WriteHash() - Write failed for record id

[attu]

when we move our job from dev to test and try to run it we get this fatal error message

Code: Select all

WriteHash() - Write failed for record id 'BRT019453K 53425'

After we re-compile the job it runs fine. It does not look like it is related to data issue because the job runs fine after we compile it. The fatal error comes in two of the hashed file stages - we are writing to hashed file and at the same time doing a lookup as well. Update action is clear file before writing and allow stage write cache is enabled for the hashed stage.

what could be the reason for job abort and why it runs fine after compiling?

Please provide feedback, Thanks

[Vinodanand]

A search on Write failed for record id would have provided you with the solution . Here is my bit in helping you out ..

http://dsxchange.com/viewtopic.php?t=11 ... ce9937713d

[dsvsinformatica]

Is the same user running job or is it different users? and tell the Ulimit settings for that users.

[sysmq]

Hi
I have the same error
I went through the forum looking for idea, checked all the recomandation mentioned in the forum (Data and over file size etc...)
yet as mentioned only re-compile solves this issue.
i cannot do this on regular beses in a production server

any other ideas ?

Here is the error message :
job.pearson.LT_After_Reject:writehash() - write failed fro record ID '4597291'

[Sathishkumarins]

Some times Hashed file size may be a problem.
Try using a 64 bit hashed file. This might solve your problem.

Because in a 64 bit hashed file largest address that can be represented is 19million TB.

[ray.wurlod]

It's definitely not a "64-bit" error. These give different errors. Write failed is usually because the key is illegal (contains mark characters or is null) or because the hashed file is internally corrupted.

[swordmood]

Hi,
I had the same error .
Maybe there is not enough memery or disk space.
And you can try.