I want to ask about rollup job. Currently, i've been created a rollup job with 1s interval and 5 fields on terms and 2 fields for metrics and 60.000 page size. And the total document for the production indexes which is the index that i want to rollup is around 100+ million in 1 day.
I define the pattern in my rollup job like this:
Which is it will match two kind of index from 18th till today:
These indexes have total of documents around 100 million. So we can say on each day, there will be 100+ million of documents from those 2 kind of indexes
The problem is:
When the job was ran for indexing documents, it's like it didn't finish his job at initiate run while other job with the same spec except terms(2 fields), it keeps running until it catch up today's data. But This broken job, only rolling up a thousand of documents and then it stop indexing documents and back to started mode
Do you think it's because the amount of documents in production index were too much? and the terms and interval are too complex? (Interval: 1s, terms: 5 fields)
Or is there any reason to explain this behaviour?
Also, i can't extend the page size from 60k to 70 or bigger than that. Because when i test it in my non prod environment, when the job has page size bigger than 60k, the job will only triggered to indexing but it will not processing any document.
If i assume this behaviour happen Because the page size is not enough for the data, what should i do? As my previous statement, i can't change it bigger than 60.000
is there any reason to explain this behaviour?
Please, i need your experience and knowledge