Reason for mutex_unlock() Error ?

[nitin376]

Hi !

I am using a funnel stage that combines data from two source. Now , the problem is that from one of the source (Transformer Stage) , I am getting the following error:

ds_ipcflush() - Error in mutex_unlock() - Not owner

Please advice me the reson for the particular error.

Thanks
Nitin

[meena]

Hi,
I think it is better you check with your DBA once. And can explain what exactly you are doing after the funnel stage(loading the data into any table or etc..).

[nitin376]

Hi,
I think it is better you check with your DBA once. And can explain what exactly you are doing after the funnel stage(loading the data into any table or etc..).

Hi Meena,
Thanks for the reply !

Well after the funnel stage'
Funnel stage -->TX--->Seq stage (target )

at the TX , performing lookups and updating tables in Oracle database, then the output goes into the seq stage (flat file)

Thanks ,
Nitin

[thumsup9]

Do you have enough space on the disk where you are writing your target file.

[nitin376]

Do you have enough space on the disk where you are writing your target file.

See I am doing a conversion from server to parallel , the job runs perfectly fine with same parameters in server.

Therefore, space wise target is fine. I am creating this job in parallel just to speed up the process.

let me know where am I wrong !
what can be done to resolve that error

[thumsup9]

Do you have any other stages like IPC or Link Partioner in your job design. Once had a similar problem and removed the IPC and it performed OK.

Another option I found in the forum is to check inter process under Enable row buffer option and increase the time out-say 60sec.

HTH

[nitin376]

Do you have any other stages like IPC or Link Partioner in your job design. Once had a similar problem and removed the IPC and it performed OK.

Another option I found in the forum is to check inter process under Enable row buffer option and increase the time out-say 60sec.

HTH

see I am working in parallel extender, there is no IPC stage in it

[ArndW]

...ds_ipcflush() - Error in mutex_unlock() - Not owner...

This message is unfortunately not particularly help. When one side of the pipe terminates and the other side can no longer communicate you get this error by the surviving process. So this is just reporting that your job has terminated and isn't itself the error. Do you have any other messages in your log that could be used to track down the error? Have you tried a reset to see if you get any additional information?

[thumsup9]

Do you have any other stages like IPC or Link Partioner in your job design. Once had a similar problem and removed the IPC and it performed OK.

Another option I found in the forum is to check inter process under Enable row buffer option and increase the time out-say 60sec.

HTH

see I am working in parallel extender, there is no IPC stage in it

Sorry Nitin, had a big fat lunch!! As ArndW suggested, do you have any other information in the log..

[thumsup9]

Do you have any other stages like IPC or Link Partioner in your job design. Once had a similar problem and removed the IPC and it performed OK.

Another option I found in the forum is to check inter process under Enable row buffer option and increase the time out-say 60sec.

HTH

see I am working in parallel extender, there is no IPC stage in it

Sorry Nitin, had a big fat lunch!! As ArndW suggested, do you have any other information in the log..

[nitin376]

...ds_ipcflush() - Error in mutex_unlock() - Not owner...

This message is unfortunately not particularly help. When one side of the pipe terminates and the other side can no longer communicate you get this error by the surviving process. So this is just reporting that your job has terminated and isn't itself the error. Do you have any other messages in your log that could be used to track down the error? Have you tried a reset to see if you get any additional information?

The Job desc is as follows:

seq stage -----> TX------>Funnel<-----TX------Oracle EE
.........................................|
.........................................|
.........................................V
........................................TX stage

the error details are as follows :

ETRValidateShip.#1.X_CURRENTSHIP.l_CurrentShipList_SeqOut-Input.l_CurrentShipList_SeqOut: ds_ipcgetnext - timeout waiting for mutex

Attempting to Cleanup after ABORT raised in stage ETRValidateShip(X_CURRENTSHIP).#1.X_CURRENTSHIP

ETRValidateShip.#0.X_CURRENTSHIP.l_CurrentShipList_SeqOut-Input.l_CurrentShipList_SeqOut: ds_ipcgetnext - timeout waiting for mutex

Attempting to Cleanup after ABORT raised in stage ETRValidateShip(X_CURRENTSHIP).#0.X_CURRENTSHIP

please advise me !
Nitin

[ray.wurlod]

Inter-process communication is automatically used in parallel jobs - indeed there is no IPC stage. Timeouts in IPC most usually result from overloaded systems or, in MPP configurations, overloaded or damaged networks.

Access to resources is controlled by semaphores, which are implemented differently on different kinds of UNIX. On your operating system semaphores are implemented as mutual exclusion (mutex) locks.

Read the error message carefully, and answer these questions based on what you read there.
1. What is the job name?
2. Was this job running in multi-instance mode at the time?
3. What is the stage name in which the error occurred?
4. What is the name of the link from which the error was reported?
5. What is the purpose of this link in the job?

Inspect your job logs. Do all instances read different source files, or is it possible that two (or more) instances might have been reading the same file? From the "environment variables" entry (second event for the job run), please advise what buffering strategy was being used.