[
https://issues.apache.org/jira/browse/IOTDB-4688?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Song Ziyang reassigned IOTDB-4688:
----------------------------------
Assignee: Song Ziyang
> Streaming MD5 replace blocking MD5
> ----------------------------------
>
> Key: IOTDB-4688
> URL: https://issues.apache.org/jira/browse/IOTDB-4688
> Project: Apache IoTDB
> Issue Type: Sub-task
> Reporter: Song Ziyang
> Assignee: Song Ziyang
> Priority: Major
>
> Currently in InstallSnapshot, statemachine has to pre-compute each file's md5
> digest and provide it in {{{}FileInfo{}}}. When snapshot file is large, this
> computing is time-consuming and may cause leader timeout.
> In this PR I propose to move the computing task along within the
> InstallSnapshot stream process. That is, for each SnapshotChunk, we can read
> the chunk & compute its moving MD5 digest in the same time. This can enable
> us to read file & compute file MD5 in one sequential scan and thus save IO
> costs.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
