Powerplay6.5: 'Cannot read memory' error

[vjmanda1]

We are currently using PowerPlay6.5 on NT. We have to migrate to Win2K as an IT policy. But when we installed this version on win2K and start the data refresh, after checking all the relations/measures/dimensions, PP is raising a 'Cannot read Memory' error.

Anyone came across this issue earlier?

 

[CognosProfessional]

Did you designate a TEMP file location?

CP

 

[vjmanda1]

Hi CP, thanks for checking my posting. Still I havent been able to resolve the issue. I have checked all the permissions and also the available space on the c:drive and everything seems to be OK.

Do you know of any patch that is available from Cognos?

I am attaching the log file, if it is of any help.

Following is the windows error that I am getting, before it closes down the app.

'trnsfrmr.exe has generated errors and will be closed by Windows. You will need to restart the program.

An error log is being created.'

And then if I try to restart the process again, its raising the following message.

'TR3419 Transformer was stopped in the process of CREATEFILES. DO you wish to continuw the processing? ' with OK, cancel buttons.

Here is the log file from transformer.

---- Start Log
PowerPlay Transformer(6.5.311.0 ) Wed Jul 09 15:49:31 2003

LogFileDirectory=c:\temp
ModelSaveDirectory=
DataSourceDirectory=
CubeSaveDirectory=
DataWorkDirectory=
ModelWorkDirectory=
MaxTransactionNum=500000
ReadCacheSize=16384
WriteCacheSize=65536

07/03/2003 03:50:05 PM 4 00000000 Start cube update.
07/03/2003 03:50:05 PM 4 00000000 Initializing categories.
07/03/2003 03:50:06 PM 4 00000000 Timing, INITIALIZING CATEGORIES,00:00:01
07/03/2003 03:50:06 PM 4 00000000 Start test build with data source 't:\iion_prd\powerplay\queries\dev\fund indicator\iion01r\fund_geo.iqd'.
07/03/2003 03:50:06 PM 4 00000000 Reading source data.
--- End log

 

[vjmanda1]

CP.. one more thing. I have also tried running the 'Great Outdoors' sample and it worked fine for me.

I open the .pyh file thru transformer and tried to check the categories, which resulted in this error.

Just wanted to give you the whole story. Hope you can help.

 

[vjmanda1]

CP, here is the entry in the event log that I have. Have any idea if this is a known issue with this version?

Application popup: PowerPlay Transformer - fund_indicator_01r.pyh: trnsfrmr.exe - Application Error : The instruction at "0x0059da88" referenced memory at "0x00000001". The memory could not be "read".

Click on OK to terminate the program
Click on CANCEL to debug the program

 

[NZTechSupp]

Hi,

I have taken a look at your log file entry above.
Your:
ReadCacheSize=16384
WriteCacheSize=65536
should be the same size. Increase your ReadCacheSize to 65536 as well.
This should be done in the Cer2.ini file found under Program files->Cognos->Cer2->Bin.
Also check to see that the following function is in this file: SORTMEMORY=512, your temp directoy as previously mentioned should also be set to C:\Temp.

Hope this helps.
Regards,
NZTechSupport

 

[NZTechSupp]

Hi,

Another thing, this may be caused by a buildup of files in the temporary directory. Find and delete any .QY? and .TMP files in all local and network drives. Confirm there is sufficient free space in the temp directory location. If necessary, reboot, or run a scandisk and defrag on the harddrive.

This error may also occur if the model is corrupt.

 

[CognosProfessional]

Does this happen with the .mdl also?
Is your TEMP directory empty?
Do you run other .pyh's and don't get error?
When you click "Modify Columns" in each Data Source in the model, do you get an error?
Can you run each data source separately without error?
How many data sources are in the model?

CP

 

[vjmanda1]

I have checked all the above recommendations and all the values are intact.

But the only difference is, I dont have a cer2.ini, instead I have a cer1.ini. Can you provide me a cer2.ini sample, if it is different?

I checked cloumns on each query, and also successfully test build with 100 rows.

When I generate categories from any query, it is checking the columns successfully and then while reading source data, after 38421 rows are read, it raises this exception.

It generated a Dr.Watson error and from the log it indicates that, the exception is raised at RootChangeToCategory, NtWaitForMultipleObjects function calls.

We have only one database that we go against and we have 8 queries (.iqd) and on NT version version everything runs OK. This issue is only when we run on Win2K server.

Hope this information helps.

 

[nagrajm]

In addition to NZTechSupport's advice find and rename TRNSFRMR.INI(to say TRNSFRMR.OLD), re-launch transformer and rebuild the cube. See if this helps.

Nagraj

 

[vjmanda1]

nagrajm,
I tried that also and still getting the same error when it reaches 38421 records. Any other thoughts?

DO you know if they have a patch available?

thanks.

 

[CognosProfessional]

cer1 is for the old version. You must be using the old version, which is fine, you don't need cer2 unless you upgraded and it automatically shows. There is not much difference.

Sounds like your 8 .iqd's is getting confused at a certain point when it tries to allocate all those categories and is using too much memory for the setting of memory. You are going to need some memory/temp/cache changes. Something along those lines must be different now than before the upgrade.

CP