[
https://issues.apache.org/jira/browse/HDFS-2136?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13061432#comment-13061432
]
Todd Lipcon commented on HDFS-2136:
-----------------------------------
Yea, that would be great. I think there are three possibilities:
1) use the AOP-based fault injection framework
2) use some static methods (see use of ErrorSimulator in the SecondaryNameNode
and TransferFsImage classes)
3) refactor the code so that a mock could be injected
Personally, my preference is #2. I've found the AOP-based fault injection tests
(#1) very difficult to develop and understand, as someone who only looks at
aspectj code once every few months. They're also easy to forget about since
they're in a different tree, they're slow to run, etc. #3 is probably ideal but
I can't think of any refactors that wouldn't be really contrived. Thoughts?
> 1073: Fault injection for StorageDirectory failures during read/write of
> FSImage/Edits files
> --------------------------------------------------------------------------------------------
>
> Key: HDFS-2136
> URL: https://issues.apache.org/jira/browse/HDFS-2136
> Project: Hadoop HDFS
> Issue Type: Sub-task
> Reporter: Matt Foley
>
> Both HDFS-1955 and HDFS-2135 have observed that it is difficult to unit test
> such failures. As a result, regression of HDFS-1955 was only found by
> careful manual review (thanks, atm!). Since 1073 is making broad changes to
> the way these files are read and written, and appropriately putting effort
> into correct error handling, I propose we make also make it possible to
> auto-test that error handling.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
