You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@flink.apache.org by "ASF GitHub Bot (Jira)" <ji...@apache.org> on 2025/03/04 13:04:00 UTC
[jira] [Updated] (FLINK-37419) ForStMapState serializeValue may get wrong result
[ https://issues.apache.org/jira/browse/FLINK-37419?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
ASF GitHub Bot updated FLINK-37419:
-----------------------------------
Labels: pull-request-available (was: )
> ForStMapState serializeValue may get wrong result
> -------------------------------------------------
>
> Key: FLINK-37419
> URL: https://issues.apache.org/jira/browse/FLINK-37419
> Project: Flink
> Issue Type: Bug
> Components: Runtime / State Backends
> Affects Versions: 2.0.0
> Reporter: Yue Ma
> Priority: Major
> Labels: pull-request-available
> Fix For: 2.0.0
>
>
> When ForStMapState write some thing , the user value is directly serialized and written to the DB. But this may be problematic for two main reasons:
> * In {*}ForStSyncMapState{*}, _serializValueNullSensitive_ is called when executing put, which adds an extra boolean to determine if the value is null. This is inconsistent with the behavior of *ForStMapState* and may result in inconsistent errors when switching async.
> * In the scenario of using ttl, *ForStMapState* will use
> {code:java}
> config=FlinkCompactionFilter Config.createForMap(ttl, queryTimeAfterNumEntries);
> {code}
> This config is used to configure the compactFilter, which also means that the compactFilter timestampOffset for ForStMapState is 1. If ForStMapState is written in the current way without a boolean, some errors may occur during data cleaning by the compactFilter
--
This message was sent by Atlassian Jira
(v8.20.10#820010)