-
Notifications
You must be signed in to change notification settings - Fork 508
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[DOC] New system property: opensearch.xcontent.fast_double_writer #4513
Comments
Hi @mgodwan - Is this being introduced in version 2.9? |
Yes, The changes have been merged already: opensearch-project/OpenSearch#8467 (comment) |
@mgodwan Where would the new documentation be published on the doc site? Should it be added to https://opensearch.org/docs/latest/install-and-configure/install-opensearch/index/#important-system-properties? Also, will you link me to the feature brief and dev docs? |
First, the issue description is incorrect: the system setting is
@vagimeli There is no feature brief to my knowledge. This is a pretty low-level performance improvement and a system setting was added to alleviate concerns around potential incompatibilities. I think it is logical to group it with the existing
By default, OpenSearch uses the floating point parsing capability built in to the Java Runtime Environment. Set this value to @mgodwan A couple questions for you:
|
Thanks @andrross for pointing to the correct name of the property.
@vagimeli Yes, this seems to the correct place since we are already adding other system properties related to xcontent in the same place.
@andrross Yes, we decided to make it an opt-in property to reduce the blast radius. This property only deals with floating point serialization (and not parsing which is always enabled as we didn't observe differences in the parsing of floating point numbers compare to java code)
Possible description we can add for this can be: The property can be set to determine whether to use standard Java code to write floats/doubles (default) or use the Schubfach algorithm which is faster. The latter approach may lead to small differences in the precision of the float/double that is written to the JSON output. |
What do you want to do?
Tell us about your request.
Provide the documentation for recently introduced new system property
opensearch.xcontent.fast_double_writer
that allows to configure using fast double serialization features which provides better performance for serialization of floating point numbers.What other resources are available? Provide links to related issues, POCs, steps for testing, etc.
Covered in issue: opensearch-project/OpenSearch#7822
The text was updated successfully, but these errors were encountered: