[
https://issues.apache.org/jira/browse/HDFS-955?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12852537#action_12852537
]
Suresh Srinivas commented on HDFS-955:
--------------------------------------
# FSImage.java
#* saveNameSpace() - in method comments, instead of "descrepancy between
directory states", some thing like "in order help recovery in case failure
during saveNamespace"
#* typo onle
#* loadFSImage() - why do we need check for directory type, while comparing
latest checkpoint times?
#* loadFSImage() - instead of "after saving some of the images", to "after
saving images in some of the storage directories" to avoid interpretation that
partial image is saved.
#* saveCurrent(), moveCurrent(), moveLastCheckpoint() could all be static.
Better still, they should all be methods in StorageDirectory.
# Storage.java
#* if (!hasCheckpointTmp) will always be true.
#* This might be a good time to document all the the get***Dir() methods
#* analyzestorage() - optional - can you add IOException after throws to fix
javadoc warning
# TestSaveNamespace - should the switch cases for MOVE_CURRENT and
MOVE_LAST_CHECKPOINT be switched?
# TestSaveNamespace - we should add other tests where save image could fail for
first dir also. Also adding failing at different iterations of
moveLastCheckPoint() and moveCurrent()?
> FSImage.saveFSImage can lose edits
> ----------------------------------
>
> Key: HDFS-955
> URL: https://issues.apache.org/jira/browse/HDFS-955
> Project: Hadoop HDFS
> Issue Type: Bug
> Affects Versions: 0.20.1, 0.21.0, 0.22.0
> Reporter: Todd Lipcon
> Assignee: Konstantin Shvachko
> Priority: Blocker
> Attachments: FSStateTransition7.htm, hdfs-955-moretests.txt,
> hdfs-955-unittest.txt, PurgeEditsBeforeImageSave.patch,
> saveNamespace-0.20.patch, saveNamespace-0.21.patch, saveNamespace.patch,
> saveNamespace.patch, saveNamespace.txt
>
>
> This is a continuation of a discussion from HDFS-909. The FSImage.saveFSImage
> function (implementing dfsadmin -saveNamespace) can corrupt the NN storage
> such that all current edits are lost.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
