[
https://issues.apache.org/jira/browse/HDFS-397?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12730752#action_12730752
]
Luca Telloli commented on HDFS-397:
-----------------------------------
Raghu, for the latest question I suggest to download
https://issues.apache.org/jira/secure/attachment/12413314/HDFS-311.patch
and have a look for instance to:
line 10: calling new EditLogFileInputStream on a StorageDirectory
line 345 and below: where constructors for new EditLogOutputStream are
instantiated and every time a newInstance() is called
About BackupStorage, yes, there might be a problem there. I have to go a bit in
deep with the Backup* architecture and might require some time. At the same
time I believe a couple of things:
- I agree that unit tests are not sufficient, but in general I think they
should fail on non-standard non-planned cases. If, given a wrong directory,
this test doesn't fail then I assume the test is broken or needs some fixing
- Since this patch deals with new EditLogFileOutputStream instantiated with a
directory, I'm guessing a change to support such a jspool dir can be easily
implemented. But at the same time, in the design document for HADOOP-4539 I
read "Journal Spool is an analog of 'edits.new'" which tells me a standard
storage directory could be used. What's your take on this?
> Incorporate storage directories into EditLogFileInput/Output streams
> --------------------------------------------------------------------
>
> Key: HDFS-397
> URL: https://issues.apache.org/jira/browse/HDFS-397
> Project: Hadoop HDFS
> Issue Type: Sub-task
> Reporter: Luca Telloli
> Assignee: Luca Telloli
> Attachments: HADOOP-6001.patch, HADOOP-6001.patch, HADOOP-6001.patch,
> HDFS-397.patch
>
>
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
