n?
Cheers,
Subroto Sanyal
Datanode didn't fail….
DFS Remaining was 0%
Cheers,
Subroto Sanyal
On Jun 20, 2012, at 12:16 PM, qhdpp wrote:
> datanode
> fail or
> dfs
> full
>
> Sent from Meizu MX
>
>
> 原始邮件
> 发件人:Subroto
> 时间:2012-6-20 15:07
> 收件人:hdfs-user
indicate the same…..
Just like to know where the point:
> 5*BLOCK_SIZE
is documented…. :-(
Thanks again...
Cheers,
Subroto Sanyal
On Jun 19, 2012, at 6:22 PM, Todd Lipcon wrote:
> Hi Subroto,
>
> Anything in the DN logs? Maybe the one DN is out of disk space? You
> need to have 5*
…..
The hadoop version being used is: 2.0.0
Cheers,
Subroto Sanyal
,
Subroto Sanyal
On May 31, 2012, at 8:29 PM, Todd Lipcon wrote:
> Hi Subroto,
>
> You'll need to implement the "getChannel()" method in such a way that
> it returns a SocketChannel implementation -- and then implement the
> appropriate methods in that class to read/
I missed to mention that my custom socket factory returns a Custom Socket as
well. :-(
Cheers,
Subroto Sanyal
On May 29, 2012, at 2:46 PM, Subroto wrote:
> Hi,
>
> As part of the fix HDFS-2129 it is mentioned:
> " If you use a custom socket factory, it must generate
here is an option to use legacy BlockReader
(dfs.client.use.legacy.blockreader) but, I am planning to upgrade my custom
socket factory rather than failing back to the old approach.
Request some guidance in this direction….
Cheers,
Subroto Sanyal
Thanks Harsh ……
Cheers
Subroto Sanyal
On May 18, 2012, at 1:52 PM, Harsh J wrote:
> Yes this is intentional, and an incompatible change, and was done via
> https://issues.apache.org/jira/browse/HADOOP-6201 to have better API
> behavior.
>
> On Fri, May 18, 2012 at 2:04 PM,
java:1456)
at org.apache.hadoop.fs.FileSystem.globStatus(FileSystem.java:1411)
at org.apache.hadoop.fs.FileSystem.globStatus(FileSystem.java:1389)
Is this an intentional change made or shall we need to file it as bug and fix
it…. ?
Cheers,
Subroto Sanyal