Page 1 of 1
how to beyond the filelimit in Unix using sequence file?
Posted: Tue Sep 18, 2007 4:12 am
by donhoff
Hi, I developed a parallel job which extracted a huge amount of data and wrote them into a sequence file. But the file was so huge that it was beyond the unix file limit(2G).
So I want to write data in a list of files, such as filename_01,filename_02,.... The number of files depends upon the data.
Could anyone tell me how to realize this function in DataStage EE?
Thanks!
Posted: Tue Sep 18, 2007 8:18 am
by Ragunathan Gunasekaran
You can use fileset stage instead of sequential file stage . The fileset stage has the option of overcoming the file size limit.
Posted: Tue Sep 18, 2007 8:41 am
by donhoff
Yes, I tried to use fileset. But Fileset produced only 6 files for I have 6 partitions. It could not produce more files and so the job failed. All 6 files reached the 2G limit but there are still data not written to file.
Is there any other solution? For example , some way that can produce 2G data files according to the actual data volume?
Thanks!
Re: how to beyond the filelimit in Unix using sequence file?
Posted: Tue Sep 18, 2007 8:55 am
by Raghumreddy
Hi
Contact your Unix admin on the file size . they can increase the limit on the file system that you are using.
HTH
Raghu M
donhoff wrote:Hi, I developed a parallel job which extracted a huge amount of data and wrote them into a sequence file. But the file was so huge that it was beyond the unix file limit(2G).
So I want to write data in a list of files, such as filename_01,filename_02,.... The number of files depends upon the data.
Could anyone tell me how to realize this function in DataStage EE?
Thanks!
Re: how to beyond the filelimit in Unix using sequence file?
Posted: Tue Sep 18, 2007 8:58 am
by donhoff
Raghumreddy wrote:Hi
Contact your Unix admin on the file size . they can increase the limit on the file system that you are using.
HTH
Raghu M
donhoff wrote:Hi, I developed a parallel job which extracted a huge amount of data and wrote them into a sequence file. But the file was so huge that it was beyond the unix file limit(2G).
So I want to write data in a list of files, such as filename_01,filename_02,.... The number of files depends upon the data.
Could anyone tell me how to realize this function in DataStage EE?
Thanks!
OK, Thanks! This is a good suggestion. Do you know what is the max file size for a HP9000 Unix Server and a IBM AIX5.3 p570 Server?
Thanks!
Re: how to beyond the filelimit in Unix using sequence file?
Posted: Tue Sep 18, 2007 9:02 am
by Raghumreddy
Default max size is 2G . your admin can make it what ever you want on a partition
or they can create a file on multiple nodes that you can use for placing huge volume of data
HTH
Raghu M
Re: how to beyond the filelimit in Unix using sequence file?
Posted: Tue Sep 18, 2007 9:14 am
by donhoff
Raghumreddy wrote:Default max size is 2G . your admin can make it what ever you want on a partition
or they can create a file on multiple nodes that you can use for placing huge volume of data
HTH
Raghu M
Hi, you said my admin can make it what ever I want on a partition. Does that mean the actual max file size can only be limited by the disk space?
Someone told me that a HP-Unix or AIX can only have a max file size of 128G. Even you have many Ts of disk space, you could not produce a file beyond 128G. Is that true? This is very important for me.
Because for my estimating, the single file I extracted from DB is at least 300G
Posted: Tue Sep 18, 2007 10:48 pm
by Raghavendra
I have seen a file of 210GB on AIX server and so I belive a file size of max 120GB might not be true.
Posted: Wed Sep 19, 2007 7:52 am
by donhoff
OK,Thanks very much
Posted: Wed Sep 19, 2007 8:07 am
by chulett
Ask your SysAdmin what the limits are.