Shell Programming and Scripting

first_1:3:4:5:6:1:3:4:5:something:notinterested
second_2:3:4:3:4:2
first_1:3:4:6:6:7:8

echo "Group	Jobs	WallClock	CPU	TotalMem	Memory"
        first_1     2              4                6              7                  8
        second_2    1              3                4              8                  3

#! /bin/sh
echo "Group	Jobs	WallClock	CPU	TotalMem	Memory"
awk -F":" ' { printf("%s	%d	%d	%d	%d	%d\n",$3,$6,$14,$37,$38) } ' $1

awk -F":" ' BEGIN{print "Group\tJobs\tWallClock\tCPU\tTotalMem\tMemory"}
{Jobs[$1]+=$3;WC[$1]+=$6;CPU[$1]+=$14;TM[$1]+=$37;Mem[$1]+=$38} 
END{for ( i in Jobs) printf "%s\t%d\t%d\t%d\t%d\t%d\n",i, Jobs[i],WC[i],CPU[i],TM[i],Mem[i]} ' $1

awk -F":" '
NF==44 {
        # $3 - group
        # $6 - jobs
        # $14 - wallclock
        # $37 - CPU
        # $38 - TotalMem
        #1024*$38/$37 - Memory

        ++gsum[$3]
        ++gcnt
        jsum[$3]+=$14
        jcnt+=$14
        csum[$3]+=$37
        ccnt+=$37
        tmsum[$3]+=$38
        tmcnt+=$38
}
END {
        printf("Gruop\tJobs\tWallclock\tCPU\tTotalMem\tMemory\n")
        for(i in jsum) {
                printf("%-10s\t%.2f%\t%.2f%.2f\t.2f\t.2f\n", i, gsum[i]*100/gcnt, jsum[i]*100/jcnt,csum[i]*100/ccnt[i]*100,tmsum[i]*100/tmcnt,1024*tmsum[i]/csum[i])
        }
}' $1

    #$3 - Group
    #$6 - Jobs
    #$14 - Wallclock
    #$37 - CPU
    #$38 - TotalMem
    #1024*$38/$37 - MedMem
echo ""
date +"DATA: %m-%d-%y"
echo ""
awk -F":" ' BEGIN{
print "Group\tN. Jobss\tWallClock(s)\tCPU(s)\tTotalMem(GB)\tMedMem(MB)"}
{Jobs[$1]+=$3;WC[$1]+=$6;CPU[$1]+=$14/1000;TM[$1]+=$37;Mem[$1]+=1024*$38/$37} 
END{for ( i in Jobs) printf "%s\t%d\t%d\t%d\t%d\t%.2f\n",i, Jobs[i],WC[i],CPU[i],TM[i]/1024,Mem[i]} ' $1

	#$3 - Grupo
	#$6 - N.Tarefas
	#$14 - Wallclock
	#$37 - CPU
	#$38 - MemTotal
	#1024*$38/$37 - Mem #estamos dividir pq temos k apresentar em GB
echo ""
date +"DATA: %m-%d-%y"
echo ""
#awk -F":" permite ir ao ficheiro neste caso $1 ( ou seja ficheiro com nome que vem depois de chamar o script
# ./ge-stat.sh <accounting-20110226> e ler cada campo separado por ":"
#Jobs[$3]++ array com os nomes dos grupos, cada vez quando nome repete-se ele mete no arrray
#WC[$3]+=$6 - WallClock de cada grupo quando repete-se soma-se os valores do mesmo grupo. Esses valores
#vai buscar ao campo $6 do ficheiro accounting
#$CPU,$TM,$Mem - igual ao wallclock, vai somar os valores com mesmo group name.
awk -F":" ' BEGIN{
print "Grupo\tN. Tarefas\tWallClock(s)\tCPU(s)\tMemTotal(GB)\tMem(MB)"}
{Jobs[$3]++;WC[$3]+=$6;CPU[$3]+=$14;TM[$3]+=$37;Mem[$3]+=1024*$38/$37} 
END{for ( i in Jobs) printf "%s\t%d\t%d\t%d\t%d\t%.2f\n",i, Jobs[i],WC[i],CPU[i],TM[i]/1024,Mem[i]} ' $1