append data to each line

Shell Programming and Scripting
1

Hi guys,

I need to investigate a memory leak on a solaris server, so what I have done is pmap'd each process on the system with a script which tar'd the directory every hour in cron. Now I need to write a script to process the pmap data.

So what I have is about 100 directories

[root@localhost ~]# ll pmaptest/pmapdata2009101
pmapdata20091012122344/ pmapdata20091013080000/ pmapdata20091013230000/ pmapdata20091014140000/ pmapdata20091015050000/ pmapdata20091015200001/
pmapdata20091012180000/ pmapdata20091013090000/ pmapdata20091014000000/ pmapdata20091014150000/ pmapdata20091015060000/ pmapdata20091015210000/
pmapdata20091012190000/ pmapdata20091013100000/ pmapdata20091014010000/ pmapdata20091014160000/ pmapdata20091015070000/ pmapdata20091015220000/
.......

inside each directory is the output from the pmap on that process

[root@localhost pmapdata]# ll
total 1172
-rw-r--r-- 1 root root    21 2009-10-12 17:00 pmap0.txt
-rw-r--r-- 1 root root 21394 2009-10-12 17:00 pmap11746.txt
-rw-r--r-- 1 root root  3148 2009-10-12 17:00 pmap1294.txt
-rw-r--r-- 1 root root  8578 2009-10-12 17:00 pmap13176.txt
-rw-r--r-- 1 root root  9316 2009-10-12 17:00 pmap14005.txt
-rw-r--r-- 1 root root  2862 2009-10-12 17:00 pmap14006.txt
-rw-r--r-- 1 root root  2134 2009-10-12 17:00 pmap141.txt
-rw-r--r-- 1 root root 13857 2009-10-12 17:00 pmap14467.txt
-rw-r--r-- 1 root root 13959 2009-10-12 17:00 pmap14468.txt
-rw-r--r-- 1 root root 20210 2009-10-12 17:00 pmap14469.txt
-rw-r--r-- 1 root root 13003 2009-10-12 17:00 pmap14470.txt
-rw-r--r-- 1 root root  7652 2009-10-12 17:00 pmap16406.txt
-rw-r--r-- 1 root root  6032 2009-10-12 17:00 pmap16436.txt
-rw-r--r-- 1 root root  6032 2009-10-12 17:00 pmap16438.txt

the format of the filename is pmap<pid>.txt

inside each file looks like this

[root@localhost pmapdata]# cat pmap985.txt
 4423 1388 grep $pid
985:    /usr/local/sbin/xinetd -f /usr/local/etc/xinetd.conf
 Address  Kbytes     RSS    Anon  Locked Mode   Mapped File
08045000      12      12       4       - rw---    [ stack ]
08050000     212     140       -       - r-x--  xinetd
08094000       4       4       -       - rwx--  xinetd
08095000      44      40      20       - rwx--    [ heap ]
FED20000      64       4       -       - rwx--    [ anon ]
FED40000      64       8       -       - rwx--    [ anon ]
FED60000    1080    1080       -       - r-x--  libc.so.1
FEE7E000      32      32      20       - rw---  libc.so.1
FEE86000       8       4       -       - rw---  libc.so.1
FEE90000     516     516       -       - r-x--  libnsl.so.1
FEF21000      20      20       4       - rw---  libnsl.so.1
FEF26000      32       8       -       - rw---  libnsl.so.1
FEF30000      44      44       -       - r-x--  libsocket.so.1
FEF40000      24      12       4       - rwx--    [ anon ]
FEF4B000       4       4       -       - rw---  libsocket.so.1
FEF50000     268      84       -       - r-x--  libm.so.2
FEFA2000      16       8       -       - rwx--  libm.so.2
FEFC0000       4       4       -       - rwx--    [ anon ]
FEFC3000     160     160       -       - r-x--  ld.so.1
FEFF0000       4       4       4       - rwx--    [ anon ]
FEFF8000       4       4       -       - rwxs-    [ anon ]
FEFFB000       8       8       4       - rwx--  ld.so.1
FEFFD000       4       4       4       - rwx--  ld.so.1
-------- ------- ------- ------- -------
total Kb    2628    2204      64       -

What I am interested in is the anon for each pmap is the total at the bottom for the Anon column. I want to output this data in csv format for use inside an excel spreadsheet where I can graph the data.

So the output format need to be

pid,anonmemtotal,anonmemtotal,anonmemtotal,anonmemtotal
pid,anonmemtotal,anonmemtotal,anonmemtotal,anonmemtotal
pid,anonmemtotal,anonmemtotal,anonmemtotal,anonmemtotal

etc

I have started writing a script to do that which is below, but I cant figure out how I will get the pid and mem total appended to the same line each time?

#!/bin/bash
ls -l /root/pmaptest/ | awk '{print $8}' > /root/pmapdirs.txt
for dir in `cat /root/pmapdirs.txt`
do
cat /root/pmaptest/pmapdata$dir/var/tmp/pmapdata/psefo.txt | awk '{print $1}' > /root/pmaptest/pmapdata$dir/var/tmp/pmapdata/pids.txt
for pid in `cat /root/pmaptest/pmapdata$dir/var/tmp/pmapdata/pids.txt`
do
tail -1 /root/pmaptest/pmapdata$dir/var/tmp/pmapdata/pmap$pid.txt  | awk {'print $5'}
done
done
2 
for i in `find /root/pmaptest -type f -name pmap*.txt`
do
  PID=`echo "$i" |awk -F/ '{print $NF}' |sed -e 's/^pmap//' -e 's/\.txt//' `
  anonmemtotal=`tail -1 $i |awk '{print $5}' `
  echo $PID "," $anonmemtotal
done

you can create another script (or add one more function) to merge anonmemtotal into same PID.

3

If I understand what you are looking to do, this awk script will take a list of pmap files of the form you presented and print a csv line with the pid and anon values found in all the pmap files. The anon values are in the order of the pmap files given on the command line.

(

FNR == 1) {

        getline   # skip to the pid
        this_pid = strtonum($1)
        if( this_pid in pid_list)
                pid_list[this_pid] = pid_list[this_pid] ","     # add a comma
        else
                pid_list[this_pid] = "%d,"              # set the format string
        next
}
/^total/ {              # look for total line

        pid_list[this_pid] = pid_list[this_pid] $5      # add anon total
}

END {
        for( i in pid_list )
                printf pid_list "\n" , i
}

To use it, put the above code in a file (pid.awk in my example).
Then use a command of this form

awk -f pid.awk <list_of_pmap_file>

The <list_of_pmap_file> should be a wildcard spec or back tick command to produce a list of files in the order you want them processed.