[
https://issues.apache.org/jira/browse/HDDS-15065?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=18075082#comment-18075082
]
Priyesh K commented on HDDS-15065:
----------------------------------
The automatic snapshot trigger is controlled by
{{{}ozone.scm.ha.ratis.snapshot.threshold{}}}, which defaults to 1000. This
means a snapshot is only auto triggered once the number of pending transactions
exceeds 1000.
As a result, the timer-based mechanism is only effective in scenarios where the
pending transactions are already below this threshold. Given this behavior, it
may be more reasonable to fix the gap to 1 and avoid exposing it as a
configurable parameter.
Otherwise, the changes introduced in this
[PR|https://github.com/apache/ozone/pull/4683] appear to be redundant.
> Reduce Ratis snapshot creation gap to eliminate SCM flush delay and remove
> redundant Ozone configuration
> --------------------------------------------------------------------------------------------------------
>
> Key: HDDS-15065
> URL: https://issues.apache.org/jira/browse/HDDS-15065
> Project: Apache Ozone
> Issue Type: Bug
> Components: SCM
> Reporter: Priyesh K
> Assignee: Priyesh K
> Priority: Major
> Labels: pull-request-available
>
> This is a follow-up to HDDS-8485.
> The issue of delayed delete block processing in SCM still persists due to
> internal behavior in Apache Ratis.
> Although a timer-based flush mechanism was introduced, its effectiveness is
> limited because Ratis enforces a *minimum transaction gap (1024)* before
> triggering certain operations such as snapshot-related checks.
> *Code Reference: [RaftServerImpl (lines
> 1317–1320)|https://github.com/apache/ratis/blob/f141e10551a1bef69c0c7387dd184bc88e821b6d/ratis-server/src/main/java/org/apache/ratis/server/impl/RaftServerImpl.java#L1317-L1320]*
> Because of this constraint, flush operations remain indirectly gated by the
> transaction gap, negating the intended benefit of the timer-based approach.
> *Proposed Fix*
> * Reduce the value of {{ozone.scm.ha.ratis.server.snapshot.creation.gap}}
> from *1024 to 1*
> * Remove this configuration from Ozone to avoid duplication and rely on
> appropriate defaults within Ratis
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
