Closed FuhuXia closed 4 days ago
It is ok to crash the process, as long as the app stays up.
Crashing fetch-consumer process on ParentNotHarvestedException is related to how child datasets are processed before parent dataset is created. Rather than changing the behaviour, we can simply restart the process right after crash so that catalog-fetch app stays up, asdone in the PR above.
Deployed on prod. Rerun some datajson sources that were previously turn off to manual. Still sees ParentNotHarvestedException, but no more app crashes.
When the parent dataset is deleted from a datajson source, each child dataset makes catalog-fetch process crash. This slows down harvest process and often make other harvest jobs stuck for days.
How to reproduce
https://github.com/GSA/data.gov/issues/4755#issuecomment-2122747761 https://github.com/GSA/data.gov/issues/4772#issuecomment-2147608839
Expected behavior
logs an error and continues to the item in the fetch queue
Actual behavior
crashes catalog-fetch
Sketch
[Notes or a checklist reflecting our understanding of the selected approach]