Disk IO issue - Longer Job run time

[ray.wurlod]

Take a look at the Resource Estimation tool (and, possibly, Performance Analysis tool) in DataStage Designer.

Have you implemented DataStage Operations Console (which was new in version 8.7)? IF so, that would show you much more useful information.

[Developer9]

The fork lookup is a potential problem with buffering and deadlocks.

@Mike ,

IBM Replied with link for $TEMP environment variable

http://www-01.ibm.com/support/docview.w ... wg21595254

This is morning we experienced one additional job delay in the same sequence.


Job1 :(As topic of discussion )
6/30/2016 7:07:21 AM main_program: APT configuration file: /appl/infoserver/Server/Configurations/2node_Admin.apt
Delay 1
6/30/2016 7:13:13 AM main_program: This step has 8 datasets:It runs 10 processes on 2 nodes.


6/30/2016 7:13:24 AM seq_import_data,0: Import complete; 69375 records imported successfully, 0 rejected. -->Data read
Delay 2
6/30/2016 7:22:34 AM lkup_import_data,0: Ignoring duplicate entry at table record 2; no further warnings will be issued for this table

Job 2 (New job with issue)

seq -->xfm (3 links) --> funnel (filter and lookup as inputs) -->xfm -->dataset

6/30/2016 7:22:56 AM fnl_import_data: When checking operator: Defaulting "LastUpdate" in transfer from "inRec" to "outRec".
Delay 3
6/30/2016 7:36:03 AMfnl_import_data: When checking operator: Defaulting "LastUpdate" in transfer from "inRec" to "outRec".


Regarding Replacing LOOKUP stage with join stage --I am working on the design changes ..

Please let me know your ideas

[Developer9]

@ ray,

I was able to verify that information

CPU -Avg -20 %
Memory -Physical/Virtual Free -35/50 GB
Disk Space --Nothing Monitored

Still waiting on IBM feedback on disk i/o issue

Thank you

[Developer9]

Hi Everyone, We got some advise from IBM to include environment variable of APT_STARTUP_STATUS=1 for better idea on why the job is taking so long to startup.

Does this approach helps in the root cause of the issue. please provide me your input

Thanks,
DS user

[bharathappriyan]

Hi,

I faced this problem couple of years back. It was due to job log. The job was called in loop(min 400 times). It started running slowly whenever there was a failure.

Cleaning up the job logs solved the issue.

Thanks.
Bharathappriyan