Re: [monit] Re: what triggers resource limit?

[Martin Pala]

Yet one thing - 100% cpu usage is not evil in general (of course it  
depends on context) ... in fact high CPU usage could be good since the  
HW is used on 100% and is not oversized (some companies watch their  
key performance indicators to make sure there is no excess of free  
resources and for example demand usage >= 50%).

Having cpu usage peaks is normal almost on every system - to prevent  
false alerts you can modify the cpu usage conditions to generate alert  
only when the cpu usage is very high for long timeframe (longer then  
expected peaks), for example:


    if cpu usage (user) > 90% for 5 cycles then alert
    if cpu usage (system) > 90% for 5 cycles then alert
    if cpu usage (wait) > 90% for 5 cycles then alert


=> if the usage is > 90% for 5 consecutive cycles, then alert is  
generated


Martin


On May 19, 2009, at 6:09 PM, Abdellah Tantan wrote:

> I get the same email notification (resource limit) from monit but not
> everyday, I am thinking that is the clamav antivirus who performs a  
> system
> scan around midnight everyday, but the notification email is not  
> always
> fired around midnight, It could be also freshclam that checks for  
> updates
> every four hours. I need to keep a close eye on the Antivirus. I  
> will try
> your suggestion. Thanks.
>
>
> Abdellah
>
> -----Original Message-----
> From: address@hidden
> [mailto:address@hidden On  
> Behalf Of
> John Dixon
> Sent: Tuesday, May 19, 2009 9:50 AM
> To: address@hidden
> Subject: [monit] Re: what triggers resource limit?
>
> What I have done on one of my servers is to add an exec line that does
> top and redirects to a file when the specific resource limit is  
> reached.
> See below my system.conf out of /etc/monit.d/
>
> check system hostname
> if loadavg (5min) > 1 then
>        exec "/bin/bash -c '/usr/bin/top -b -n 1
> > > /etc/monit.d/status_log;/bin/date >> /etc/monit.d/status_log'"
>    if memory usage > 50% then
>        exec "/bin/bash -c '/usr/bin/top -b -n 1
> > > /etc/monit.d/status_log;/bin/date >> /etc/monit.d/status_log'"
>    if cpu usage (user) > 30% then
>        exec "/bin/bash -c '/usr/bin/top -b -n 1
> > > /etc/monit.d/status_log;/bin/date >> /etc/monit.d/status_log'"
>    if cpu usage (system) > 40% then
>        exec "/bin/bash -c '/usr/bin/top -b -n 1
> > > /etc/monit.d/status_log;/bin/date >> /etc/monit.d/status_log'"
>    if cpu usage (wait) > 35% then
>        exec "/bin/bash -c '/usr/bin/top -b -n 1
> > > /etc/monit.d/status_log;/bin/date >> /etc/monit.d/status_log'"
>    group system
>
>
> With this setup you will get notified and you will get top output in  
> the
> status_log file.  I'm sure there's a better way to write this same  
> thing
> using the monit config syntax, but this works even if it's not pretty.
> You may need to run this setup for a few days before you catch what is
> using the resources because of timing, etc.  Remember to clear out  
> your
> status_log periodically or add it to a logrotate setup.
> > Hello,
> > i've got a problem with monit, i configure it to alert me when cpu  
> > usage
> is too
> > important, and i've got this mail :
> >
> > Resource limit matched Service Intranet
> >         Date:        Mon, 18 May 2009 04:13:24 +0200
> >         Action:      alert
> >         Host:        tamtam
> >         Description: 'Intranet' cpu user usage of 70.4% matches resource
> limit [cpu
> > user usage>70.0%]
> >
> > Resource limit matched Service Intranet
> >         Date:        Mon, 18 May 2009 04:13:25 +0200
> >         Action:      alert
> >         Host:        tamtam
> >         Description: 'Intranet' loadavg(5min) of 2.2 matches resource limit
> > [loadavg(5min)>2.0]
>
>
>
>
> --
> To unsubscribe:
> http://lists.nongnu.org/mailman/listinfo/monit-general
>
>
>
>
>
>
>
> --
> To unsubscribe:
> http://lists.nongnu.org/mailman/listinfo/monit-general