TIME_OUT error while executing an ABAP report

[Former Member]

Hi All,

I have an ABAP report in which the code looks like below

LOOP AT pernr_tab . " *Internal table consists of  around 25k records*  
   IF last_run_date IS INITIAL. " Initial Load
      PERFORM fill_data USING pernr_tab-pernr last_run_date CHANGING lv_flag.
      lv_flag = 'x'.
      initial_count = initial_count + 1.
    ELSE.
      PERFORM fill_data USING pernr_tab-pernr last_run_date CHANGING lv_flag.
      IF pernr_tab-status EQ '3' AND pernr_tab-begda GE today. " Future Hire records during delta load
        lv_flag = 'x'.
      ENDIF.
      IF lv_flag EQ 'x'.
        delta_change_count = delta_change_count + 1.
      ENDIF.
    ENDIF.
    IF lv_flag = 'x'. "initial load or changes in some infotype for delta load
      PERFORM append_data.
      CLEAR: p0000_tab[],p0000,p0001_tab[],p0001,p0002_tab[],p0002,p0006_tab[],p0006,p0006_phone,
p0007_tab[],p0007,p0027_tab[],p0027,p0032_tab[],p0032,p0105_tab_usr[],p0041_tab[],p0041.
      CLEAR: p0105_usr,p0105_tab_ven[],p0105_ven,p0105_tab_mgr[],p0105_mgr,p0709_tab[],p0709,pos_id,
functional_title,p0521_tab[],p0521,p2001_tab[],p2001,corporate_band_code,corporate_title_code,wa_pernr_tab.
    ENDIF.
  ENDLOOP.

inside the loop i am fetching data from 10 different infotypes and adding into another internal table.When I execute this program it is giving TIME_OUT error.Is there any way i can resolve this issue(I dont want to run this program in the background)

Thanks

Bala Duvvuri

Edited by: Bala Duvvuri on Jul 13, 2010 8:25 AM

Edited by: Thomas Zloch on Jul 13, 2010 4:58 PM - line width reduced

[Former Member]

You will have to paste the code from the routines FILL_DATA and APPEND_DATA for analysis.

[Former Member]

fill_data

reading 10 infotypes

call function 'HR_READ_INFOTYPE'
  exporting
*     TCLAS                 = 'A'
    pernr                 = p_pernr_tab_pernr
    infty                 = '0000'
   begda                 = '18000101'
   endda                 = '99991231'
*   bypass_buffer         = 'X'
  tables
    infty_tab             = p0000_tab
 exceptions
   infty_not_found       = 1
   others                = 2
          .
  if sy-subrc <> 0.
  endif.
  if p0000_tab[] is not initial.
    sort p0000_tab by endda  descending.
    read table p0000_tab into p0000 index 1.
    if p0000-aedtm ge last_run_date.
      lv_flag = 'X' ." Infotype 0 has been changed
    endif.
  endif.

append_data

assigning 50 different fields from above infotypes to ls_emp_data

ls_emp_data-functional_title = functional_title.
  ls_emp_data-last_update = today.
  if ( p2001-awart = '5006' and p2001-begda <= today and p2001-endda >= today ) or
 ( p0521-atzph = 'AB' AND p0521-begda <= today AND p0521-endda >= today ).
    ls_emp_data-absent = text-007.
  ELSE.
    ls_emp_data-absent = text-008.
  ENDIF.
  ls_emp_data-corporate_band_code = corporate_band_code.
  ls_emp_data-corporate_title_code = corporate_title_code.
  APPEND ls_emp_data TO lt_emp_data.
  CLEAR:ls_emp_data,lv_flag.

Thanks

Bala Duvvuri

Edited by: Thomas Zloch on Jul 14, 2010 2:48 PM - code tags added

[brad_bohn]

That sounds like alot of data to process in the foreground and I don't know much about HR. I would suspect there's a performance limiter in that function call which could be tracked with a trace. If there isn't and you absolutely have to run the thing in the foreground, you can always bypass the TIME_OUT with a strategically placed commit work. Not that I recommend it as a performance workaround; I'm just saying it's an option .

[Former Member]

I followed this thread and it got solved

Thanks

Bala Duvvuri

[Former Member]

25000 sorts in the LOOP can easily kill you, and of cours the Call function coding is still missing

>sort p0000_tab by endda descending.

A code sicussion will not help you, you must analyse the execution! There is definitely nonlinear coding, you must figure out where.

The following 3 blogs can help you:

Nonlinearity: The problem and background

/people/siegfried.boes/blog/2007/02/12/performance-problems-caused-by-nonlinear-coding

Z_SE30_COMPARE

/people/siegfried.boes/blog/2008/01/15/a-tool-to-compare-runtime-measurements-zse30compare

Nonlinearity Check

/people/siegfried.boes/blog/2008/01/24/nonlinearity-check-using-the-zse30compare

You must solve the problem otherwise it will even run in background much too long. Whether it fits in a dialog step depends on what is done and how.

Siegfried