Too many columns in a Record

rafidwh · Post by **rafidwh** » Wed Jan 18, 2006 9:37 pm

Hi All,
I get flat files daily generated from COBOL extract,the job was running smoothly ,but yesterday in the prod the job is showing a lot of warnings like "Too many columns in a record".Due to this number of warnings they have stopped running the job and asking me to analyze.i am using a comma delimiter in the squential file stage.
it is looking strange to me that it has run smoothly before,but now it is showing this warnings.
Pl suggest me

Andal · Post by **Andal** » Wed Jan 18, 2006 9:55 pm

There may be problems with the file you have got yesterday. Check if there is any extra commas in the columns itself. And also check the line terminator settings of the Sequential file. It should be unix in your case.

chulett · Post by **chulett** » Wed Jan 18, 2006 10:10 pm

This can happen when string fields are not quoted and there is a comma (or whatever the delimiter of choice is) in the data for a field. Or fields.

rafidwh · Post by **rafidwh** » Thu Jan 19, 2006 9:17 pm

what will be the reason if I am geting the same warning"Too many columns in a record" ,when I am selecting "fixed width column". and not any delimeters

jzparad · Post by **jzparad** » Thu Jan 19, 2006 9:23 pm

There was a recent discussion about a similar problem.

If all your files look OK and the job used to run OK, check if the job has active stages in sequence and, if so, try placing an IPC between them.

chulett · Post by **chulett** » Thu Jan 19, 2006 9:24 pm

Then your metadata is wrong. The Display Size of your fields don't add up to the actual length of the records.

bmadhav · Post by **bmadhav** » Thu Jan 19, 2006 9:41 pm

Since u are reading COBOL output, u might be running into EBCDIC to ASCII conversion issues.

rafidwh · Post by **rafidwh** » Thu Jan 19, 2006 9:54 pm

But chullet,The problem is it is running well without any warnings for some files,but for the last file it gave warnings ,so is that the metadata issue?

chulett · Post by **chulett** » Thu Jan 19, 2006 10:02 pm

Then check the file that is giving you the error. There must be something different about it from the ones that run without error.

Since you say they are fixed width, then it should be easy to tell. Each record should always be X bytes in length, the sum of the Display Sizes defined on your Columns tab. The bad file should have records > X bytes long.

rafidwh · Post by **rafidwh** » Thu Jan 19, 2006 10:41 pm

chulet,
I have checked the cobol program it shows that the record lengths are fixed in the program as well,so there wont be any Byte changes in the extract,
I am surprised how does I got the warning messages in the previous file

.

ray.wurlod · Post by **ray.wurlod** » Thu Jan 19, 2006 11:13 pm

Don't check the program, check the file!

DataStage isn't complaining about the program, it's complaining about the file.

What's the row length of the file? What is the total size of the file (use the wc command)? Is one an exact multiple of the other? If not there is at least one aberrant row in the file.

srinagesh · Post by **srinagesh** » Fri Jan 20, 2006 5:25 am

We got a similar error before and we checked the , in the Source file stage, "supress row truncation warnings" (in the format tab) and Make the "Incomplete column" (in the columns tab) (for each element of the row, except the last one) as "Retain"

Hope this helps

Nagesh