Sequential files are not databases. There is no
concept of a "row" of data, it is just bytes of data.
A or is just data. If you have two
processes spooling data into the same file, the
operating system (if it lets you) will merge on a
block of bytes basis, not rows. This is readily
verifiable with a some simple testing.
So, when DataStage tries to write output to the same
file twice simultaneously via completely separate and
independent streams, then your data will probably be
garbage. This is not the same as two output links
writing to the same file, I'm talking about
independent streams which can be achieved with
simultaneous jobs or a single job with an IPC stage
that is detaching processes for simultaneity.
Since the solution of using a hash file is a database
construct, your operation will work.
---
gbarsky@osde.com.ar wrote:
>
>
>
>
> Craig:
>
> Thank you. I followed your instructions and
> replacing the seq file with a
> hash table, it worked fine.
>
> But I stil do not understand what does "... write
> three output streams to
> the *same* sequential file simultaneously ..." mean.
> Are they really
> simultaneously, or perhaps they are so so close in
> time (with a difference
> of microseconds), that the machine"thinks" that this
> is simultaneously.
>
> Are the three links arriving to the seq file, really
> arriving at the same
> time ? How is this translated and interpreted when
> compiled ?
>
> Thanks, and have a nice weekend.
>
>
> Guillermo P. Barsky -
gbarsky@osde.com.ar
> Gerencia de Sistemas - Desarrollo
>
> OSDE Binario - Filial Metropolitana
> Alem 1067, Piso 16
> TE (5411)4510-4330, Fax (5411)4510-5480
>
http://www.osde.com.ar
>
>
>
>
>
> "Craig Hulett"
>
>
> Para: "'DataStage Users Discussion List'"
>
> et>
> cc:
>
> Enviado por:
> Asunto: RE: Execution order in a server job
>
> datastage-users-bounces
>
>
> @oliver.com
>
>
>
>
>
>
>
>
> 26/02/2004 11:44 p.m.
>
>
> Por favor, responda a
>
>
> DataStage Users
>
>
> Discussion List
>
>
>
>
>
>
>
>
>
>
>
>
> Guillermo,
>
> This isn't an 'execution order' problem. The error
> is caused by trying to
> write three output streams to the *same* sequential
> file simultaneously.
> This cannot be done, period. And before you ask,
> this has nothing to do
> with
> DataStage but is due to the fundamental nature of a
> 'sequential' file.
>
> Change your job to write to three separate files and
> then combine them in
> an
> after-job routine. Or, write the records to a hash
> file instead and then
> spin the hash file out to a sequential file after
> it's been populated.
>
> -craig
>
> -----Original Message-----
> From:
datastage-users-bounces@oliver.com
> [mailto:
datastage-users-bounces@oliver.com] On
> Behalf Of
>
gbarsky@osde.com.ar
> Sent: Thursday, February 26, 2004 1:35 PM
> To:
datastage-users@oliver.com
> Subject: Re: Execution order in a server job
>
>
> Ray:
>
> Thank you for your answer.
>
> The problem I'm having, is that a sequential file is
> fed from and
> aggregator, and from other two transformers at the
> same time. All records
> are "appended" to the file. For the run (and for the
> test data), there are
> no records, so it seems that three links are
> arriving at the same time at
> the seq file, and an error occurs: "DSD.SEQOpen
> Unable to create file
> ......".
>
> But the file in the error, was created in a previous
> batch job (job A),
> which first is creating the file, and then is
> calling the server job (job
> B).
>
> I really don't know why the server job is trying to
> create the file.
>
> Could you give me any hint to follow and solve the
> error ?
>
> Thanks again.
>
>
> Guillermo P. Barsky -
gbarsky@osde.com.ar
> Gerencia de Sistemas - Desarrollo
>
> OSDE Binario - Filial Metropolitana
> Alem 1067, Piso 16
> TE (5411)4510-4330, Fax (5411)4510-5480
>
http://www.osde.com.ar
>
>
> _______________________________________________
> datastage-users mailing list
>
datastage-users@oliver.com
>
http://www.oliver.com/mailman/listinfo/datastage-users
>
>
>
> _______________________________________________
> datastage-users mailing list
>
datastage-users@oliver.com
>
http://www.oliver.com/mailman/listinfo/datastage-users
__________________________________
Do you Yahoo!?
Get better spam protection with Yahoo! Mail.
http://antispam.yahoo.com/tools
_______________________________________________
datastage-users mailing list
datastage-users@oliver.com
http://www.oliver.com/mailman/listinfo/datastage-users