[
https://issues.apache.org/jira/browse/HDDS-12008?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17909789#comment-17909789
]
Tsz-wo Sze commented on HDDS-12008:
-----------------------------------
bq. Any possible ideas regarding the truncation logic of a "linked" LogEntry?
Do we need to delete the files, etc?
Checked the code. The async case also won't delete the block file after
truncation, i.e.
- WriteChunk at index 200: a local block file F is created.
- truncate to index 190: the file F will not be deleted (the file will be
deleted by the block scanner later).
For the link(..), we should change it not to do putBlock (HDDS-12007). Then,
we don't have to delete the local file during truncation.
> Handle ContainerStateMachine#truncate for Streaming Write Pipeline
> ------------------------------------------------------------------
>
> Key: HDDS-12008
> URL: https://issues.apache.org/jira/browse/HDDS-12008
> Project: Apache Ozone
> Issue Type: Sub-task
> Reporter: Ivan Andika
> Priority: Major
>
> StateMachine#link is called in WriteLog. Meaning that there might be a chance
> that the log entry will be truncated (e.g. INCONSISTENCY, etc). We need to
> see whether we should handle the ContainerStateMachine#truncate logic for
> Streaming Write Pipeline.
>
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]