Re: HDFS open file limit

2014-01-27 Thread sudhakara st
There is no open file limitation for HDFS. The 'Too many open file'  limit
is for OS file system. Increase *system-wide maximum number of open files,
Per-User/Group/Process file descriptor limits.*


On Mon, Jan 27, 2014 at 1:52 AM, Bertrand Dechoux decho...@gmail.comwrote:

 At least for each machine, there is the *ulimit *that need to be verified.

 Regards

 Bertrand

 Bertrand Dechoux


 On Sun, Jan 26, 2014 at 6:32 PM, John Lilley john.lil...@redpoint.netwrote:

  I have an application that wants to open a large set of files in HDFS
 simultaneously.  Are there hard or practical limits to what can be opened
 at once by a single process?  By the entire cluster in aggregate?

 Thanks

 John









-- 

Regards,
...Sudhakara.st


Re: HDFS open file limit

2014-01-27 Thread Harsh J
Hi John,

There is a concurrent connections limit on the DNs that's set to a default
of 4k max parallel threaded connections for reading or writing blocks. This
is also expandable via configuration but usually the default value suffices
even for pretty large operations given the replicas help spread read load
around.

Beyond this you will mostly just run into configurable OS limitations.
On Jan 26, 2014 11:03 PM, John Lilley john.lil...@redpoint.net wrote:

  I have an application that wants to open a large set of files in HDFS
 simultaneously.  Are there hard or practical limits to what can be opened
 at once by a single process?  By the entire cluster in aggregate?

 Thanks

 John







RE: HDFS open file limit

2014-01-27 Thread John Lilley
What exception would I expect to get if this limit was exceeded?
john

From: Harsh J [mailto:ha...@cloudera.com]
Sent: Monday, January 27, 2014 8:12 AM
To: user@hadoop.apache.org
Subject: Re: HDFS open file limit


Hi John,

There is a concurrent connections limit on the DNs that's set to a default of 
4k max parallel threaded connections for reading or writing blocks. This is 
also expandable via configuration but usually the default value suffices even 
for pretty large operations given the replicas help spread read load around.

Beyond this you will mostly just run into configurable OS limitations.
On Jan 26, 2014 11:03 PM, John Lilley 
john.lil...@redpoint.netmailto:john.lil...@redpoint.net wrote:
I have an application that wants to open a large set of files in HDFS 
simultaneously.  Are there hard or practical limits to what can be opened at 
once by a single process?  By the entire cluster in aggregate?
Thanks
John