[
https://issues.apache.org/jira/browse/FLUME-1186?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13270521#comment-13270521
]
Brock Noland commented on FLUME-1186:
-------------------------------------
I was thinking we'd implement it as follows but I am very open to other ideas:
1) Have a digester open for each file and all bytes written to this file are
added to the hash.
2) When the file is closed we write out a side .md5 file.
3) When starting/replaying we'd calculate the hash of a log file if a .md5 file
exists. If the file does not exist we'd just replay.
I think you should be able to turn off all the code above via configuration and
have no checksumming.
> FileChannel logs should be checksumed
> -------------------------------------
>
> Key: FLUME-1186
> URL: https://issues.apache.org/jira/browse/FLUME-1186
> Project: Flume
> Issue Type: Improvement
> Components: Channel
> Affects Versions: v1.2.0
> Reporter: Brock Noland
> Priority: Minor
>
> While writing a FileChannel WAL we should be calculating the hash of the data
> and then when closed that hash should be written out. Perhaps a side file.
> This can be verified when replaying.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira