LANG Environment variable

reachmexyz · Post by **reachmexyz** » Tue Mar 15, 2011 1:42 am

Hello all.

I have a process where i pick data from tables, process them and create files.
When i run the process in produciton, i got the warning "Invalid character(s) ([x8000]) found converting Unicode string (code point(s): 2.00[x0002]0[x0080][x8000]3619108.3619114.715535.2056639.3919244.2056615.11202413.10971540[x0006]FOG[x000E]860...) to codepage ISO-8859-1, substituting.".
But when i run the same process in development, i do not get this error. In Development and production i was extracting the data from same table.
Later i checked the environment variables list and i found "LANG=en_US.UTF-8" variable on devlopment job log but not in the log of production process. Is this the culprit.
But when i go into linux environment and type "echo $LANG", i get "LANG=en_US.UTF-8" in production too. But i cannot see this in Job log ( Director). Is this environment varaiable not being respected by Production DataStage server. Please help me out.

reachmexyz · Post by **reachmexyz** » Tue Mar 15, 2011 9:25 am

Can somebody please respond to this?

ray.wurlod · Post by **ray.wurlod** » Tue Mar 15, 2011 4:35 pm

As a general rule, if someone can contribute something they will. Learn patience.

chulett · Post by **chulett** » Tue Mar 15, 2011 6:50 pm

It doesn't matter what *you* see from the command line, what matters is what exists in the job's environment when it runs... and even with 'the same user' they're not the same thing. Since it seems to be missing there on production, add it to the environment of that server at whatever level seems appropriate to you. For every job in every project, set your LANG (or NLS_LANG) variable in your dsenv file. Or constrain it to just that one Project or even just that one job by adding it at that level.

ajay.vaidyanathan · Post by **ajay.vaidyanathan** » Wed Mar 16, 2011 12:43 pm

Hi,
The UNICODE error is a very strange error. If you have defined the STRING fields in your job as Unicode, try making them as 'Non-Unicode' and reimport the metadata back to the job. This should fix your problem.