[
https://issues.apache.org/jira/browse/SPARK-51252?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Jungtaek Lim resolved SPARK-51252.
----------------------------------
Fix Version/s: 4.1.0
Resolution: Fixed
Issue resolved by pull request 50030
[https://github.com/apache/spark/pull/50030]
> Adding state store level metrics for last uploaded snapshot version in HDFS
> State Stores
> ----------------------------------------------------------------------------------------
>
> Key: SPARK-51252
> URL: https://issues.apache.org/jira/browse/SPARK-51252
> Project: Spark
> Issue Type: Improvement
> Components: Structured Streaming
> Affects Versions: 4.0.0, 4.1.0
> Reporter: Zeyu Chen
> Assignee: Zeyu Chen
> Priority: Minor
> Labels: pull-request-available
> Fix For: 4.1.0
>
>
> Similarly to SPARK-51097, we would also like to introduce a similar level of
> observability to HDFSBackedStateStore.
> The introduction of state store "instance" metrics to StreamingQueryProgress
> to track the latest snapshot version uploaded in HDFS state stores should
> address three challenges in observability:
> * Uneven partition starvation, where we need to identify partitions with
> slow state maintenance,
> * Finding missing snapshots across versions, so we minimize extensive
> replays during recovery,
> * Identify performance instability, such as gaining insights into snapshot
> upload patterns
> The instance metrics should be kept as generalized as possible, so that
> future instance metrics for observability can be added with minimal
> refactoring.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
