Hi James,

We had a very similar block of records in our OAI-PMH interface.

ExLibris/Primo were very quick to say it was a problem with our repository.

It took me a while to convince them that setting a slightly longer timeout on a harvest request would make it work again.

In our case, I could demonstrate that the repository reliably took ~ 65 seconds to generate the response. At the time, the timeout was 60 seconds.

To resolve this issue:

• There might be an option in Primo to set a longer timeout for the harvest: https://knowledge.exlibrisgroup.com/Primo/Knowledge_Articles/How_can_I_configure_OAI_harvesting_for_Primo_Publishing_Platform_Pipes%3F. This would be my first option.
• A custom Primo metadata profile is something we have used in the past to add ORCIDs to the data Primo harvests. Give it another name, and configure Primo. This may be worth considering to provide additional data to Primo*.
• Your ‘other’ idea would be OK until something (e.g. a system update) makes all those records appear to have changed at the same time. Then you’d have to run through the review/live process all over again.

* There may be other things you can 'fix' with a custom profile e.g. how Primo understands/displays open, restricted, requestable, oa-location or no-fulltext records.

Cheers,

John

Hi everyone,

Our harvest from EPrints to Primo via the OAI harvester keeps failing. We think we have pinned down the issue to a page under one of the resumption tokens that contains ~17 physics records that each have ~3000 authors. We think this causes the resumption token url to take too long to load and breaks the harvest.

As a potential solution, we have decided to try creating a new export plugin that limits the number of authors being sent to the harvester. Ideally only our Primo instance would use this plugin.

Does this sound like a reasonable approach? If anyone has experienced something similar, would you be willing to share your experience or solutions?

I have had another idea where we move the problem records into review, get a successful harvest, then gradually move them back into the live archive and let the harvester grab them since they would have a more recent last modified date. This is less of a solution and more of a way through it for now.

Thanks,

James