[
https://issues.apache.org/jira/browse/HDFS-1751?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13008157#comment-13008157
]
Daryn Sharp commented on HDFS-1751:
-----------------------------------
(I'll avoid the design dispute, and explain why the implementation is the way
it is)
The main quota method, updateCount(), is too low in the call stack and is
designed to handle disk and inode changes. Allowing updateCount() to perform
the limit checks will cause issues because too many other operations call it,
and updateCount() can't discern why it's been invoked. A few examples of
issues that would occur:
1) adding or removing a block from a file will fail if the directory item limit
has been reached
2) changing the replication factor on a pre-existing file that exceeds either
of the limits will fail
3) updating the disk quota counts via updateSpaceConsumed() will fail if either
of the limits are reached
...etc...
To address these issues, at a minimum, a boolean will need to be passed to
updateCount() to indicate if a filesystem directory update is occurring (ie.
only addChild() will pass true). The new INode will also need to be passed to
updateCount to check the component length. This would be a more complex
change, that places an undue burden on the callers of updateCount() to pass the
right args, just to avoid having addChild perform the fs limit checks.
Please let me know if I'm overlooking anything in my analysis.
> Intrinsic limits for HDFS files, directories
> --------------------------------------------
>
> Key: HDFS-1751
> URL: https://issues.apache.org/jira/browse/HDFS-1751
> Project: Hadoop HDFS
> Issue Type: New Feature
> Components: data-node
> Affects Versions: 0.22.0
> Reporter: Daryn Sharp
> Assignee: Daryn Sharp
> Fix For: 0.23.0
>
> Attachments: HDFS-1751-2.patch, HDFS-1751-3.patch, HDFS-1751-4.patch,
> HDFS-1751.patch
>
>
> Enforce a configurable limit on:
> the length of a path component
> the number of names in a directory
> The intention is to prevent a too-long name or a too-full directory. This is
> not about RPC buffers, the length of command lines, etc. There may be good
> reasons for those kinds of limits, but that is not the intended scope of this
> feature. Consequently, a reasonable implementation might be to extend the
> existing quota checker so that it faults the creation of a name that violates
> the limits. This strategy of faulting new creation evades the problem of
> existing names or directories that violate the limits.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
