OpenSearch/client
Costin Leau f9c15d0fec EQL: Introduce sequencing fetch size (#59063)
The current internal sequence algorithm relies on fetching multiple results and then paginating through the dataset. Depending on the dataset and memory, setting a larger page size can yield better performance at the expense of memory.
This PR makes this behavior explicit by decoupling the fetch size from size, the maximum number of results desired.
As such, use in testing a minimum fetch size which exposed a number of bugs:

Jumping across data across queries causing valid data to be seen as a gap.
Incorrectly resuming searching across pages (again causing data to be discarded).
which have been addressed.

(cherry picked from commit 2f389a7724790d7b0bda67264d6eafcfa8b2116e)
2020-07-06 19:14:26 +03:00
..
benchmark Replace compile configuration usage with api (7.x backport) (#58721) 2020-06-30 15:57:41 +02:00
client-benchmark-noop-api-plugin Deduplicate Empty InternalAggregations (#58386) (#59032) 2020-07-04 14:02:16 +02:00
rest Replace compile configuration usage with api (7.x backport) (#58721) 2020-06-30 15:57:41 +02:00
rest-high-level EQL: Introduce sequencing fetch size (#59063) 2020-07-06 19:14:26 +03:00
sniffer Replace compile configuration usage with api (7.x backport) (#58721) 2020-06-30 15:57:41 +02:00
test Replace compile configuration usage with api (7.x backport) (#58721) 2020-06-30 15:57:41 +02:00
transport Replace compile configuration usage with api (7.x backport) (#58721) 2020-06-30 15:57:41 +02:00