Extracting order of execution of objects from DSX file

[clarcombe]

It has been requested that we create an excel spreadsheet with all of the job sequences and parallel jobs in order of execution

Thus
S1 - J1- OCI In - OCI Out
S1 - J2- DS In - OCI Out - DS Out
S2 - S3
S3 - J3 - FF In - OCI Out
etc

At present my colleague is doing this by hand and we have tens of job sequences to evaluate.

I have tried reading the dsx in Excel and filtering on *** Activity and jb$ but I don't always get the objects in the right order

e.g. the dsx has

Code: Select all

*** Activity "Repair_PS_S_COMBO_CF_DEFN": Initialize job
jb$V120S7 = "OCN_J_Stage_HASH_CRC_REPAIR":'.':("Repair_PS
*** Activity "Repair_PS_S_BUS_UNIT_AP": Initialize job
jb$V121S0 = "OCN_J_Stage_HASH_CRC_REPAIR":'.':("Repair_PS

However, the calling order in the DSX is actually

Repair_PS_PAYMENT_TBL
Repair_PS_S_BUS_UNIT_AP

Questions
1) If I have a compiled dsx, is there anyway from reading the generated BASIC that I can get the objects out in the right order ?
2) Are there any other ways apart from reading the dsx that I can get this information.

Thanks

Colin

[qt_ky]

First ask why. You can call jobs in parallel from a sequence job and those can be difficult to represent in a spreadsheet.

Go into Designer under the File menu and generate a job report for each sequence job. A picture is worth a thousand words. It will produce documentation for you by generating the report as a web page.

One great point with DataStage is that it's a GUI. By reducing a GUI job design back to text only format, you're making it more difficult for other people to understand.

[ray.wurlod]

There's no such thing as a compiled DSX, so option 1 is out. The best way is to look at the sequence log - near the end there's a "summary of sequence run" event.

[clarcombe]

Ray, thanks for that. Thats what I needed.

As for the compiled dsx, I meant a job that has been compiled and exported as a dsx (so will have the Universe Basic in )as opposed to a job that has just been saved with no previous compilation.

[eph]

Hi!

Following qt_ky's idea, after a generation of all job's report, it's quite easy to parse the results into a csv

First step : Generate the documentation using DsJobReport (cf Kim Duke's website)
Then execute the following script on unix shell, using the job sequence naming convention (like Jq_, Seq_):

Code: Select all

#!/bin/ksh
# Arg 1 : PATH of the documentation
# Arg 2 : Job Sequence pattern in naming convention



find $1 -name "*$2*.htm*" > listDocs

cat listDocs | while read file;
        do
                echo "File processed : ${file}";
                outputfile=$(echo "${file}" | sed -e 's/htm.*//');
                echo "Type;Stage;Link info;Stage I/O" >  "${outputfile}csv";
                grep -E "</A></B></TR>|Input from|Outputs to"  ${file} > temp1;
                cat temp1 | while read line;
                do
                        case "${line}" in
                                *A\ name*)
                                                stage=$(echo ${line} | sed -e 's/.*">//;s/<.*//')
                                                echo "Stage;${stage}" >> "${outputfile}csv"
                                                ;;
                                *A\ href*)
                                                inOut=$(echo ${line} | sed -e 's/.*<dd>//;s/<.*//')
                                                stageInOut=$(echo ${line} | sed -e 's/.*">//;s/<.*//')
                                                echo "Link;${stage};${inOut};${stageInOut}" >> "${outputfile}csv"
                                                ;;
                        esac;
                done;
        done;

Results like :

Code: Select all

Type	      Stage	                                        Link info	                                                Stage I/O
Stage      uv_Connection		
Link	      uv_Connection	                        Outputs to UserVariables Activity stage 	uv_ConnectionPS
Stage      Jx_SRV_RDD_CONTEXTE_DUP		
Link        Jx_SRV_RDD_CONTEXTE_DUP	Input from Job Activity stage 	                Jx_RD_T_ME_OPD_CLIENT_DUP_001
Link        Jx_SRV_RDD_CONTEXTE_DUP	Outputs to Job Activity stage 	                Jx_RD_T_ME_OPD_CLIENT_DUP_002

It's just a quick&dirty script, tell me if you improve it, I'm interested

Eric