RT_SCTEMP - error number is 1

[Amos.Rosmarin]

Hi,

I got the number 1 error

Error reading record CO_NLIV_XREF.22..fifo from file RT_SCTEMP - error number is 1
Parallel job reports failure (code -99)

I think I should be happy but, does anyone knows what does it mean ?



Amos

[ray.wurlod]

This is probably the operating system error number 1. I can't remember for sure what 1 is; 2 is "file not found", so 1 might be "file does not exist".

[ArndW]

Amos,

is your error reproduceable? The RT_SCTEMP is a hash file used to store records for locking/syncronization during a Px job execution, but the error 1 is not a DataStage read error.

The RT_SCTEMP file should be empty when no jobs are running.

[Amos.Rosmarin]

Thanks Ray,

It's a basic job :

datasaet --> modify --> loopup fileset


why is it that the fifo file does not exist ?

it runs in multiple instance, other instances were ok .




Amos

[ray.wurlod]

Are you really running a parallel job in multiple instances? In that case, it may be that one job had already created a named pipe for inter-process communication, and another instance could not, therefore, create the same named pipe. I'm guessing here, but it highlights one of the many dangers when running parallel jobs as multi-instance. You really don't have to do this. Reserve multi-instance for server jobs.

[Eric]

"multiple instances", does this mean job invocation?

the fifo should have a uniquie name so you should be able to run multiple invocations with no problem, although if they are all trying to write to the same fileset at the same time you will have a problem!

[Amos.Rosmarin]

thanks all,


as you can a unique fifo was created

CO_NLIV_XREF.22..fifo

when 22 is the invocation id

I do not agree with you that in PX job there is no place for multiple invocations, what if you have different files with the same schema? using multiple invocations is just an elegant way for extending code re-use.
The source and target file names are both parameters so there is no chance that 2 jobs will use the same name (and I double checked it just to be sure)

the fifo file is an internal file that the px creates, I did not know it exists until it failed.



Cheers,
amos

[ray.wurlod]

I was unaware that the invocation ID becomes part of the named pipe file. Given that it is, I retract my previous guess and leave you to research what O/S error code 1 means and what might have happened to prevent the file from being created.

[vam]

Hi,
i got the same error today and it went away after re-run.. do you think we need to do any changes in the setting in datastage to avoid these kinds of errors..

Error reading record PRD_NEW..fifo from file RT_SCTEMP - error number is 1

This is the parallel job. reads the data from flat file and loads into mysql using odbc stage.

Thank you
vam