ON AMI ingest we presort the ADOs to ensure Top level/parents get ingested first. But we don't traverse deep enough when sorting (specially when using UUIDs for the semantic relationships) leading to the queue having children first. We also have a safety feature for that (a retry) but on very heavily nested structures, the number of retries is not enough
The task is to make the Sorting by parent algorithm faster/better and maybe just as failsafe add an extra retry
DiegoPino
commented
4 weeks ago
What?
ON AMI ingest we presort the ADOs to ensure Top level/parents get ingested first. But we don't traverse deep enough when sorting (specially when using UUIDs for the semantic relationships) leading to the queue having children first. We also have a safety feature for that (a retry) but on very heavily nested structures, the number of retries is not enough
The task is to make the Sorting by parent algorithm faster/better and maybe just as failsafe add an extra retry
@alliomeria as observed with scrapbooks