CDR Tickets
| Issue Number | 4735 |
|---|---|
| Summary | Find out how summary rewrites are handled |
| Created | 2019-11-23 14:43:20 |
| Issue Type | Inquiry |
| Submitted By | Kline, Bob (NIH/NCI) [C] |
| Assigned To | Kline, Bob (NIH/NCI) [C] |
| Status | Closed |
| Resolved | 2019-11-25 21:17:58 |
| Resolution | Fixed |
| Path | /home/bkline/backups/jira/ocecdr/issue.252781 |
I was trying to run the weekly publishing test this morning on QA2, and while the export went smoothly, the push job blew up. The cause turned out to be that the job was trying to remove CDR62770 (the English HP summary on breast cancer treatment during pregnancy) which failed because CDR256749 (the Spanish translation of that summary) would have been left without its English original, and Drupal doesn't allow that to happen. I had (perhaps mistakenly) been under the impression that when a summary gets an overhaul which requires the rewrite to take place in a separate CDR document, the new content from that document is transplanted to the document for the original summary which is being rewritten (using the Replace Old Document With New One tool from the CDR Admin menus), so that the original summary is able to continue life under its original CDR ID. However, at least in this case, that's not how things were done. If we want to change how the summary rewrites are published (or if my picture of how things used to work was wrong in the first place, and it's always been done the way it's happening in this case), we may need to establish a careful process for the sequence of events. When a rewrite of an English summary is going to be implemented by blocking the old document and publishing the new one, and there's a Spanish translation on the web site for that summary, I believe we will need to perform a sequence of hot fixes, first removing the Spanish summary, then removing the old English original summary, then publishing the new English version, and then restoring the Spanish translation of the summary. It might get tricky (perhaps very tricky) if there are things which depend on (link to) the Spanish summary, but I imagine we'll be able to come up with some creative solutions.
Added some watchers.
~oseipokuw and ~juther - which of you would be the most appropriate for discussing this issue with?
~bkline You can discuss this with me. However, I should say that you're seeing this problem because this is QA data, and because we use QA for training and testing, we have to expect some inconsistencies although it should be minimal (This was partly my argument on Thursday for QA refresh :-)). Looking at the comments, it appears we used this summary to test the Replace function albeit incorrectly. Your description of the replace doc process is correct and that is exactly how we use it. That is, we always retain the original summary and the new/temp summary is blocked. (We recently discussed deleting the blocked summaries from PROD since we don't use them for anything.)
Ah, that's encouraging. So the publishing job failure was the result of a mistake in the usage of that tool. I think I understand the benefits of fresher data (and the risks from refreshing at this point in the game). At least in this case, it's hard to see how having more up-to-date data would have made any difference for this mistake. I have modified the instructions on the tool's form page in the hopes of making this mistake less likely to occur (let me know if you agree).
At any rate, ~volker, I think you should be able to resume your testing of publishing, after first fixing the problem caused by the incorrect use of the tool (that is, unblocking CDR62770). You might want to do a hotfix of just that summary separately first, though if I had to bet, I would think that wouldn't be necessary. But unblocking that summary is necessary. 🙂
The instructions looks good. There seems to be a minor typo in the second sentence.
Good catch. Fixed.
William confirmed that my understanding of how major summary rewrites are supposed to be handled is correct, and that the use of the Replace tool which broke publishing was a mistake.
| File Name | Posted | User |
|---|---|---|
| ReplaceDoc Instructions.PNG | 2019-11-25 20:03:07 | Osei-Poku, William (NIH/NCI) [C] |
Elapsed: 0:00:00.002425