search

scientist-softserv / britishlibrary

Other
3 stars 0 forks source link

ActiveFedora Error in exporter / can't edit record #374

Open grahamjevon opened 1 year ago

grahamjevon commented 1 year ago

I'm trying to understand why an export contained failures - including a record that can be found in the repository, but cannot be edited. I have outlined the scenario below.

In the nms.iro.bl.uk tenant, I generated an export. Parameters: Metadata Only, All. This generated 4 csv files, but the status was "Complete (with failures)". On page 45 of the export, I found a failure for record id: 5725375b-2a45-4d0a-9458-e8e0093bf15c. There could be more failures, but I am not aware of an easy way to identify them (other than scrolling through each page of the export results), so I will use just this one as an example.

Image

When I click on the record id, I receive the following error:

Image

I successfully accessed this record by adding the id to a base url

Image

I also successfully found this record by searching for the title: "The politics of fashion and the pleasures of youth: Young men and their clothes, 1814-1914".

Image

However, although I can see the public facing record, when I click on Edit, the page apparently does not exist:

Image

Image

ShanaLMoore commented 1 year ago

I believe this sentry error is related:

image

cziaarm commented 1 year ago

Hi @ShanaLMoore ,

That Sentry rntry This looks like a book record that is missing from britishmuseum, I'm not spotting any sentry errors that relate to 5725375b-2a45-4d0a-9458-e8e0093bf15c , though I can confirm the behaviour that Graham has reported. Do think they are related somehow?

cziaarm commented 1 year ago

Perhaps this item is only existing in solr and not in fcrepo.

ShanaLMoore commented 1 year ago

Hi @ShanaLMoore ,

That Sentry rntry This looks like a book record that is missing from britishmuseum, I'm not spotting any sentry errors that relate to 5725375b-2a45-4d0a-9458-e8e0093bf15c , though I can confirm the behaviour that Graham has reported. Do think they are related somehow?

Ah oops! That's my mistake. I must not have been paying close enough attention. Disregard! It's probably not related.

grahamjevon commented 1 year ago

For info, I suspect there may be three instances like this in the NMS repo because the NMS search results returns a total of 3,265 results, but the csv metadata exports only contains 3,262. Is there an easy way to identify what the other 2 missing affected works are?

Also, could this be linked to #369? In #369, though, the count discrepancy is the opposite (i.e. the exporter has 2 more records than the public search).

cziaarm commented 1 year ago

Don't think those are related, will investigate and post on that other issue at a later point.

For this issue, it seems that some some reason the record is in solr but not in fedora.

(don't follow these links they are for devs who have port-forwarding on the go)

solr: http://localhost:8888/solr/#/df04cda6-7f4e-447a-9503-4304d9dc60eb/query?q=id:5725375b-2a45-4d0a-9458-e8e0093bf15c&q.op=OR&indent=true

fcrepo: http://localhost:8889/rest/df04cda6-7f4e-447a-9503-4304d9dc60eb/57/25/37/5b/5725375b-2a45-4d0a-9458-e8e0093bf15c

What to do? Not entirely sure, but have a vague plan...

j-basford commented 1 year ago

Hi @cziaarm you mentioned a plan, we do still see an issue with a discrepancy between export and UI item counts and this ticket seems related? please can you share info on what you think could be done for it? thanks (we understand it could be due to messy metadata but we just need to be able to clean that messy metadata)