[
https://issues.apache.org/jira/browse/HDFS-1791?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13012200#comment-13012200
]
Suresh Srinivas commented on HDFS-1791:
---------------------------------------
# DFSAdmin.java:
#* Use same formatting with \t for deleteBlockPool description
#* Should we mention in the usage, for deleteBlockPool to work, the namenode
service must not be running in datanode and add pointer to refreshNamenodes?
# FSDatasetInterface.java - document coudl be clearer on distinction between
removeBlockPool() Vs deleteBlockPool(). Should we call it shutdownBlockPool()
instead of removeBlockPool()? Also talking about rbw, tmp etc may not be the
right thing in FSDatasetInterface.java.
# FSDataset.java
#* Move FSDataset#deleteBPDirectories() and FSDataset#isBPDirEmpry() to
FSVolume.
#* FSDataset#deleteBlockPool() - move force flag check to outside the for loop
where you are checking for isBPDirEmpty()
#* FSVolume#removeBlockPool() should access map directly to get BlockPoolSlice
and not throw IOException. Similarly methods calling this up the chain should
not throw IOException.
# Datanode.java
#* Instead of ignoring the exception thrown by removeBlockPool(), please log
it, using LOG.warn
# ReplicasMap#cleanupBlockPool() - is checkBlockPool() needed? With the
exception thrown from this, FSDataset#removeBlockPool() many not call
volumes.removeBlockPool()
# TestDeleteBlockPool
#* Class javadoc shoudl mention it tests both refreshNamenodes and
deleteBlockPool functionality
#* Remove first test for "Must not delete a running blockpool", as it is done
twice. Delete this block for dn2 also, as this test has already been done with
dn1.
#* Instead of comments in catch blocks such as expected and ignored, you could
also use catch (IOException expected) { } or catch (IOException ignored) { }
> Hdfs Federation: Add command to delete block pool directories from a datanode.
> ------------------------------------------------------------------------------
>
> Key: HDFS-1791
> URL: https://issues.apache.org/jira/browse/HDFS-1791
> Project: Hadoop HDFS
> Issue Type: Sub-task
> Components: name-node
> Affects Versions: Federation Branch
> Reporter: Jitendra Nath Pandey
> Assignee: Jitendra Nath Pandey
> Fix For: Federation Branch
>
> Attachments: HDFS-1791.1.patch
>
>
> A command to delete block pool directories from the datanode is required.
> Once a datanode stops serving a block pool, this command can be used to
> delete the block pool directories. A force option will delete the block pool
> directory on the datanode along with all its contents. If force is not
> specified directories will be deleted only if no block files exist for that
> block pool.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
