Hi,

I am a student doing a experiment. I need to obtain the memory usage (RSS or real) of Apache httpd process every couple of seconds and write it to a log file.

I need to record the time from beginning to receive http requests from the only client to the time when the Apache httpd memory reaches 90% of total memory.

I'm thinking of writing a perl script. What commands and/or monitoring software are available for this task?

Thanks,

My experiment includes one host: Ubuntu Apahce2, one client: Ubuntu connected by switch

pmap would work but it's kind of noisy

pmap -x and just look at the mid number of the last line that's total RSS

Better to use /proc/<pid>/status and grep VmRSS - lot easier to handle. Both include the shared libraries, so you could debate whether it *really* is the real usage of that task. And when/if it gets to 90% of total will be highly contentious.

1 members found this post helpful.

Thank you! by the way, can I get the memory used every couple of seconds?

/proc is not a real filesystem despite what it appears. It is a pseudo filesystem, the contents of which are created ("filled in" if you like) by kernel code when you "read" the file.
Read it as often as you wish - the contents will always be correct at that instant in time.

Here is a script written by someone. How can I revise this one to meet my requirement?

I am sorry for basic questions, but I am a newbie....


#!/bin/sh
PROCNAME=httpd
SYSDATE=`date -d today +"%Y-%m-%d %H:%M:%S"`
NUM=$(ps -ef | grep $PROCNAME | wc -l)
echo -n $SYSDATE >> /root/apache_monitor.log
echo -n ' NUM:' >> /root/apache_monitor.log
echo $NUM >> /root/apache_monitor.log


Maybe I can change some commands in it, can make this run every few seconds, but how can I achieve this? Thanks alot

You're not going to learn much if others do it all for you.
We're here to nudge you in the right direction.

I am running the following code and receiving the warning :


Use of uninitialized value in pattern match (m//) at processMonitor.pl line 38


\processMonitor.pl

#!/usr/bin/perl -w
#
#name: processMonitor.pl
#written by trizsolo
#usage: ./processMonitor.pl args args args
#you can add as many processes as needed for args
#or use the defaults... syslogd is always checked
#----------------------------
my $user='root';
my $host=`hostname`;
chomp($host);
my $date=`date`;
chomp($date);
my $logfile="$host" . '.log';
our @proc_list=@ARGV;
# Default list of processes
if(@proc_list==0){ # check for arguments
@proc_list=('inetd', 'sendmail', 'chkMounts.pl');
}
push(@proc_list, 'syslogd'); # add syslogd to check

# ---------- open logfile and determine Operating System --------------
open(LOG, ">>$logfile") or die "Can't open $logfile to write: $!";
print LOG "Searching for: @proc_list on $date!\n";
chomp($os=`uname -r`);
print LOG "OS on this server is $os\n";
close(LOG);

# ---------------- call "ps" & analyse output --------------------
foreach my $process (@proc_list) {
$event_count{$process} = 0;
}
open(PS, "/bin/ps auwx |") or die "Can't run ps: __FILE__ $!";
$/="\n"; # record seperator
while(
) {
foreach my $process(@proc_list) {
if($_ =~(m/$process/i)) {
$event_count{$process}++;
}
}
}
close(PS);
open(LOG,">>$logfile") or die "Cannot open $logfile: $!";
# Add line to log to nagios if needed
foreach $process(@proc_list) {
if($event_count{$process} ==0) {
print LOG "Process $process is NOT running!\n";
system("/bin/logger -p warn Process: $process is NOT running!\n")
== 0 or die "Cannot complete cmd: $! : $?";
}else{
print LOG "Process $process occurred $event_count{$process} times!\n";
}
}
close(LOG);
exit 0;
#EOF

please help

I'd guess it's because you don't have a space between $process and (@proc_list) though you'd think in that case it'd say the foreach was bad. Also Code tags help keep formatting [ CODE ] [ /CODE ] but without the extra internal spaces