For the import of decision/process XMLs, we go over the process definition index in Elasticsearch and fetch all definitions that don't have an XML yet. For those, we fetch the XMLs from the engine and then add an import job. As Optimize goes over the index regularly there's no point to use scrolling here.
- the process and decision definition XML import does not use Elasticsearch scrolling any longer
There are (possibly) two options on how we could improve that:
- We could just use simply perform a query over the definitions that don't have an XML and sorted by the date those definitions were imported to Elasticsearch. This way we could page over the result.
- the process definition XML import iterates over the engine definitions sorted by deployment time and fetches the XML for each definition.