, QUEUED, IN_PROGRESS, ERRORED, FAILED, COMPLETED.
* The initial state is READY, and the final states are FAILED, and COMPLETED.
*
* There are 2 primary systems in play during Batch2 Jobs. A Maintenance Job and the Batch2 Job Notification topic.
*
* <h>The Maintenance Job</h>
*
* This runs every minute and does the following:
*
* <ul>
* <li> Chunks are created QUEUED (NB - should be READY or WAITING) and notification is posted to the channel for non-gated steps.</li>
* <li>Moves READY work chunks to QUEUE and publishes it to the Batch2 Notification topic</li>
* <li>Calculates job progress (% of workchunks in complete status).</li>
* <li>If the job is finished, purges any left over work chunks.</li>
* <li>Cleans up any complete, failed, or cancelled jobs.</li>
* <li>Moves any gated jobs onto their next step.</li>
* <li>If the final step of a (gated) job is a reduction step, a reduction step execution will be triggered.</li>
* </ul>
*
* <h>Processing the Messages</h>
*
* <ul>
* <li>Change the work chunk from QUEUED to IN_PROGRESS</li>
* <li>Change the Job Instance status from QUEUED to IN_PROGRESS</li>
* <li>If the Job Instance is cancelled, change the status to CANCELLED and abort processing</li>
* <li>If the step creates new work chunks, each work chunk will be created in the READY state</li>
* <li>If the step succeeds, the work chunk status is changed from IN_PROGRESS to COMPLETE</li>
* <li>If the step fails, the work chunk status is changed from IN_PROGRESS to either ERRORED or FAILED depending on the severity of the error</li>
* </ul>
*
* <h>The job lifecycle</h>
*
* <ul>
* <li> Chunks are created READY (NB - should be READY or WAITING) and notification is posted to the channel for non-gated steps.</li>
* <li>
* Workers receive a notification and advance QUEUED->IN_PROGRESS.
* {@link ca.uhn.fhir.batch2.api.IWorkChunkPersistence#onWorkChunkDequeue(String)}