Page 1 of 1

Timeout waiting for mutex Error in Server Shared Container

Posted: Wed Sep 23, 2009 10:26 pm
by rajeev_prabhuat
Hi,

We have a parallel job from which we are calling server shared container to generate CRC and the job is failing with the following errors mentioned below:

stdhub_Account_Pack_Hub.#0.stdhubCRCAcctidDF.Output_Acct_Table-Input.Input: ds_ipcgetnext() - timeout waiting for mutex

Attempting to Cleanup after ABORT raised in job stdhub_Account_Pack_Hub(stdhubCRCAcctidDF).#0.stdhubCRCAcctidDF.Transformer_0

Abnormal termination of stage stdhub_Account_Pack_Hub(stdhubCRCAcctidDF).#0.stdhubCRCAcctidDF.Transformer_0 detected

Please provide your inputs on how this can be resolved, we had searched the forum for Mutex and SPINSLEEP, SPINTRIES. But could not find resolution, please help those who had resolved this problem. Also please provide help on how to use SPINSLEEP & SPINTRIES.

Regards
Rajeev Prabhu

Posted: Wed Sep 23, 2009 10:49 pm
by chulett
What operating system?

Posted: Wed Sep 23, 2009 11:18 pm
by prasad.bodduluri
just increase ipc stage bufffer and time out values. i think it will work.

Posted: Wed Sep 23, 2009 11:20 pm
by ray.wurlod
Prevention is always better than cure. Make sure that whatever it was that timed out does not need to do so. Often timeouts are caused by overloaded servers. In this particular case the problem appears to be in the IPC interface between the parallel job and the server shared container.

Posted: Wed Sep 23, 2009 11:26 pm
by rajeev_prabhuat
chulett wrote:What operating system? ...
Hi Ray,

It is Linux box, with Datastage 8.0.1.

Regards
Rajeev Prabhu

Posted: Wed Sep 23, 2009 11:30 pm
by rajeev_prabhuat
prasad.bodduluri wrote:just increase ipc stage bufffer and time out values. i think it will work.
Hi Chulett,

We are not using IPC stage below is the flow of the jobs:

ODBC--> Transformer-->Container --> ODBC

In Container

inport --> CRC generation (transformer) --> outport


Are you telling to use IPC stage instead of Transformer? But thing is that this job is running for for some loads and not running for others!!! :-( Please help.

Regards
Rajeev Prabhu

Posted: Wed Sep 23, 2009 11:32 pm
by rajeev_prabhuat
ray.wurlod wrote:Prevention is always better than cure. Make sure that whatever it was that timed out does not need to do so. Often timeouts are caused by overloaded servers. In this particular case the problem app ...
Hi Ray,

Sorry to say that I am not premium user, can you put the comments in nomal text :), so that I can understand the solution that you have provided. :(

Regards
Rajeev Prabhu

Posted: Thu Sep 24, 2009 12:06 am
by ray.wurlod
Premium membership is one of the ways that DSXchange's hosting and bandwidth costs are funded. I do not propose to establish any precedent that undermines the ongoing operation of DSXchange.

The alternative is for you to obtain a premium membership. It's not expensive, at less than 30c (Rs 12) per day.