BODS 4.2 SP1 - Job fails when set on Pageable cache

[Former Member]

Hello Friends,

We are upgraded from 3.1 to 4.2 SP1 .

Within a job, at few dataflows, the job fails when it tries to use pageable cache.

However the surprising thing is, the same job when we run on our old BODS 3.1 it always works fine with pageable cache.

BODS 3.1 (unix environment) has - using PAGEABLE Cache with <3531 MB> buffer pool.

where as BODS 4.2 (window 2008) has more - using PAGEABLE Cache with <8192 MB> buffer pool. How come the job fails inspite of having more PAGEABLE Cache memory.

The machine has more then 18GB of memory allocated for cache.

Has anyone faced similar problem before or is there any other place where we need to increase the cache (any intermediate layer which is used by job server to pass on the data to the machine & again get it back)

I am getting 2 types of error,

1st Error -

RUN-050406: |Session STG_IMS_MAIN_LOAD_Job|Work flow WF_IMS_MAIN_DW_LOAD_DUMMY|Work flow WF_DW_INVENTORY_IMS|Work flow WF_DW_INVENTORY_IMS_1|Work flow WF_DW_INV_PRG_LD_IMS_Mod|Data flow ORDR_ReUse_DF_DW_INV_PRG_LD_IMS_Mod1

Data flow <ORDR_ReUse_DF_DW_INV_PRG_LD_IMS_Mod1> received a bad system message. Message text from the child process is

<a ==========================================================

Collect the following and send to Customer Support:

1. Log files(error_*, monitor_*, trace_*) associated with this failed job.

2. Exported ATL file of this failed job.

3. DDL statements of tables referenced in this failed job.

4. Data to populate the tables referenced in the failed job. If not possible, get the last few rows (or sample of them) when

the job failed.

5. Core dump, if any, generated from this failed job.

==========================================================>. The process executing data flow

<ORDR_ReUse_DF_DW_INV_PRG_LD_IMS_Mod1> has died abnormally. For NT, check errorlog.txt. For HPUX, check stack_trace.txt. Also,

notify Technical Support.

RUN-050409: |Session STG_IMS_MAIN_LOAD_Job|Work flow WF_IMS_MAIN_DW_LOAD_DUMMY|Work flow WF_DW_INVENTORY_IMS|Work flow WF_DW_INVENTORY_IMS_1|Work flow WF_DW_INV_PRG_LD_IMS_Mod

The job process could not communicate with the data flow <ORDR_ReUse_DF_DW_INV_PRG_LD_IMS_Mod1> process. For details, see

previously logged error <50406>.

2nd Error -

System Exception <ACCESS_VIOLATION> occurred. Process dump option is off. Process is not dumped.

Call stack:

0x0000000080514B01, PageableCacheBPT_impl::getPageableStats()+0033 byte(s),

d:\im_ds_4.2_sp_rel\src\dataservices\dataintegrator\codeline\code\src\eval\pageablecachebpt_impl.cpp, line 0566+0007 byte(s)

0x00000000804CE47F, gatherPageableStats::getAvgRowSize()+0015 byte(s),

d:\im_ds_4.2_sp_rel\src\dataservices\dataintegrator\codeline\code\src\eval\pageablecache_impl.cpp, line 0492

0x00000000801BE0BC, XDataflow_desc::executeInline()+1404 byte(s),

d:\im_ds_4.2_sp_rel\src\dataservices\dataintegrator\codeline\code\src\core\xnet.cpp, line 1157+0011 byte(s)

0x00000000801C777D, XDataflow_info::execute()+0925 byte(s),

d:\im_ds_4.2_sp_rel\src\dataservices\dataintegrator\codeline\code\src\core\net.cpp, line 0365

0x00000000801BAA22, XDataflow_desc::execute()+0114 byte(s),

d:\im_ds_4.2_sp_rel\src\dataservices\dataintegrator\codeline\code\src\core\xnet.cpp, line 0838+0016 byte(s)

0x00000000801C4E47, XDataflow_info::compute()+0343 byte(s),

d:\im_ds_4.2_sp_rel\src\dataservices\dataintegrator\codeline\code\src\core\net.cpp, line 0538+0013 byte(s)

0x00000000801C1F16, XDataflow_info::compute()+0326 byte(s),

d:\im_ds_4.2_sp_rel\src\dataservices\dataintegrator\codeline\code\src\core\net.cpp, line 0658+0059 byte(s)

0x0000000080066148, AE_Main_Process_Options()+60424 byte(s),

d:\im_ds_4.2_sp_rel\src\dataservices\dataintegrator\codeline\code\src\xterniface\actamainexp.cpp, line 5205

0x0000000080076946, AE_Main()+3238 byte(s),

d:\im_ds_4.2_sp_rel\src\dataservices\dataintegrator\codeline\code\src\xterniface\actamainexp.cpp, line 1422+0036 byte(s)

0x0000000040002266

0x0000000040002ADA

0x000000004000F9F0

0x000000007763652D

0x00000000779BC541

Registers:

RAX=00000000817858E8 RBX=000000000D09ED50 RCX=0000000000720065 RDX=000000000D09ED50 RSI=000000000E0BCF90

RDI=000000000DC6B710 RBP=000000000D09ED50 RSP=0000000000206BD0 RIP=0000000080514B01 FLG=0000000000010206

R8=000000000BFC001E R9=000000000C20001F R10=0000000000460044 R11=0000000081780440 R12=000000000D788E20

R13=000000000D789038 R14=000000000D7B9530 R15=0000000000000000

Exception code: C0000005 ACCESS_VIOLATION

Fault address: 0000000180514B01 01:0000000000513B01 D:\BI-IPS\Data Services\bin\acta.dll

==========================================================

Collect the following and send to Customer Support:

1. Log files(error_*, monitor_*, trace_*) associated with this failed job.

2. Exported ATL file of this failed job.

3. DDL statements of tables referenced in this failed job.

4. Data to populate the tables referenced in the failed job. If not possible, get the last few rows (or sample of them) when

the job failed.

5. Core dump, if any, generated from this failed job.

==========================================================

[Tdelaloy]

Hello All,

After the migration we had exactly the same issue on one dataflow which used a lot of lookup functions in one Query (6 lookups on huge tables).

It was not sufficient to switch the dataflow from "pageable" to "in memory" mode. In addition I edited our lookup functions to move from "On demand cache" to "No Cache" and activate the calculation in a distinct memory thread. With those 2 parameters we were able to execute the dataflow without additional rework or split.

Regards,

Thomas

[former_member197676]

Hi Dvn,

I have the same issue that you have with:

RUN-050406: |Session STG_IMS_MAIN_LOAD_Job|Work flow WF_IMS_MAIN_DW_LOAD_DUMMY|Work flow WF_DW_INVENTORY_IMS|Work flow WF_DW_INVENTORY_IMS_1|Work flow WF_DW_INV_PRG_LD_IMS_Mod|Data flow ORDR_ReUse_DF_DW_INV_PRG_LD_IMS_Mod1

Data flow <ORDR_ReUse_DF_DW_INV_PRG_LD_IMS_Mod1> received a bad system message. Message text from the child process is

<a ==========================================================

Collect the following and send to Customer Support:

1. Log files(error_*, monitor_*, trace_*) associated with this failed job.

2. Exported ATL file of this failed job.

3. DDL statements of tables referenced in this failed job.

4. Data to populate the tables referenced in the failed job. If not possible, get the last few rows (or sample of them) when

the job failed.

5. Core dump, if any, generated from this failed job.

==========================================================>. The process executing data flow

<ORDR_ReUse_DF_DW_INV_PRG_LD_IMS_Mod1> has died abnormally. For NT, check errorlog.txt. For HPUX, check stack_trace.txt.

We have done the upgrade from DS 4.0 to DS 4.2 and this error has started. The answer in the other post was that it is a bug identified by SAP, but have you found any work around?

My customer has their jobs stopped because we have this issue, and it is critical to have it solved asap.

I will appreciate any comment or suggestion,

Regards,

[Former Member]

Hello,

The work around for now is that you run the dataflows in In-memory cache.

Whichever dataflows are giving error, right click on them --> go to properties --> change the cache to In-memory & DOP (degree of parallelism to 1).

Even after doing so if it gives error then split your dataflow which is giving error into 2 or more parts (depending on how much memory intensive operation it have inside like table comparison, order by, group by , lookup etc) one of my dataflow was so huge that I have to split it in 6 dataflows & then run it all with in-memory as cache option & DOP=1.

Hope this helps.

[Former Member]

Splitting the data flow where the job is failing abnormally, resolved the issue for me.