Kevan Jahanshahi created UNOMI-776:
--------------------------------------
Summary: Avoid the PersistenceService auto scroll when doing
queries
Key: UNOMI-776
URL: https://issues.apache.org/jira/browse/UNOMI-776
Project: Apache Unomi
Issue Type: Bug
Affects Versions: unomi-1.8.0, unomi-2.2.0
Reporter: Kevan Jahanshahi
recently discovered in: UNOMI-764
The persistence service is doing an automatic scroll queries when doing queries
with limit {*}-1{*}:
{code:java}
persistenceService.query(propertyCondition, null, Profile.class); {code}
Nothing in the Java Doc is mentioning this behavior.
And it's dangerous because if this query is done a lot of time it will create a
lot of scroll id in RAM and kill the memory.
ch scroll query creates a cursor on the Elasticsearch server, which consumes
system resources, including memory, CPU, and disk I/O.If you have a high number
of concurrent scroll queries running simultaneously, it can lead to increased
resource utilization and potentially cause performance issues or even
out-of-memory errors if the system is not adequately configured to handle the
load.
This is especially true when dealing with large indices with a high volume of
data. (lot of profiles, lot of sessions in case of the merge)
*To me we should remove this auto-scroll done in the persistence service, and
create a dedicated method in the persistence service to let the implementation
choose between scrolling or not.*
*TODO:*
* *remove auto scroll*
* *create a new dedicated method in the service to do scroll queries.*
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
