The job is very simple. But when DB2 API Stage was used and the target table name was like SDS7.SDS_I_143_RAW_MATERIAL, the job just kept running and hanged on over there. The job process became phantom process left there without parent process.
For more information, we can look into the files started with DSD.RUN and DSD.StageRun in details under project &PH& directory. We found the below error messages.
*** glibc detected *** malloc(): memory corruption (fast): 0x08646f60 ***
Abnormal termination of DataStage.
Fault type is 6. Layer type is BASIC run machine.
Fault occurred in BASIC program DSP.Open at address 574.
With a bit of work the actual statement that caused the failure can be found, but it is quicker to approach the problem from the other side. Is it just this table that is causing the problem? What data types are in this table?
I'm suspecting it's an environment issue. Maybe the installation or some component has some problem.
The strange thing is that the issue will go away after I changed the target table name, for whatever change. Or after I changed the schema name, then the issue has gone.
I did a test from the WinXP Information Server v8.1 environment on my Laptop, same job, connect to the same target table on the same database which was cataloged on my WinXP. There is no any issue.
The Information Server is version 8.1.
Red Hat Linux release is Red Hat Enterprise Linux ES release 4 (Nahant Update 4). Version is Linux version 2.6.9-42.ELsmp (bhcompile@hs20-bc1-1.build.redhat.com) (gcc version 3.4.6 20060404 (Red Hat 3.4.6-2)) #1 SMP Wed Jul 12 23:27:17 EDT 2006
DB2 is DB2/LINUX 9.1.3.
Just changing the name fixed it? Very odd indeed. Was the malloc() error caused by memory filling up? I suppose it is too late now to narrow down the cause.
After we replaced the DB2 UDB API stage by hashed file stage or sequential file stage, the job has no any problem. Is there anyway to trace into more detail so that I can identify what may be the root cause?
What solved the problem - renaming the table or using a different stage? You need to get back to where the problem is reproduceable in order to find out the cause by changing small things until the error goes away.
Currently, we changed all target table names to just make sure they're a little bit different from old names. But it's not the final solution, some time the same issue still occur. For instance, in one job, if the source sequential file is empty, then the job will hang on over there and the same error message appear again. If the source sequential file is not empty, the issue gone. Or we changed the NLS to none for all stages, then there is no issue even for empty source sequential file.
The job is very simple. But when DB2 API Stage was used and the target table name was like SDS7.SDS_I_143_RAW_MATERIAL, the job just kept running and hanged on over there. The job process became phantom process left there without parent process.
For more information, we can look into the files started with DSD.RUN and DSD.StageRun in details under project &PH& directory. We found the below error messages.
*** glibc detected *** malloc(): memory corruption (fast): 0x08646f60 ***
Abnormal termination of DataStage.
Fault type is 6. Layer type is BASIC run machine.
Fault occurred in BASIC program DSP.Open at address 574.
Thanks
Did you ever find a permanent solution for this? I have recently ran into the same problem, the parallel job mapping doesn't necessarily matter - whether it be dataset -> transformer -> db2 api or just row generator -> db2 api. One thing I've noticed as you did is changing the table name "fixes" this - furthermore what I found especially odd was the error is only thrown where the table name is 27 characters (ie DB2ADMIN.TF_PAID_PERCENTAGE).
We are running IIS 8.1 on RH 4.1.2 Linux 2.6.18 gcc 4.1.2 glibc 2.5-34 and db2 v9.7.
The job is very simple. But when DB2 API Stage was used and the target table name was like SDS7.SDS_I_143_RAW_MATERIAL, the job just kept running and hanged on over there. The job process became phantom process left there without parent process.
For more information, we can look into the files started with DSD.RUN and DSD.StageRun in details under project &PH& directory. We found the below error messages.
*** glibc detected *** malloc(): memory corruption (fast): 0x08646f60 ***
Abnormal termination of DataStage.
Fault type is 6. Layer type is BASIC run machine.
Fault occurred in BASIC program DSP.Open at address 574.
Thanks
Did you ever find a permanent solution for this? I have recently ran into the same problem, the parallel job mapping doesn't necessarily matter - whether it be dataset -> transformer -> db2 api or just row generator -> db2 api. One thing I've noticed as you did is changing the table name "fixes" this - furthermore what I found especially odd was the error is only thrown where the table name is 27 characters (ie DB2ADMIN.TF_PAID_PERCENTAGE).
We are running IIS 8.1 on RH 4.1.2 Linux 2.6.18 gcc 4.1.2 glibc 2.5-34 and db2 v9.7.
I am in the same issue now, Job is very simple Its a DB2 UDB stage to a Sequential file.
1) Fetching only one field and its an Integer( Tried changing to Char also) 2) Volume is 325944 ( Tried fetching first 10 rows also)
3) having this error with only one table. I have plenty of other jobs working fine.changing the table in the same stage works fine ,Perticularly if I use one table name DNB_SCORE it throws this error.(With schema it willbe TDB2CPMD.DNB_SCORE- here My table name is less than 27 charecters I dont think its because of the name length)
rajan.n wrote:
I am in the same issue now, Job is very simple Its a DB2 UDB stage to a Sequential file.
1) Fetching only one field and its an Integer( Tried changing to Char also) 2) Volume is 325944 ( Tried fetching first 10 rows also)
3) having this error with only one table. I have plenty of other jobs working fine.changing the table in the same stage works fine ,Perticularly if I use one table name DNB_SCORE it throws this error.(With schema it willbe TDB2CPMD.DNB_SCORE- here My table name is less than 27 charecters I dont think its because of the name length)
If some one had any solution can please help in this ?
Thanks,
Rajan
Hi Rajan,
The table name was an issue for me when the db2 api stage was the input. Although even more recently I am now seeing the same issue as you - where it is not the table name and the stage is a source/output. Are you entering custom sql? Do you have NLS enabled? What are your gcc, glibc, db, linux, etc versions? Just trying to get a little grounds for comparison here - seems a lot of people are experiencing this issue but no resolutions are being posted.
rajan.n wrote:
I am in the same issue now, Job is very simple Its a DB2 UDB stage to a Sequential file.
1) Fetching only one field and its an Integer( Tried changing to Char also) 2) Volume is 325944 ( Tried fetching first 10 rows also)
3) having this error with only one table. I have plenty of other jobs working fine.changing the table in the same stage works fine ,Perticularly if I use one table name DNB_SCORE it throws this error.(With schema it willbe TDB2CPMD.DNB_SCORE- here My table name is less than 27 charecters I dont think its because of the name length)
If some one had any solution can please help in this ?
Thanks,
Rajan
Hi Rajan,
The table name was an issue for me when the db2 api stage was the input. Although even more recently I am now seeing the same issue as you - where it is not the table name and the stage is a source/output. Are you entering custom sql? Do you have NLS enabled? What are your gcc, glibc, db, linux, etc versions? Just trying to get a little grounds for comparison here - seems a lot of people are experiencing this issue but no resolutions are being posted.
Thanks
Thomas, Thanks for getting back on this.
Both ways I am trying, entering Custom SQL and builing the SQL from SQLBuilder, Currently its in Custom mode as I am trying to chage the variable type.
here are the version details,
glibc - 2.4- 31.2,gcc 4.1.0,Linux 2.6.16.54-0.2.12-bigsmp and I dont have permit to find the DB2 version , can provide it later.
Assuming you can afford to disable NLS for this particular stage I have come up with a temporary workaround until the root cause is identified. In the DB2 API stage under the NLS Map tab rather than selecting from the list simply type 'OFF'.
Fortunately for this particular project NLS wasn't required so shutting it off on a stage level doesn't affect anything I'm doing - hopefully this is an option for you.
Assuming you can afford to disable NLS for this particular stage I have come up with a temporary workaround until the root cause is identified. In the DB2 API stage under the NLS Map tab rather than selecting from the list simply type 'OFF'.
Fortunately for this particular project NLS wasn't required so shutting it off on a stage level doesn't affect anything I'm doing - hopefully this is an option for you.
DJ
Amazing Man, Job ran sucessful , Thanks a ton. I dont know what it has to do with NLS internally, but the job ran fine thats all I wanted , will need to figure it out, Thanks again.
</a>
, will need to figure it out, Thanks again.