monit-general
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Memory totals not adding up


From: Darhl Thomason
Subject: Re: Memory totals not adding up
Date: Thu, 1 Sep 2011 11:47:16 -0700

Thanks Martin!!

Sent from my AT&T wireless handheld

On Sep 1, 2011, at 9:27 AM, Martin Pala <address@hidden> wrote:

You are right, int he case of the process-based worker pool such as Apache it is confusing to display only the CPU and Memory usage of the parent/control process.

We'll modify the overview page to display the resource usage including children (will work fine even for services without children - in their case the total memory usage and memory usage are equal).

Best regards,
Martin


On Aug 30, 2011, at 4:33 AM, Darhl Thomason wrote:

Martin,
 
Thanks for the reply and I now see the 'Total Memory Usage (incl. children)' on the detail page for the Apache process.  It's a bit confusing because the top line of the main page shows total memory used, but when you add up the memory displayed on that page for each process it doesn't add up.
 
Are there any plans or is there any possibility for the main page to show 'Total Memory Usage (incl. children)'?  I also just noticed that CPU usage is reported the same way.  The main page only shows CPU for the single process, but the detail page shows 'Total CPU Usage (incl. children)'.
 
If you have a wishlist, I'd like to add the Total Memory Usage (incl. children) and Total CPU Usage (incl. children) on the main page.  Maybe have this as an option in the config file so you can do something like this:
 
check process Apache with pidfile /var/run/apache2.pid
   group www
   show totalmem
   show totalcpu
   start program = "/etc/init.d/apache2 start"
   stop program  = "/etc/init.d/apache2 stop"
   if totalcpu > 60% for 2 cycles then alert
   if totalcpu > 80% for 5 cycles then restart
   if totalmem > 500 MB for 5 cycles then restart
   if children > 250 then restart
   if loadavg(5min) greater than 10 for 8 cycles then restart
   if 3 restarts within 5 cycles then timeout
 
Thanks!!!
 
Darhl

On Mon, Aug 29, 2011 at 12:40 PM, Martin Pala <address@hidden> wrote:
Hi,

sorry for long delay.

The configuration and the reported values are OK … the problem is simple - the attached snapshot of the Monit httpd interface report is from the main page (overview) which doesn't show the total memory usage, but only the memory usage of the process itself … in your case it matches the 'ps aux' output for the parent apache process with PID 24120 (10692kB => 0.6%). To display the total memory usage (including children) you need to click on the "Apache" link which will open details for the process and the total memory will be displayed in the table row with the title "Total memory usage (incl. children)". 

Regards,
Martin



On Aug 24, 2011, at 3:09 PM, Darhl Thomason wrote:

Hi Martin,
 
Wondering if you had any suggestions for me...
 
Thanks!
 
Darhl

On Fri, Aug 19, 2011 at 11:27 AM, Darhl Thomason <address@hidden> wrote:
Hi Martin,
 
Sorry for the delay.  Apache portion of monitrc below:

check process Apache with pidfile /var/run/apache2.pid
   group www
   start program = "/etc/init.d/apache2 start"
   stop program  = "/etc/init.d/apache2 stop"
   if totalcpu > 60% for 2 cycles then alert
   if totalcpu > 80% for 5 cycles then restart
   if totalmem > 500 MB for 5 cycles then restart
   if children > 250 then restart
   if loadavg(5min) greater than 10 for 8 cycles then restart
   if 3 restarts within 5 cycles then timeout
 
 
 

 
On Sat, Aug 13, 2011 at 7:52 AM, Martin Pala <address@hidden> wrote:
Please can you post your monit configuration for apache service?

If you're using the pidfile based monitoring, the pidfile should contain the pid of the apache master process which is parent to all apache childs. Monit then summarizes the memory of all its childs.

If you're using the pattern matching process monitoring and multiple processes match the pattern, then monit checks only the first matching process - if this process is child, then it wont see the total memory usage of the sibblings.

Regards,
Martin



On Aug 10, 2011, at 8:20 PM, Darhl Thomason wrote:

I'm using Monit 5.2.5 running on Debian Squeeze

When I look at my Monit web portal I see items similar to this:

Apache running 9d 3h 27m  0.0% 0.6% [10692 kB]

But when I do 'ps aux | grep apache' I get:
address@hidden:/# ps aux | grep apache
www-data  2009  0.0  0.8  44528 13404 ?        S    Aug09   0:22 /usr/sbin/apache2 -k start
www-data  4941  0.0  0.8  44524 13248 ?        S    Aug09   0:07 /usr/sbin/apache2 -k start
root      5194  0.0  0.0   3300   740 pts/1    S+   11:11   0:00 grep apache
www-data  5733  0.0  0.8  44524 13096 ?        S    Aug09   0:18 /usr/sbin/apache2 -k start
www-data  8829  0.0  0.8  44524 13108 ?        S    Aug09   0:16 /usr/sbin/apache2 -k start
www-data  9411  0.0  0.8  44524 13112 ?        S    Aug09   0:15 /usr/sbin/apache2 -k start
www-data  9423  0.0  0.8  44524 13104 ?        S    Aug09   0:28 /usr/sbin/apache2 -k start
www-data 13388  0.0  0.8  44256 12500 ?        S    06:05   0:01 /usr/sbin/apache2 -k start
www-data 13484  0.0  0.8  44524 12940 ?        S    06:06   0:00 /usr/sbin/apache2 -k start
www-data 17911  0.0  0.3  38296  5736 ?        S    06:22   0:00 /usr/sbin/apache2 -k start
www-data 22802  0.0  0.8  44524 13120 ?        S    Aug09   0:10 /usr/sbin/apache2 -k start
root     24120  0.0  0.6  38160 10692 ?        Ss   Aug01   1:08 /usr/sbin/apache2 -k start

it adds up to 8.1% memory.  Why is it only reporting 0.6% Memory?

There are other/similar items examples Amavis, Spamassassin, etc.

Thanks!

Darhl

--
To unsubscribe:
https://lists.nongnu.org/mailman/listinfo/monit-general


--
To unsubscribe:
https://lists.nongnu.org/mailman/listinfo/monit-general


--
To unsubscribe:
https://lists.nongnu.org/mailman/listinfo/monit-general


--
To unsubscribe:
https://lists.nongnu.org/mailman/listinfo/monit-general

--
To unsubscribe:
https://lists.nongnu.org/mailman/listinfo/monit-general


reply via email to

[Prev in Thread] Current Thread [Next in Thread]