Cube refresh: Transformer fails, no log created

[ampt]

We are refreshing cubes using this Windows batch command:

COGTR -nologo -n1 -dLogFileDirectory="D:\Cognos BI\Log Files" -p"D:\Cognos BI\ModelWork\%_var1%.pyj"

- where _var1 is a paramenter passed in = the name of the cube (also the name of the .pyj file)

98% of the time this works perfectly, but sometimes Transformer hangs without creating a log file.

Transformer 8.4.1, Windows Server 2008, 4G RAM

Has anyone run into this?

[bdbits]

I have had that happen from time to time. Unfortunately it can be hard to track down. A few suggestions... Check the Windows event logs. Check to see what else if anything might be running at the time like backups, virus scans, etc. or anything that might put a lock on your data sources. Make sure your current directory is where cogtr.exe is located, even if that is on the path. Add "-r4" to your command line (should be default but might help). If running this as a scheduled task, you may want to use -n2, also check the credentials, and the task history might yield something. Also if you have had abnormal terminations of cogtr, it sometimes helps to go in and clean out %temp% and the various working folders you have configured in Transformer.

I know a lot of this is a bit trivial, but sometimes it's the little things. Hope something helps out.

[Arsenal]

how big is the cube and how many records on the DB? Do you have logging enabled? (preferences sheet)

The times it fails, do you know if the DB runs some activities at the time, such as backup etc.? I ask because of the possibility of a timeout during read althoughb it should still generate a log

What is the size of the hard drive - might run out of temp space at the time of writing records to the cube

The ram seems a tad small at 4 GB but again, that would depend on the number of records it reads, and the categories it generates

[jonathan.webb@raytheon.co]

I've had issues before when the command line waits for a response.. and when you're running your cube build as a batch job, no screen is there to give it a response, so it just hangs... maybe try adding an echo off to your script. Also, your model isn't prompting for security credentials? Make they're embedded in the sign-on screen on the model as auto logon

[gene.beljaeff]

We have noticed this same problem for many month, now. We can't find the root cause. I've read some things that might point a finger at the Virus Scanner, but we've added some directories and some file extensions to the scanner's Exclusions list -- but that didn't seem to fix anything.

In our script, we have this command:
COGTR -nologo -n1 -dLogFileDirectory="D:\Cognos BI\Log Files" -p"D:\Cognos BI\ModelWork\%_var1%.pyj"

followed a bit later by this one:
COGTR -nologo -dLogFileDirectory="D:\Cognos BI\Log Files\Deployment Logs" -n1 -ox -g -p"D:\Cognos BI\ModelWork\%_var1%.pyj"

It works 98% of the time.