dheepa/Retail_ETL_Control_Scripts
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

Retail ETL jobs

Browse Source
This commit is contained in:
dheepa
2021-11-10 03:46:16 +00:00
commit a1b0f0facc
9 changed files with 248140 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

128
retail_update_run.sh Executable file
View File

@@ -0,0 +1,128 @@
#!/bin/bash -e
#Declare the Associative array
declare -A credentials
HOME_DIR=$HOME
echo $HOME_DIR
CONF_DIR=$HOME_DIR/conf
#Server credentials file path
input_file=$CONF_DIR/server_credentials.txt
#Read the credentials from the server_credentials from conf folders
#Remove Blank lines and comment starting with #
while IFS="=" read -r key value; do
credentials[$key]=$value
done < <(sed -e '/^\s*$/ d' -e '/^#/ d' $input_file)
# done < <(sed -e '/^\s*$/ d' -e '/^#/ d' $input_file)
#CODS credentials
host=${credentials[host]}
#echo ${credentials[host]}
user=${credentials[user]}
#echo ${credentials[user]}
pwd=${credentials[password]}
#echo ${credentials[password]}
port=${credentials[port]}
dbname=${credentials[dbname1]}
psql_base_arg="postgresql://$user:$pwd@$host/$dbname"
echo $psql_base_arg
echo "Initiating a job"
date
/home/compegence/dev/retail/retail_monthly_job_load.sh
#while loop for each run_frequency
while IFS="|" read -a outer
do
client_id=${outer[0]}
function_id=${outer[1]}
run_frequency=${outer[2]}
echo "$run_frequency"
#Update start_time for fw_jobctl_runschedule
psql $psql_base_arg -c "update fw_core.fw_jobctl_runschedule set start_time=now(),begin_status='started' where client_id=$client_id and function_id=$function_id and run_frequency='$run_frequency' and end_status is null "
#while loop for each job & step
while IFS="|" read -a inner
do
job_name=${inner[4]}
step_name+=(${inner[5]})
job_script=${inner[6]}
job_function_name=${inner[7]}
echo "$job_function_name"
#remove first and last char - curly braces
job_fun_param=${inner[8]:1:-1}
data_from_date==${inner[10]}
data_to_date=${inner[11]}
job_scope=${inner[12]}
job_scope_qualifier=${inner[13]}
#echo $data_from_date,$data_to_date,$job_scope,$job_scope_qualifier
param=''
var1=' '
echo "$job_fun_param"
echo "$job_script"
#reading array column and spliting each var and appending $ & space
IFS=',' read -r -a ADDR <<< "$job_fun_param"
for i in "${ADDR[@]}"; do
echo $i
param=$param$i$var1
done
echo $param
if [ $job_script == "R" ] ; then
echo " $job_function_name started"
psql $psql_base_arg -c "update fw_core.fw_jobctl_runschedule_jobstep set jobstep_start_time=now(),begin_status='started',job_status='started' where job_function_name='$job_function_name' and client_id=$client_id and function_id=$function_id and run_frequency='$run_frequency' and latest_flag='true' and job_script='R'"
#Rscript /home/compegence/R_code/$job_function_name $host $port $user $pwd $dbname $param $client_id $function_id $data_from_date $data_to_date $job_scope $job_scope_qualifier
echo "$job_scope_qualifier"
Rscript /home/compegence/R_code/$job_function_name $host $port $user $pwd $dbname $param $client_id $function_id "2017-01-01" "2020-12-31" $job_scope $job_scope_qualifier
psql $psql_base_arg -c "update fw_core.fw_jobctl_runschedule_jobstep set jobstep_end_time=now(),end_status='completed',job_status='completed',latest_flag='false' where job_function_name='$job_function_name' and client_id=$client_id and function_id=$function_id and run_frequency='$run_frequency' and latest_flag='true' and job_script='R'"
echo " $job_function_name Ended"
elif [ $job_script == "SQL" ] ; then
echo " $job_function_name started"
psql $psql_base_arg -c "update fw_core.fw_jobctl_runschedule_jobstep set jobstep_start_time=now(),begin_status='started',job_status='started' where job_function_name='$job_function_name' and client_id=$client_id and function_id=$function_id and run_frequency='$run_frequency' and latest_flag='true' and job_script='SQL'"
# psql $psql_base_arg -a -f /home/compegence/R_code/$job_function_name $host $port $user $pwd $dbname $param $client_id $function_id $data_from_date $data_to_date $job_scope $job_scope_qualifier
psql $psql_base_arg -a -f $job_function_name $host $port $user $pwd $dbname $param $client_id $function_id $data_from_date $data_to_date $job_scope $job_scope_qualifier
psql $psql_base_arg -c "update fw_core.fw_jobctl_runschedule_jobstep set jobstep_end_time=now(),end_status='completed',job_status='completed',latest_flag='false' where job_function_name='$job_function_name' and client_id=$client_id and function_id=$function_id and run_frequency='$run_frequency' and latest_flag='true' and job_script='SQL'"
echo " $job_function_name Ended"
elif [ $job_script == "JAVA" ] ; then
echo " $job_function_name started"
psql $psql_base_arg -c "update fw_core.fw_jobctl_runschedule_jobstep set jobstep_start_time=now(),begin_status='started',job_status='started' where job_function_name='$job_function_name' and client_id=$client_id and function_id=$function_id and run_frequency='$run_frequency' and latest_flag='true' and job_script='JAVA'"
# java -jar /home/compegence/mmt_code/generic_table/target/ex1-1.0-SNAPSHOT-shaded.jar
# java -jar $job_function_name
psql $psql_base_arg -c "update fw_core.fw_jobctl_runschedule_jobstep set jobstep_end_time=now(),end_status='completed',job_status='completed',latest_flag='false' where job_function_name='$job_function_name' and client_id=$client_id and function_id=$function_id and run_frequency='$run_frequency' and latest_flag='true' and job_script='JAVA'"
echo " $job_function_name Ended"
fi
done < <(psql $psql_base_arg -At -c"select client_id,function_id,run_frequency,run_schedule_id,job_name,step_name,job_script,job_function_name,job_fun_param,job_step_run_dependency_seuqence,data_from_date,data_to_date,job_scope,job_scope_qualifier from fw_core.fw_jobctl_runschedule_jobstep where latest_flag='true' and client_id=$client_id and function_id=$function_id and run_frequency='$run_frequency' order by run_schedule_id,job_step_run_dependency_seuqence; ")
#Update end_time for fw_jobctl_runschedule
psql $psql_base_arg -c "update fw_core.fw_jobctl_runschedule set end_time=now(),end_status='stg1_completed' where client_id=$client_id and function_id=$function_id and run_frequency='$run_frequency' and end_status is null "
done < <(psql $psql_base_arg -At -c"select distinct client_id,function_id,run_frequency from fw_core.fw_jobctl_runschedule_jobstep where latest_flag='true' order by client_id,function_id,run_frequency ")