Closed mjgiarlo closed 1 year ago
The DeleteService has methods for deleting files from stacks and purl: https://github.com/sul-dlss/dor-services-app/blob/main/app/services/delete_service.rb#L38-L46
And the UnpublishService: https://github.com/sul-dlss/dor-services-app/blob/main/app/services/unpublish_service.rb
Is that good enough?
@andrewjbtw @sul-dlss/infrastructure-team ☝🏻
Decision from standup on 8/30: @andrewjbtw to write up instructions on how to do this using bulk actions.
@andrewjbtw Given :point_up:, is it OK if I assign this to you (and toss at In Progress)?
Putting the instructions in comments. The goal is to make a good faith effort to pull back SDR content from Access systems. The instructions are based on the takedown instructions in Consul, but applied to all of the items in stage/QA.
Note that we are not concerned with Exhibits in sdr-stage/QA.
First step: unrelease SDR objects from SearchWorks
Click on the facet for all items currently released to SW
Click on the "Bulk Actions" button just above the search results
From the Bulk actions page, click on "New Bulk Action"
Manage release should be the already selected bulk action
Choose the radio button for "Do not release this object"
Select "SearchWorks" as the release to option (should be already selected)
Click on "Populate with previous search"
Wait for the druid box to be populated. It could take a while for a long list.
Submit the bulk action
This will update all of the Purls with a tag that says they should be unreleased. This should remove them from FOLIO records (if they have MARC) and from SW. The complete process can take an hour or so. It may not be 100% successful in stage because of data issues but at least it's a good faith effort to clean up.
Next step: make all the druids dark
When druids are made dark:
A couple of notes specific to the stage/QA reset:
Steps to open the items:
This will take quite a while for a large number of druids.
Steps to open the collections - do the same as above but select collections instead of items in the Object Type facet.
Once the druids have been opened, you can change the rights.
When the bulk action completes, close the objects.
Accessioning will run on all of the druids. The accessionWF "shelve" step will delete the files from Stacks and "publish" will delete the Purls.
There may be a more efficient way to do all of the above programmatically but those are the steps I would follow to remove content from access systems using Argo.
With the new process for unreleasing using FOLIO APIs via a single-threaded queue, I think we should expect that step to take longer. An earlier test had 100 druids taking about 9 mins. So for 7,273 druids that could take ~11 hours.
I ran unrelease on the 7273 druids on stage last week to see if anyone would notice. So far no one has. It took about 1 hour.
I did notice that about 1500 druids in the list aren't actually released and only show that way because of problems in the indexing logic.
That's good to hear! I'm wondering if unrelease is generally faster because of the checks involved? Thanks for testing.
Part of the QA/stage reset process involves yanking SDR items from Access systems, and our mechanism for this involves unreleasing items (from SearchWorks) and assigning them dark access rights (effectively removing them from stacks).
There are bulk actions to support this in Argo, but we wonder if there may be a more efficient/automatable way to do this (possibly directly in DSA). Figuring that out, or deciding to punt on it and use Argo, is the point of this issue.