Add ADR 0005-dynamoDB.md

docs/adrs/0005-dynamoDB.md

@@ -0,0 +1,30 @@
+# 5. Storing DOI state in DynamoDB
+
+## Status
+
+Proposed
+
+## Context
+
+Given that Wiley will not provide a monthly list of only new DOIs and will instead just update a master list of DOIs, we need a system to ensure that we do not deposit the same content twice.
+
+## Decision
+
+We will create an AWS DynamoDB to track the status of each Wiley-provided DOI in the workflow. The steps are outlined below:
+
+* The database will consist of a single table with rows containing information about DOIs that are processed in this workflow.
+* Each row will contain the DOI, the status of the DOI in the workflow, and the number of attempts to process the DOI.
+* The values for the status are:
+    * Processing
+    * Success
+    * Failed and will retry
+    * Failed after too many attempts
+* When a new master list is deposited by Wiley, the workflow will iterate through the list, adding each new DOI and setting the status to `Processing`.
+* Each time the workflow is run, it will attempt to process every DOI with a status of `Processing` and `Failed and will retry`. If the DSpace Submission Service returns a success message to indicate a successful deposit, the status is changed to `Success` and it will not be processed again. If the DSpace Submission Service returns an error message, the status is changed to `Failed and will retry` and it will be processed again the next time the workflow is run.
+* The workflow attempts to process every DOI with the status of `Failed and will retry` until the number of attempts reaches the decided-upon threshold. Once the threshold is reached, the status of the DOI is changed to `Failed after too many attempts`.
+* When a DOI reaches the status of `Failed after too many attempts`, the workflow will not process that DOI in future runs and a notifcation will be sent to stakeholders to let them know that the metadata and content for that DOI will need to be deposited manually into DSpace. 
+
+## Consequences
+
+* There is currently minimal data recorded for each DOI which influenced the decision to use a simple solution like DynamoDB. If the workflow changes and more data needs to be recorded, we may need to choose a more complex database solution.
+* Decisions need to be made on how frequently this workflow will be run and the threshold for the number of attempts before the workflow stops trying to process a DOI.