SQL query Taking long time to execute










-2















Im trying to execute the below query and it is taking long time to execute. Is there anyway i can make changes in the below query? It is taking more than 60mins for executing and sometimes it is getting hanged.



select distinct z.por_s as Portfolio_Number,z.lse_s as 
Account_Number,y.les_s as Lessee_Number,z.unit_S as Asset,to_char(x.d_com_s,
'MM/DD/YYYY') as Start_Date, to_char(x.d_mat_s,'MM/DD/YYYY') as
Maturity_Date, x.p_con_d as Yield,k.nam_long_s as Venodr_Name,z.dlr_s as
Venodr_Number,y.client_s_name_s as Customer_Name,y.short_add_s as
Billing_Address,y.short_cty_s as City,y.st_s as State,z.eq_add_s as
Equipment_Address, x.trm_l as Original_Term,(x.TRM_L - x.PMT_L) as
Remaining_Term,z.id_S as EquipmentSerialID,z.org_d as Original_Amount,
(NVL(x.lsd_unit_d, 0.0) - NVL(x.acl_depr_d, 0.0) - NVL(x.sec_d, 0.0) -
NVL(SUM(b.bal_d), 0.0) + NVL(SUM(NVL(a.pamt_b_d, 0.0) + NVL(a.iamt_b_d, 0.0)
- NVL(a.pamt_p_d, 0.0) - NVL(a.iamt_p_d, 0.0)), 0.0) - NVL(x.op_bill_adj_d,
0.0) + NVL(x.unrec_in_d, 0.0)) as OPR_Remaining_Net_Investment,
(NVL(x.rec_d, 0.0) + NVL(SUM(NVL(d.aamt_b_d, 0.0) - NVL(d.aamt_p_d, 0.0)),
0.0) + NVL(SUM(NVL(a.pamt_b_d, 0.0) + NVL(a.iamt_b_d, 0.0) -
NVL(a.pamt_p_d, 0.0) - NVL(a.iamt_p_d, 0.0)), 0.0) + NVL(x.tot_res_d, 0.0) -
NVL(x.res_guar_d, 0.0) - NVL(x.l_ue_d, 0.0) + NVL(x.r_ue_d, 0.0) +
NVL(x.i_ue_d, 0.0) + NVL(x.d_ue_d, 0.0) + NVL(x.l_in_ptd_d, 0.0) +
NVL(x.r_in_ptd_d, 0.0) + NVL(x.i_in_ptd_d, 0.0) + NVL(x.d_in_ptd_d, 0.0) -
NVL(x.sec_d, 0.0) - NVL(SUM(b.bal_d), 0.0) - NVL(x.susp_l_acc_d, 0.0) +
NVL(x.susp_r_acc_d, 0.0) + NVL(x.susp_d_acc_d, 0.0) + NVL(x.susp_i_acc_d,
0.0)) as PRE_Remaining_Net_Investment
,z.res_d as Residual_Asset ,n.bus_phn_s as Business_Phone, n.hom_phn_s as
Home_phone,o.field19_s as Credit_Guarantee, o.field20_s as
Residual_Guarantee,m.add_s as Guarantor_Address,m.add2_s as
Guarantor_Address2,m.cty_s as Guarantor_City,m.st_s as
Guarantor_State,m.zip_s as Guarantor_Zip
from rgc c
FULL join rls x on c.APP_LSE_S = x.LSE_S
FULL Join rar a on x.LSE_S = a.LSE_S
FULL Join rsu b on x.LSE_S = b.LSE_S
FULL Join rarb d on x.por_s = d.por_s
FULL join vrcx_les y on x.les_s = y.les_s
FULL join req z on x.LSE_S = z.LSE_S
FULL join rlr k on z.dlr_s = k.dlr_s
FULL join ral n on x.les_s = n.les_s
FULL join rug o on x.lse_s = o.lse_s
FULL JOIN RGR m ON c.CLIENT_ID_S = m.GRN_S
where x.act_s in ('AACQ', 'AASM', 'AEXM', 'ANOR', 'ARBK', 'ARED', 'AREN',
'AXFR') and x.d_mat_s > to_date ('07/01/2018', 'MM/DD/YYYY')
and a.por_s = x.por_s
AND a.com_s = x.com_s
AND a.reg_s = x.reg_s
AND a.off_s = x.off_s
AND a.lse_s = x.lse_s
AND d.invo_s = a.invo_s
AND d.a_typ_s in ('PROR', 'IDR2', 'SECD')
AND NVL(a.income_accrued_c, ' ') <> 'N'
group by z.por_s,z.lse_s,y.les_s,z.unit_S,x.d_com_s,
x.d_mat_s,x.p_con_d,k.nam_long_s,
z.dlr_s,y.client_s_name_s,y.short_add_s,
y.short_cty_s,y.st_s,z.eq_add_s,x.trm_l,x.PMT_L,
z.id_S,z.org_d,x.lsd_unit_d,x.acl_depr_d,x.sec_d,
b.bal_d,a.pamt_b_d,a.iamt_b_d,a.pamt_p_d,a.iamt_p_d,
x.op_bill_adj_d,x.unrec_in_d,x.rec_d,x.res_guar_d,
d.aamt_b_d,d.aamt_p_d,a.pamt_b_d,a.iamt_b_d,a.pamt_p_d,
a.iamt_p_d,x.tot_res_d,x.res_guar_d,x.l_ue_d,x.r_ue_d,
x.i_ue_d,x.d_ue_d,x.l_in_ptd_d,x.r_in_ptd_d,x.i_in_ptd_d,
x.d_in_ptd_d,x.susp_l_acc_d,x.susp_r_acc_d,x.susp_d_acc_d,
x.susp_i_acc_d,z.res_d,n.bus_phn_s,n.hom_phn_s,o.field19_s,
o.field20_s,m.add_s,m.add2_s,m.cty_s,m.st_s,m.zip_s;


Any changes needs to made on this query just help me out on this.










share|improve this question



















  • 2





    You are doing FULL OUTER JOIN on eleven tables: why is that? You appear to have some join criteria in the WHERE clause, so probably those joins will produce the same result set as INNER JOINs. So maybe it doesn't make any difference. But a query which is expressive of intent is easier to understand and diagnose.

    – APC
    Nov 13 '18 at 18:42






  • 2





    Beyond that, we have to ask the usual query tuning questions. How many rows in each table? What percentage of the rows will be returned? Which columns are indexed? Is the data distribution skewed? How fresh are the table stats? At the very least you should post an Explain Plan

    – APC
    Nov 13 '18 at 18:46






  • 1





    first thing is to run an explain plan.

    – OldProgrammer
    Nov 13 '18 at 19:43












  • Or even better, a SQL Monitor report.

    – BobC
    Nov 22 '18 at 4:08















-2















Im trying to execute the below query and it is taking long time to execute. Is there anyway i can make changes in the below query? It is taking more than 60mins for executing and sometimes it is getting hanged.



select distinct z.por_s as Portfolio_Number,z.lse_s as 
Account_Number,y.les_s as Lessee_Number,z.unit_S as Asset,to_char(x.d_com_s,
'MM/DD/YYYY') as Start_Date, to_char(x.d_mat_s,'MM/DD/YYYY') as
Maturity_Date, x.p_con_d as Yield,k.nam_long_s as Venodr_Name,z.dlr_s as
Venodr_Number,y.client_s_name_s as Customer_Name,y.short_add_s as
Billing_Address,y.short_cty_s as City,y.st_s as State,z.eq_add_s as
Equipment_Address, x.trm_l as Original_Term,(x.TRM_L - x.PMT_L) as
Remaining_Term,z.id_S as EquipmentSerialID,z.org_d as Original_Amount,
(NVL(x.lsd_unit_d, 0.0) - NVL(x.acl_depr_d, 0.0) - NVL(x.sec_d, 0.0) -
NVL(SUM(b.bal_d), 0.0) + NVL(SUM(NVL(a.pamt_b_d, 0.0) + NVL(a.iamt_b_d, 0.0)
- NVL(a.pamt_p_d, 0.0) - NVL(a.iamt_p_d, 0.0)), 0.0) - NVL(x.op_bill_adj_d,
0.0) + NVL(x.unrec_in_d, 0.0)) as OPR_Remaining_Net_Investment,
(NVL(x.rec_d, 0.0) + NVL(SUM(NVL(d.aamt_b_d, 0.0) - NVL(d.aamt_p_d, 0.0)),
0.0) + NVL(SUM(NVL(a.pamt_b_d, 0.0) + NVL(a.iamt_b_d, 0.0) -
NVL(a.pamt_p_d, 0.0) - NVL(a.iamt_p_d, 0.0)), 0.0) + NVL(x.tot_res_d, 0.0) -
NVL(x.res_guar_d, 0.0) - NVL(x.l_ue_d, 0.0) + NVL(x.r_ue_d, 0.0) +
NVL(x.i_ue_d, 0.0) + NVL(x.d_ue_d, 0.0) + NVL(x.l_in_ptd_d, 0.0) +
NVL(x.r_in_ptd_d, 0.0) + NVL(x.i_in_ptd_d, 0.0) + NVL(x.d_in_ptd_d, 0.0) -
NVL(x.sec_d, 0.0) - NVL(SUM(b.bal_d), 0.0) - NVL(x.susp_l_acc_d, 0.0) +
NVL(x.susp_r_acc_d, 0.0) + NVL(x.susp_d_acc_d, 0.0) + NVL(x.susp_i_acc_d,
0.0)) as PRE_Remaining_Net_Investment
,z.res_d as Residual_Asset ,n.bus_phn_s as Business_Phone, n.hom_phn_s as
Home_phone,o.field19_s as Credit_Guarantee, o.field20_s as
Residual_Guarantee,m.add_s as Guarantor_Address,m.add2_s as
Guarantor_Address2,m.cty_s as Guarantor_City,m.st_s as
Guarantor_State,m.zip_s as Guarantor_Zip
from rgc c
FULL join rls x on c.APP_LSE_S = x.LSE_S
FULL Join rar a on x.LSE_S = a.LSE_S
FULL Join rsu b on x.LSE_S = b.LSE_S
FULL Join rarb d on x.por_s = d.por_s
FULL join vrcx_les y on x.les_s = y.les_s
FULL join req z on x.LSE_S = z.LSE_S
FULL join rlr k on z.dlr_s = k.dlr_s
FULL join ral n on x.les_s = n.les_s
FULL join rug o on x.lse_s = o.lse_s
FULL JOIN RGR m ON c.CLIENT_ID_S = m.GRN_S
where x.act_s in ('AACQ', 'AASM', 'AEXM', 'ANOR', 'ARBK', 'ARED', 'AREN',
'AXFR') and x.d_mat_s > to_date ('07/01/2018', 'MM/DD/YYYY')
and a.por_s = x.por_s
AND a.com_s = x.com_s
AND a.reg_s = x.reg_s
AND a.off_s = x.off_s
AND a.lse_s = x.lse_s
AND d.invo_s = a.invo_s
AND d.a_typ_s in ('PROR', 'IDR2', 'SECD')
AND NVL(a.income_accrued_c, ' ') <> 'N'
group by z.por_s,z.lse_s,y.les_s,z.unit_S,x.d_com_s,
x.d_mat_s,x.p_con_d,k.nam_long_s,
z.dlr_s,y.client_s_name_s,y.short_add_s,
y.short_cty_s,y.st_s,z.eq_add_s,x.trm_l,x.PMT_L,
z.id_S,z.org_d,x.lsd_unit_d,x.acl_depr_d,x.sec_d,
b.bal_d,a.pamt_b_d,a.iamt_b_d,a.pamt_p_d,a.iamt_p_d,
x.op_bill_adj_d,x.unrec_in_d,x.rec_d,x.res_guar_d,
d.aamt_b_d,d.aamt_p_d,a.pamt_b_d,a.iamt_b_d,a.pamt_p_d,
a.iamt_p_d,x.tot_res_d,x.res_guar_d,x.l_ue_d,x.r_ue_d,
x.i_ue_d,x.d_ue_d,x.l_in_ptd_d,x.r_in_ptd_d,x.i_in_ptd_d,
x.d_in_ptd_d,x.susp_l_acc_d,x.susp_r_acc_d,x.susp_d_acc_d,
x.susp_i_acc_d,z.res_d,n.bus_phn_s,n.hom_phn_s,o.field19_s,
o.field20_s,m.add_s,m.add2_s,m.cty_s,m.st_s,m.zip_s;


Any changes needs to made on this query just help me out on this.










share|improve this question



















  • 2





    You are doing FULL OUTER JOIN on eleven tables: why is that? You appear to have some join criteria in the WHERE clause, so probably those joins will produce the same result set as INNER JOINs. So maybe it doesn't make any difference. But a query which is expressive of intent is easier to understand and diagnose.

    – APC
    Nov 13 '18 at 18:42






  • 2





    Beyond that, we have to ask the usual query tuning questions. How many rows in each table? What percentage of the rows will be returned? Which columns are indexed? Is the data distribution skewed? How fresh are the table stats? At the very least you should post an Explain Plan

    – APC
    Nov 13 '18 at 18:46






  • 1





    first thing is to run an explain plan.

    – OldProgrammer
    Nov 13 '18 at 19:43












  • Or even better, a SQL Monitor report.

    – BobC
    Nov 22 '18 at 4:08













-2












-2








-2








Im trying to execute the below query and it is taking long time to execute. Is there anyway i can make changes in the below query? It is taking more than 60mins for executing and sometimes it is getting hanged.



select distinct z.por_s as Portfolio_Number,z.lse_s as 
Account_Number,y.les_s as Lessee_Number,z.unit_S as Asset,to_char(x.d_com_s,
'MM/DD/YYYY') as Start_Date, to_char(x.d_mat_s,'MM/DD/YYYY') as
Maturity_Date, x.p_con_d as Yield,k.nam_long_s as Venodr_Name,z.dlr_s as
Venodr_Number,y.client_s_name_s as Customer_Name,y.short_add_s as
Billing_Address,y.short_cty_s as City,y.st_s as State,z.eq_add_s as
Equipment_Address, x.trm_l as Original_Term,(x.TRM_L - x.PMT_L) as
Remaining_Term,z.id_S as EquipmentSerialID,z.org_d as Original_Amount,
(NVL(x.lsd_unit_d, 0.0) - NVL(x.acl_depr_d, 0.0) - NVL(x.sec_d, 0.0) -
NVL(SUM(b.bal_d), 0.0) + NVL(SUM(NVL(a.pamt_b_d, 0.0) + NVL(a.iamt_b_d, 0.0)
- NVL(a.pamt_p_d, 0.0) - NVL(a.iamt_p_d, 0.0)), 0.0) - NVL(x.op_bill_adj_d,
0.0) + NVL(x.unrec_in_d, 0.0)) as OPR_Remaining_Net_Investment,
(NVL(x.rec_d, 0.0) + NVL(SUM(NVL(d.aamt_b_d, 0.0) - NVL(d.aamt_p_d, 0.0)),
0.0) + NVL(SUM(NVL(a.pamt_b_d, 0.0) + NVL(a.iamt_b_d, 0.0) -
NVL(a.pamt_p_d, 0.0) - NVL(a.iamt_p_d, 0.0)), 0.0) + NVL(x.tot_res_d, 0.0) -
NVL(x.res_guar_d, 0.0) - NVL(x.l_ue_d, 0.0) + NVL(x.r_ue_d, 0.0) +
NVL(x.i_ue_d, 0.0) + NVL(x.d_ue_d, 0.0) + NVL(x.l_in_ptd_d, 0.0) +
NVL(x.r_in_ptd_d, 0.0) + NVL(x.i_in_ptd_d, 0.0) + NVL(x.d_in_ptd_d, 0.0) -
NVL(x.sec_d, 0.0) - NVL(SUM(b.bal_d), 0.0) - NVL(x.susp_l_acc_d, 0.0) +
NVL(x.susp_r_acc_d, 0.0) + NVL(x.susp_d_acc_d, 0.0) + NVL(x.susp_i_acc_d,
0.0)) as PRE_Remaining_Net_Investment
,z.res_d as Residual_Asset ,n.bus_phn_s as Business_Phone, n.hom_phn_s as
Home_phone,o.field19_s as Credit_Guarantee, o.field20_s as
Residual_Guarantee,m.add_s as Guarantor_Address,m.add2_s as
Guarantor_Address2,m.cty_s as Guarantor_City,m.st_s as
Guarantor_State,m.zip_s as Guarantor_Zip
from rgc c
FULL join rls x on c.APP_LSE_S = x.LSE_S
FULL Join rar a on x.LSE_S = a.LSE_S
FULL Join rsu b on x.LSE_S = b.LSE_S
FULL Join rarb d on x.por_s = d.por_s
FULL join vrcx_les y on x.les_s = y.les_s
FULL join req z on x.LSE_S = z.LSE_S
FULL join rlr k on z.dlr_s = k.dlr_s
FULL join ral n on x.les_s = n.les_s
FULL join rug o on x.lse_s = o.lse_s
FULL JOIN RGR m ON c.CLIENT_ID_S = m.GRN_S
where x.act_s in ('AACQ', 'AASM', 'AEXM', 'ANOR', 'ARBK', 'ARED', 'AREN',
'AXFR') and x.d_mat_s > to_date ('07/01/2018', 'MM/DD/YYYY')
and a.por_s = x.por_s
AND a.com_s = x.com_s
AND a.reg_s = x.reg_s
AND a.off_s = x.off_s
AND a.lse_s = x.lse_s
AND d.invo_s = a.invo_s
AND d.a_typ_s in ('PROR', 'IDR2', 'SECD')
AND NVL(a.income_accrued_c, ' ') <> 'N'
group by z.por_s,z.lse_s,y.les_s,z.unit_S,x.d_com_s,
x.d_mat_s,x.p_con_d,k.nam_long_s,
z.dlr_s,y.client_s_name_s,y.short_add_s,
y.short_cty_s,y.st_s,z.eq_add_s,x.trm_l,x.PMT_L,
z.id_S,z.org_d,x.lsd_unit_d,x.acl_depr_d,x.sec_d,
b.bal_d,a.pamt_b_d,a.iamt_b_d,a.pamt_p_d,a.iamt_p_d,
x.op_bill_adj_d,x.unrec_in_d,x.rec_d,x.res_guar_d,
d.aamt_b_d,d.aamt_p_d,a.pamt_b_d,a.iamt_b_d,a.pamt_p_d,
a.iamt_p_d,x.tot_res_d,x.res_guar_d,x.l_ue_d,x.r_ue_d,
x.i_ue_d,x.d_ue_d,x.l_in_ptd_d,x.r_in_ptd_d,x.i_in_ptd_d,
x.d_in_ptd_d,x.susp_l_acc_d,x.susp_r_acc_d,x.susp_d_acc_d,
x.susp_i_acc_d,z.res_d,n.bus_phn_s,n.hom_phn_s,o.field19_s,
o.field20_s,m.add_s,m.add2_s,m.cty_s,m.st_s,m.zip_s;


Any changes needs to made on this query just help me out on this.










share|improve this question
















Im trying to execute the below query and it is taking long time to execute. Is there anyway i can make changes in the below query? It is taking more than 60mins for executing and sometimes it is getting hanged.



select distinct z.por_s as Portfolio_Number,z.lse_s as 
Account_Number,y.les_s as Lessee_Number,z.unit_S as Asset,to_char(x.d_com_s,
'MM/DD/YYYY') as Start_Date, to_char(x.d_mat_s,'MM/DD/YYYY') as
Maturity_Date, x.p_con_d as Yield,k.nam_long_s as Venodr_Name,z.dlr_s as
Venodr_Number,y.client_s_name_s as Customer_Name,y.short_add_s as
Billing_Address,y.short_cty_s as City,y.st_s as State,z.eq_add_s as
Equipment_Address, x.trm_l as Original_Term,(x.TRM_L - x.PMT_L) as
Remaining_Term,z.id_S as EquipmentSerialID,z.org_d as Original_Amount,
(NVL(x.lsd_unit_d, 0.0) - NVL(x.acl_depr_d, 0.0) - NVL(x.sec_d, 0.0) -
NVL(SUM(b.bal_d), 0.0) + NVL(SUM(NVL(a.pamt_b_d, 0.0) + NVL(a.iamt_b_d, 0.0)
- NVL(a.pamt_p_d, 0.0) - NVL(a.iamt_p_d, 0.0)), 0.0) - NVL(x.op_bill_adj_d,
0.0) + NVL(x.unrec_in_d, 0.0)) as OPR_Remaining_Net_Investment,
(NVL(x.rec_d, 0.0) + NVL(SUM(NVL(d.aamt_b_d, 0.0) - NVL(d.aamt_p_d, 0.0)),
0.0) + NVL(SUM(NVL(a.pamt_b_d, 0.0) + NVL(a.iamt_b_d, 0.0) -
NVL(a.pamt_p_d, 0.0) - NVL(a.iamt_p_d, 0.0)), 0.0) + NVL(x.tot_res_d, 0.0) -
NVL(x.res_guar_d, 0.0) - NVL(x.l_ue_d, 0.0) + NVL(x.r_ue_d, 0.0) +
NVL(x.i_ue_d, 0.0) + NVL(x.d_ue_d, 0.0) + NVL(x.l_in_ptd_d, 0.0) +
NVL(x.r_in_ptd_d, 0.0) + NVL(x.i_in_ptd_d, 0.0) + NVL(x.d_in_ptd_d, 0.0) -
NVL(x.sec_d, 0.0) - NVL(SUM(b.bal_d), 0.0) - NVL(x.susp_l_acc_d, 0.0) +
NVL(x.susp_r_acc_d, 0.0) + NVL(x.susp_d_acc_d, 0.0) + NVL(x.susp_i_acc_d,
0.0)) as PRE_Remaining_Net_Investment
,z.res_d as Residual_Asset ,n.bus_phn_s as Business_Phone, n.hom_phn_s as
Home_phone,o.field19_s as Credit_Guarantee, o.field20_s as
Residual_Guarantee,m.add_s as Guarantor_Address,m.add2_s as
Guarantor_Address2,m.cty_s as Guarantor_City,m.st_s as
Guarantor_State,m.zip_s as Guarantor_Zip
from rgc c
FULL join rls x on c.APP_LSE_S = x.LSE_S
FULL Join rar a on x.LSE_S = a.LSE_S
FULL Join rsu b on x.LSE_S = b.LSE_S
FULL Join rarb d on x.por_s = d.por_s
FULL join vrcx_les y on x.les_s = y.les_s
FULL join req z on x.LSE_S = z.LSE_S
FULL join rlr k on z.dlr_s = k.dlr_s
FULL join ral n on x.les_s = n.les_s
FULL join rug o on x.lse_s = o.lse_s
FULL JOIN RGR m ON c.CLIENT_ID_S = m.GRN_S
where x.act_s in ('AACQ', 'AASM', 'AEXM', 'ANOR', 'ARBK', 'ARED', 'AREN',
'AXFR') and x.d_mat_s > to_date ('07/01/2018', 'MM/DD/YYYY')
and a.por_s = x.por_s
AND a.com_s = x.com_s
AND a.reg_s = x.reg_s
AND a.off_s = x.off_s
AND a.lse_s = x.lse_s
AND d.invo_s = a.invo_s
AND d.a_typ_s in ('PROR', 'IDR2', 'SECD')
AND NVL(a.income_accrued_c, ' ') <> 'N'
group by z.por_s,z.lse_s,y.les_s,z.unit_S,x.d_com_s,
x.d_mat_s,x.p_con_d,k.nam_long_s,
z.dlr_s,y.client_s_name_s,y.short_add_s,
y.short_cty_s,y.st_s,z.eq_add_s,x.trm_l,x.PMT_L,
z.id_S,z.org_d,x.lsd_unit_d,x.acl_depr_d,x.sec_d,
b.bal_d,a.pamt_b_d,a.iamt_b_d,a.pamt_p_d,a.iamt_p_d,
x.op_bill_adj_d,x.unrec_in_d,x.rec_d,x.res_guar_d,
d.aamt_b_d,d.aamt_p_d,a.pamt_b_d,a.iamt_b_d,a.pamt_p_d,
a.iamt_p_d,x.tot_res_d,x.res_guar_d,x.l_ue_d,x.r_ue_d,
x.i_ue_d,x.d_ue_d,x.l_in_ptd_d,x.r_in_ptd_d,x.i_in_ptd_d,
x.d_in_ptd_d,x.susp_l_acc_d,x.susp_r_acc_d,x.susp_d_acc_d,
x.susp_i_acc_d,z.res_d,n.bus_phn_s,n.hom_phn_s,o.field19_s,
o.field20_s,m.add_s,m.add2_s,m.cty_s,m.st_s,m.zip_s;


Any changes needs to made on this query just help me out on this.







sql oracle performance oracle11g query-optimization






share|improve this question















share|improve this question













share|improve this question




share|improve this question








edited Nov 13 '18 at 18:47









APC

119k15118229




119k15118229










asked Nov 13 '18 at 17:51









VikasVikas

11




11







  • 2





    You are doing FULL OUTER JOIN on eleven tables: why is that? You appear to have some join criteria in the WHERE clause, so probably those joins will produce the same result set as INNER JOINs. So maybe it doesn't make any difference. But a query which is expressive of intent is easier to understand and diagnose.

    – APC
    Nov 13 '18 at 18:42






  • 2





    Beyond that, we have to ask the usual query tuning questions. How many rows in each table? What percentage of the rows will be returned? Which columns are indexed? Is the data distribution skewed? How fresh are the table stats? At the very least you should post an Explain Plan

    – APC
    Nov 13 '18 at 18:46






  • 1





    first thing is to run an explain plan.

    – OldProgrammer
    Nov 13 '18 at 19:43












  • Or even better, a SQL Monitor report.

    – BobC
    Nov 22 '18 at 4:08












  • 2





    You are doing FULL OUTER JOIN on eleven tables: why is that? You appear to have some join criteria in the WHERE clause, so probably those joins will produce the same result set as INNER JOINs. So maybe it doesn't make any difference. But a query which is expressive of intent is easier to understand and diagnose.

    – APC
    Nov 13 '18 at 18:42






  • 2





    Beyond that, we have to ask the usual query tuning questions. How many rows in each table? What percentage of the rows will be returned? Which columns are indexed? Is the data distribution skewed? How fresh are the table stats? At the very least you should post an Explain Plan

    – APC
    Nov 13 '18 at 18:46






  • 1





    first thing is to run an explain plan.

    – OldProgrammer
    Nov 13 '18 at 19:43












  • Or even better, a SQL Monitor report.

    – BobC
    Nov 22 '18 at 4:08







2




2





You are doing FULL OUTER JOIN on eleven tables: why is that? You appear to have some join criteria in the WHERE clause, so probably those joins will produce the same result set as INNER JOINs. So maybe it doesn't make any difference. But a query which is expressive of intent is easier to understand and diagnose.

– APC
Nov 13 '18 at 18:42





You are doing FULL OUTER JOIN on eleven tables: why is that? You appear to have some join criteria in the WHERE clause, so probably those joins will produce the same result set as INNER JOINs. So maybe it doesn't make any difference. But a query which is expressive of intent is easier to understand and diagnose.

– APC
Nov 13 '18 at 18:42




2




2





Beyond that, we have to ask the usual query tuning questions. How many rows in each table? What percentage of the rows will be returned? Which columns are indexed? Is the data distribution skewed? How fresh are the table stats? At the very least you should post an Explain Plan

– APC
Nov 13 '18 at 18:46





Beyond that, we have to ask the usual query tuning questions. How many rows in each table? What percentage of the rows will be returned? Which columns are indexed? Is the data distribution skewed? How fresh are the table stats? At the very least you should post an Explain Plan

– APC
Nov 13 '18 at 18:46




1




1





first thing is to run an explain plan.

– OldProgrammer
Nov 13 '18 at 19:43






first thing is to run an explain plan.

– OldProgrammer
Nov 13 '18 at 19:43














Or even better, a SQL Monitor report.

– BobC
Nov 22 '18 at 4:08





Or even better, a SQL Monitor report.

– BobC
Nov 22 '18 at 4:08












1 Answer
1






active

oldest

votes


















0














 select distinct z.por_s as Portfolio_Number,
z.lse_s as Account_Number,
...

from rgc c
FULL join rls x on c.APP_LSE_S = x.LSE_S #match nro account (1)
FULL Join rar a on x.LSE_S = a.LSE_S #match nro account (2)
FULL Join rsu b on x.LSE_S = b.LSE_S #match nro account (3)
FULL Join rarb d on x.por_s = d.por_s #match (4)
FULL join vrcx_les y on x.les_s = y.les_s #match nro less
FULL join req z on x.LSE_S = z.LSE_S #match nro account

FULL join rlr k on z.dlr_s = k.dlr_s

FULL join ral n on x.les_s = n.les_s #match nro less
FULL join rug o on x.lse_s = o.lse_s #match nro account

FULL JOIN RGR m ON c.CLIENT_ID_S = m.GRN_S
where x.act_s in
('AACQ', 'AASM', 'AEXM', 'ANOR', 'ARBK', 'ARED', 'AREN', 'AXFR') #use this in the full join (2)
and x.d_mat_s > to_date ('07/01/2018', 'MM/DD/YYYY') #use this in the full join (2)
and a.por_s = x.por_s #use this in the full join (2)
AND a.com_s = x.com_s #use this in the full join (2)
AND a.reg_s = x.reg_s #use this in the full join (2)
AND a.off_s = x.off_s #use this in the full join (2)
AND a.lse_s = x.lse_s #use this in the full join (2)
AND d.invo_s = a.invo_s #use this in the full join (4)
AND d.a_typ_s in ('PROR', 'IDR2', 'SECD')
AND NVL(a.income_accrued_c, ' ') <> 'N' #nvl is not necesary
group by z.por_s,z.lse_s,y.les_s,z.unit_S,x.d_com_s,
...


I would put the restrictions in the joins, so you dont combine all the rows (1000 x 1000 for example), just the rows you need (100 x 1000) Like this:



 from rgc c 
FULL join rls x on c.APP_LSE_S = x.LSE_S #match nro account (1)
FULL Join rar a on x.LSE_S = a.LSE_S #match nro account (2)
and x.act_s in
('AACQ', 'AASM', 'AEXM', 'ANOR', 'ARBK', 'ARED', 'AREN', 'AXFR') #use this in the full join (2)
and x.d_mat_s > to_date ('07/01/2018', 'MM/DD/YYYY') #use this in the full join (2)
and a.por_s = x.por_s #use this in the full join (2)
AND a.com_s = x.com_s #use this in the full join (2)
AND a.reg_s = x.reg_s #use this in the full join (2)
AND a.off_s = x.off_s #use this in the full join (2)
AND a.lse_s = x.lse_s #use this in the full join (2)
AND a.income_accrued_c <> 'N' #nvl is not necesary, if its null its <> from 'N'

FULL Join rsu b on x.LSE_S = b.LSE_S #match nro account (3)
FULL Join rarb d on x.por_s = d.por_s #match (4)
and d.invo_s = a.invo_s #use this in the full join (4)
AND d.a_typ_s in ('PROR', 'IDR2', 'SECD')
FULL join vrcx_les y on x.les_s = y.les_s #match nro less
FULL join req z on x.LSE_S = z.LSE_S #match nro account

FULL join rlr k on z.dlr_s = k.dlr_s

FULL join ral n on x.les_s = n.les_s #match nro less
FULL join rug o on x.lse_s = o.lse_s #match nro account

FULL JOIN RGR m ON c.CLIENT_ID_S = m.GRN_S
where 1=1
group by ....


And I agree with the comments, if you want a more helpful answer, you have to post a description of every table used in the query as and a brief explanation of the fields in them. Also the functional explanation of the query, what kind of information you want to extract.






share|improve this answer






















    Your Answer






    StackExchange.ifUsing("editor", function ()
    StackExchange.using("externalEditor", function ()
    StackExchange.using("snippets", function ()
    StackExchange.snippets.init();
    );
    );
    , "code-snippets");

    StackExchange.ready(function()
    var channelOptions =
    tags: "".split(" "),
    id: "1"
    ;
    initTagRenderer("".split(" "), "".split(" "), channelOptions);

    StackExchange.using("externalEditor", function()
    // Have to fire editor after snippets, if snippets enabled
    if (StackExchange.settings.snippets.snippetsEnabled)
    StackExchange.using("snippets", function()
    createEditor();
    );

    else
    createEditor();

    );

    function createEditor()
    StackExchange.prepareEditor(
    heartbeatType: 'answer',
    autoActivateHeartbeat: false,
    convertImagesToLinks: true,
    noModals: true,
    showLowRepImageUploadWarning: true,
    reputationToPostImages: 10,
    bindNavPrevention: true,
    postfix: "",
    imageUploader:
    brandingHtml: "Powered by u003ca class="icon-imgur-white" href="https://imgur.com/"u003eu003c/au003e",
    contentPolicyHtml: "User contributions licensed under u003ca href="https://creativecommons.org/licenses/by-sa/3.0/"u003ecc by-sa 3.0 with attribution requiredu003c/au003e u003ca href="https://stackoverflow.com/legal/content-policy"u003e(content policy)u003c/au003e",
    allowUrls: true
    ,
    onDemand: true,
    discardSelector: ".discard-answer"
    ,immediatelyShowMarkdownHelp:true
    );



    );













    draft saved

    draft discarded


















    StackExchange.ready(
    function ()
    StackExchange.openid.initPostLogin('.new-post-login', 'https%3a%2f%2fstackoverflow.com%2fquestions%2f53286852%2fsql-query-taking-long-time-to-execute%23new-answer', 'question_page');

    );

    Post as a guest















    Required, but never shown

























    1 Answer
    1






    active

    oldest

    votes








    1 Answer
    1






    active

    oldest

    votes









    active

    oldest

    votes






    active

    oldest

    votes









    0














     select distinct z.por_s as Portfolio_Number,
    z.lse_s as Account_Number,
    ...

    from rgc c
    FULL join rls x on c.APP_LSE_S = x.LSE_S #match nro account (1)
    FULL Join rar a on x.LSE_S = a.LSE_S #match nro account (2)
    FULL Join rsu b on x.LSE_S = b.LSE_S #match nro account (3)
    FULL Join rarb d on x.por_s = d.por_s #match (4)
    FULL join vrcx_les y on x.les_s = y.les_s #match nro less
    FULL join req z on x.LSE_S = z.LSE_S #match nro account

    FULL join rlr k on z.dlr_s = k.dlr_s

    FULL join ral n on x.les_s = n.les_s #match nro less
    FULL join rug o on x.lse_s = o.lse_s #match nro account

    FULL JOIN RGR m ON c.CLIENT_ID_S = m.GRN_S
    where x.act_s in
    ('AACQ', 'AASM', 'AEXM', 'ANOR', 'ARBK', 'ARED', 'AREN', 'AXFR') #use this in the full join (2)
    and x.d_mat_s > to_date ('07/01/2018', 'MM/DD/YYYY') #use this in the full join (2)
    and a.por_s = x.por_s #use this in the full join (2)
    AND a.com_s = x.com_s #use this in the full join (2)
    AND a.reg_s = x.reg_s #use this in the full join (2)
    AND a.off_s = x.off_s #use this in the full join (2)
    AND a.lse_s = x.lse_s #use this in the full join (2)
    AND d.invo_s = a.invo_s #use this in the full join (4)
    AND d.a_typ_s in ('PROR', 'IDR2', 'SECD')
    AND NVL(a.income_accrued_c, ' ') <> 'N' #nvl is not necesary
    group by z.por_s,z.lse_s,y.les_s,z.unit_S,x.d_com_s,
    ...


    I would put the restrictions in the joins, so you dont combine all the rows (1000 x 1000 for example), just the rows you need (100 x 1000) Like this:



     from rgc c 
    FULL join rls x on c.APP_LSE_S = x.LSE_S #match nro account (1)
    FULL Join rar a on x.LSE_S = a.LSE_S #match nro account (2)
    and x.act_s in
    ('AACQ', 'AASM', 'AEXM', 'ANOR', 'ARBK', 'ARED', 'AREN', 'AXFR') #use this in the full join (2)
    and x.d_mat_s > to_date ('07/01/2018', 'MM/DD/YYYY') #use this in the full join (2)
    and a.por_s = x.por_s #use this in the full join (2)
    AND a.com_s = x.com_s #use this in the full join (2)
    AND a.reg_s = x.reg_s #use this in the full join (2)
    AND a.off_s = x.off_s #use this in the full join (2)
    AND a.lse_s = x.lse_s #use this in the full join (2)
    AND a.income_accrued_c <> 'N' #nvl is not necesary, if its null its <> from 'N'

    FULL Join rsu b on x.LSE_S = b.LSE_S #match nro account (3)
    FULL Join rarb d on x.por_s = d.por_s #match (4)
    and d.invo_s = a.invo_s #use this in the full join (4)
    AND d.a_typ_s in ('PROR', 'IDR2', 'SECD')
    FULL join vrcx_les y on x.les_s = y.les_s #match nro less
    FULL join req z on x.LSE_S = z.LSE_S #match nro account

    FULL join rlr k on z.dlr_s = k.dlr_s

    FULL join ral n on x.les_s = n.les_s #match nro less
    FULL join rug o on x.lse_s = o.lse_s #match nro account

    FULL JOIN RGR m ON c.CLIENT_ID_S = m.GRN_S
    where 1=1
    group by ....


    And I agree with the comments, if you want a more helpful answer, you have to post a description of every table used in the query as and a brief explanation of the fields in them. Also the functional explanation of the query, what kind of information you want to extract.






    share|improve this answer



























      0














       select distinct z.por_s as Portfolio_Number,
      z.lse_s as Account_Number,
      ...

      from rgc c
      FULL join rls x on c.APP_LSE_S = x.LSE_S #match nro account (1)
      FULL Join rar a on x.LSE_S = a.LSE_S #match nro account (2)
      FULL Join rsu b on x.LSE_S = b.LSE_S #match nro account (3)
      FULL Join rarb d on x.por_s = d.por_s #match (4)
      FULL join vrcx_les y on x.les_s = y.les_s #match nro less
      FULL join req z on x.LSE_S = z.LSE_S #match nro account

      FULL join rlr k on z.dlr_s = k.dlr_s

      FULL join ral n on x.les_s = n.les_s #match nro less
      FULL join rug o on x.lse_s = o.lse_s #match nro account

      FULL JOIN RGR m ON c.CLIENT_ID_S = m.GRN_S
      where x.act_s in
      ('AACQ', 'AASM', 'AEXM', 'ANOR', 'ARBK', 'ARED', 'AREN', 'AXFR') #use this in the full join (2)
      and x.d_mat_s > to_date ('07/01/2018', 'MM/DD/YYYY') #use this in the full join (2)
      and a.por_s = x.por_s #use this in the full join (2)
      AND a.com_s = x.com_s #use this in the full join (2)
      AND a.reg_s = x.reg_s #use this in the full join (2)
      AND a.off_s = x.off_s #use this in the full join (2)
      AND a.lse_s = x.lse_s #use this in the full join (2)
      AND d.invo_s = a.invo_s #use this in the full join (4)
      AND d.a_typ_s in ('PROR', 'IDR2', 'SECD')
      AND NVL(a.income_accrued_c, ' ') <> 'N' #nvl is not necesary
      group by z.por_s,z.lse_s,y.les_s,z.unit_S,x.d_com_s,
      ...


      I would put the restrictions in the joins, so you dont combine all the rows (1000 x 1000 for example), just the rows you need (100 x 1000) Like this:



       from rgc c 
      FULL join rls x on c.APP_LSE_S = x.LSE_S #match nro account (1)
      FULL Join rar a on x.LSE_S = a.LSE_S #match nro account (2)
      and x.act_s in
      ('AACQ', 'AASM', 'AEXM', 'ANOR', 'ARBK', 'ARED', 'AREN', 'AXFR') #use this in the full join (2)
      and x.d_mat_s > to_date ('07/01/2018', 'MM/DD/YYYY') #use this in the full join (2)
      and a.por_s = x.por_s #use this in the full join (2)
      AND a.com_s = x.com_s #use this in the full join (2)
      AND a.reg_s = x.reg_s #use this in the full join (2)
      AND a.off_s = x.off_s #use this in the full join (2)
      AND a.lse_s = x.lse_s #use this in the full join (2)
      AND a.income_accrued_c <> 'N' #nvl is not necesary, if its null its <> from 'N'

      FULL Join rsu b on x.LSE_S = b.LSE_S #match nro account (3)
      FULL Join rarb d on x.por_s = d.por_s #match (4)
      and d.invo_s = a.invo_s #use this in the full join (4)
      AND d.a_typ_s in ('PROR', 'IDR2', 'SECD')
      FULL join vrcx_les y on x.les_s = y.les_s #match nro less
      FULL join req z on x.LSE_S = z.LSE_S #match nro account

      FULL join rlr k on z.dlr_s = k.dlr_s

      FULL join ral n on x.les_s = n.les_s #match nro less
      FULL join rug o on x.lse_s = o.lse_s #match nro account

      FULL JOIN RGR m ON c.CLIENT_ID_S = m.GRN_S
      where 1=1
      group by ....


      And I agree with the comments, if you want a more helpful answer, you have to post a description of every table used in the query as and a brief explanation of the fields in them. Also the functional explanation of the query, what kind of information you want to extract.






      share|improve this answer

























        0












        0








        0







         select distinct z.por_s as Portfolio_Number,
        z.lse_s as Account_Number,
        ...

        from rgc c
        FULL join rls x on c.APP_LSE_S = x.LSE_S #match nro account (1)
        FULL Join rar a on x.LSE_S = a.LSE_S #match nro account (2)
        FULL Join rsu b on x.LSE_S = b.LSE_S #match nro account (3)
        FULL Join rarb d on x.por_s = d.por_s #match (4)
        FULL join vrcx_les y on x.les_s = y.les_s #match nro less
        FULL join req z on x.LSE_S = z.LSE_S #match nro account

        FULL join rlr k on z.dlr_s = k.dlr_s

        FULL join ral n on x.les_s = n.les_s #match nro less
        FULL join rug o on x.lse_s = o.lse_s #match nro account

        FULL JOIN RGR m ON c.CLIENT_ID_S = m.GRN_S
        where x.act_s in
        ('AACQ', 'AASM', 'AEXM', 'ANOR', 'ARBK', 'ARED', 'AREN', 'AXFR') #use this in the full join (2)
        and x.d_mat_s > to_date ('07/01/2018', 'MM/DD/YYYY') #use this in the full join (2)
        and a.por_s = x.por_s #use this in the full join (2)
        AND a.com_s = x.com_s #use this in the full join (2)
        AND a.reg_s = x.reg_s #use this in the full join (2)
        AND a.off_s = x.off_s #use this in the full join (2)
        AND a.lse_s = x.lse_s #use this in the full join (2)
        AND d.invo_s = a.invo_s #use this in the full join (4)
        AND d.a_typ_s in ('PROR', 'IDR2', 'SECD')
        AND NVL(a.income_accrued_c, ' ') <> 'N' #nvl is not necesary
        group by z.por_s,z.lse_s,y.les_s,z.unit_S,x.d_com_s,
        ...


        I would put the restrictions in the joins, so you dont combine all the rows (1000 x 1000 for example), just the rows you need (100 x 1000) Like this:



         from rgc c 
        FULL join rls x on c.APP_LSE_S = x.LSE_S #match nro account (1)
        FULL Join rar a on x.LSE_S = a.LSE_S #match nro account (2)
        and x.act_s in
        ('AACQ', 'AASM', 'AEXM', 'ANOR', 'ARBK', 'ARED', 'AREN', 'AXFR') #use this in the full join (2)
        and x.d_mat_s > to_date ('07/01/2018', 'MM/DD/YYYY') #use this in the full join (2)
        and a.por_s = x.por_s #use this in the full join (2)
        AND a.com_s = x.com_s #use this in the full join (2)
        AND a.reg_s = x.reg_s #use this in the full join (2)
        AND a.off_s = x.off_s #use this in the full join (2)
        AND a.lse_s = x.lse_s #use this in the full join (2)
        AND a.income_accrued_c <> 'N' #nvl is not necesary, if its null its <> from 'N'

        FULL Join rsu b on x.LSE_S = b.LSE_S #match nro account (3)
        FULL Join rarb d on x.por_s = d.por_s #match (4)
        and d.invo_s = a.invo_s #use this in the full join (4)
        AND d.a_typ_s in ('PROR', 'IDR2', 'SECD')
        FULL join vrcx_les y on x.les_s = y.les_s #match nro less
        FULL join req z on x.LSE_S = z.LSE_S #match nro account

        FULL join rlr k on z.dlr_s = k.dlr_s

        FULL join ral n on x.les_s = n.les_s #match nro less
        FULL join rug o on x.lse_s = o.lse_s #match nro account

        FULL JOIN RGR m ON c.CLIENT_ID_S = m.GRN_S
        where 1=1
        group by ....


        And I agree with the comments, if you want a more helpful answer, you have to post a description of every table used in the query as and a brief explanation of the fields in them. Also the functional explanation of the query, what kind of information you want to extract.






        share|improve this answer













         select distinct z.por_s as Portfolio_Number,
        z.lse_s as Account_Number,
        ...

        from rgc c
        FULL join rls x on c.APP_LSE_S = x.LSE_S #match nro account (1)
        FULL Join rar a on x.LSE_S = a.LSE_S #match nro account (2)
        FULL Join rsu b on x.LSE_S = b.LSE_S #match nro account (3)
        FULL Join rarb d on x.por_s = d.por_s #match (4)
        FULL join vrcx_les y on x.les_s = y.les_s #match nro less
        FULL join req z on x.LSE_S = z.LSE_S #match nro account

        FULL join rlr k on z.dlr_s = k.dlr_s

        FULL join ral n on x.les_s = n.les_s #match nro less
        FULL join rug o on x.lse_s = o.lse_s #match nro account

        FULL JOIN RGR m ON c.CLIENT_ID_S = m.GRN_S
        where x.act_s in
        ('AACQ', 'AASM', 'AEXM', 'ANOR', 'ARBK', 'ARED', 'AREN', 'AXFR') #use this in the full join (2)
        and x.d_mat_s > to_date ('07/01/2018', 'MM/DD/YYYY') #use this in the full join (2)
        and a.por_s = x.por_s #use this in the full join (2)
        AND a.com_s = x.com_s #use this in the full join (2)
        AND a.reg_s = x.reg_s #use this in the full join (2)
        AND a.off_s = x.off_s #use this in the full join (2)
        AND a.lse_s = x.lse_s #use this in the full join (2)
        AND d.invo_s = a.invo_s #use this in the full join (4)
        AND d.a_typ_s in ('PROR', 'IDR2', 'SECD')
        AND NVL(a.income_accrued_c, ' ') <> 'N' #nvl is not necesary
        group by z.por_s,z.lse_s,y.les_s,z.unit_S,x.d_com_s,
        ...


        I would put the restrictions in the joins, so you dont combine all the rows (1000 x 1000 for example), just the rows you need (100 x 1000) Like this:



         from rgc c 
        FULL join rls x on c.APP_LSE_S = x.LSE_S #match nro account (1)
        FULL Join rar a on x.LSE_S = a.LSE_S #match nro account (2)
        and x.act_s in
        ('AACQ', 'AASM', 'AEXM', 'ANOR', 'ARBK', 'ARED', 'AREN', 'AXFR') #use this in the full join (2)
        and x.d_mat_s > to_date ('07/01/2018', 'MM/DD/YYYY') #use this in the full join (2)
        and a.por_s = x.por_s #use this in the full join (2)
        AND a.com_s = x.com_s #use this in the full join (2)
        AND a.reg_s = x.reg_s #use this in the full join (2)
        AND a.off_s = x.off_s #use this in the full join (2)
        AND a.lse_s = x.lse_s #use this in the full join (2)
        AND a.income_accrued_c <> 'N' #nvl is not necesary, if its null its <> from 'N'

        FULL Join rsu b on x.LSE_S = b.LSE_S #match nro account (3)
        FULL Join rarb d on x.por_s = d.por_s #match (4)
        and d.invo_s = a.invo_s #use this in the full join (4)
        AND d.a_typ_s in ('PROR', 'IDR2', 'SECD')
        FULL join vrcx_les y on x.les_s = y.les_s #match nro less
        FULL join req z on x.LSE_S = z.LSE_S #match nro account

        FULL join rlr k on z.dlr_s = k.dlr_s

        FULL join ral n on x.les_s = n.les_s #match nro less
        FULL join rug o on x.lse_s = o.lse_s #match nro account

        FULL JOIN RGR m ON c.CLIENT_ID_S = m.GRN_S
        where 1=1
        group by ....


        And I agree with the comments, if you want a more helpful answer, you have to post a description of every table used in the query as and a brief explanation of the fields in them. Also the functional explanation of the query, what kind of information you want to extract.







        share|improve this answer












        share|improve this answer



        share|improve this answer










        answered Nov 14 '18 at 3:13









        YazsidYazsid

        1555




        1555





























            draft saved

            draft discarded
















































            Thanks for contributing an answer to Stack Overflow!


            • Please be sure to answer the question. Provide details and share your research!

            But avoid


            • Asking for help, clarification, or responding to other answers.

            • Making statements based on opinion; back them up with references or personal experience.

            To learn more, see our tips on writing great answers.




            draft saved


            draft discarded














            StackExchange.ready(
            function ()
            StackExchange.openid.initPostLogin('.new-post-login', 'https%3a%2f%2fstackoverflow.com%2fquestions%2f53286852%2fsql-query-taking-long-time-to-execute%23new-answer', 'question_page');

            );

            Post as a guest















            Required, but never shown





















































            Required, but never shown














            Required, but never shown












            Required, but never shown







            Required, but never shown

































            Required, but never shown














            Required, but never shown












            Required, but never shown







            Required, but never shown







            Popular posts from this blog

            Use pre created SQLite database for Android project in kotlin

            Darth Vader #20

            Ondo