I meant to respond sooner.

To start, I am very confused in this statement:    
"What I was hoping to do was to split this so that data could go to two  
machines during the course of the experiment bearing in mind that the  write 
CIFS folder is changing over the course of the two weeks  continuously."

Please define:  
- Split the file (end after X bytes, create new file?)
- "the write CIFS folder is changing" (is it changing the mapping/mount?)

I'm curious what you actually "need" here -- just two systems reading the same 
file?  Splitting the files across multiple systems?

This is really a classic issue of the program and protocol, not any services.  
I 
run into this regularly with clustered nodes and file systems, they do not 
solve 
issues that can only be addressed at the program/client.  I'm also curious if 
you're running up against 32-bit limitations in the SMB client (2GiB file size 
limitation), hence why you need to change the system (?).

In the POSIX (UNIX/Linux) DAQ world, this is a classic case/use for 
buffers/pipes/sockets.  Instead of writing to a file, you write to a special 
file, which is then acted upon or serviced by another program.  So while one 
program is writing to what it thinks is a file, it's actually writing to a 
solution on the other end which can split the data up as it sees fit.  This is 
very commonly used for continuous logging, but also for data when long batch 
jobs are the norm too.

Although Microsoft raided SCO Xenix to add basic UNIX-like redirection and 
piping for MS-DOS 2.0, they did not bring over most of these features, much 
less 
NTFS and SMB do not support such.  However, it might be interesting to see what 
would happen if you created such a file on the UNIX side that is shared out via 
SMB.  I've never attempted such myself, as I don't know how SMB handles 
addressing blocks in such special inodes.

-- Bryan

P.S.  I've done a lot of DAQ in my time, with Windows, Linux, VxWorks and a 
couple of UNIX flavors (including RT).  I'm curious to know more about your 
instrument, drivers, etc...

-- 
Bryan J  Smith             Professional, Technical Annoyance 
Linked Profile:           http://www.linkedin.com/in/bjsmith 
------------------------------------------------------------ 
"Now if you own an automatic ... sell it!
You are totally missing out on the coolest part of driving"
-- Johnny O'Connell



----- Original Message ----
From: Bryan Hepworth <[email protected]>
To: "[email protected]" <[email protected]>
Sent: Thu, December 30, 2010 8:11:17 PM
Subject: [rhelv5-list] OT Question - Directory and File duplication

Hi All

I'm hoping someone may be able to point me to a specific utility for this 
problem. So far I haven't seen anything, so input is appreciated. I know this 
is 
totally off topic so just pointers is fine.

Problem is as follows..

I have a machine that creates sequencing data over a typical 10 working days 
period of time. The machine writes to a data area and then uses this data over 
the course of the experiment to analyse and re-write files - we end up with a 
file around 0.5TB once it's all done. The instrument -Windows -  needs a CIFS 
share which is fine, it writes to a samba share on a designated server. What I 
was hoping to do was to split this so that data could go to two machines during 
the course of the experiment bearing in mind that the write CIFS folder is 
changing over the course of the two weeks continuously. All suggestions would 
be 
most welcome.

Bryan

_______________________________________________
rhelv5-list mailing list
[email protected]
https://www.redhat.com/mailman/listinfo/rhelv5-list

Reply via email to