| Issue Number | 314 |
|---|---|
| Summary | Prevent future record duplications |
| Created | 2015-08-20 20:00:10 |
| Issue Type | Bug |
| Submitted By | alan |
| Assigned To | Kline, Bob (NIH/NCI) [C] |
| Status | Closed |
| Resolved | 2016-09-09 14:01:31 |
| Resolution | Fixed |
| Path | /home/bkline/backups/jira/oceebms/issue.168185 |
A unknown failure during a recent import caused two ebms_article records to be created for one single Pubmed record. It's hard to see just how this happened since there is EBMS software to prevent it, but Bob has pointed out that we should have made the indexes on the Pubmed record IDs and journal IDs unique so that the database would enforce data integrity as a hard backstop for any checks done in application software.
To do that will require that we remove the conflict in the database - which probably requires that we remove all references to one of the records, and then remove the record itself (there are other approaches such as changing the "source" for one of the records, but it may create more problems down the road), and then reconstruct the indexes to require "unique" index entries.
This issue is for the work to do that.
The work on this issue will require the creation of a script that can do the entire job to be given to CBIIT to run. The DEV database has been refreshed from PROD since the error occurred so we will be able to test on DEV - though it is possible that PROD has changes to the records that were made after the refresh.
LOE of 8 story points reflect the need to do this on all four tiers, and with the assistance of CBIIT on two of those tiers.
Done on DEV and QA.
Elapsed: 0:00:00.000394