Re: CPU usage in Monit

[Lawrence, Wayne]

Hi Martin,

 

I have tried various methods to dientify the cause of this and took your advice and used vmstat. I simply restarted the httpd process from the monit web interface while the comand was running and got the following warning.

 

      Description: cpu system usage of 50.0% matches resource limit [cpu system usage>30.0%]

 

But vmstat doesnt show that level of usage at the point of alert. As you can see there is some usage in the 3rd line of the output when i restarted the httpd service but it doesnt seem enough to trigger an alert.

 

vmstat 1 10
procs -----------memory---------- ---swap-- -----io---- --system-- -----cpu-----
 r  b   swpd   free   buff  cache   si   so    bi    bo   in   cs us sy id wa st
 0  0      0 859596 114684 856908    0    0     4     6   81   77  0  0 99  0  0
 0  0      0 859448 114684 856916    0    0     0     0  100   94  1  0 99  0  0
 0  0      0 898352 114692 815600    0    0     0   168  555  605 23 15 61  1  0

 

Not sure if there are any other tests i can run to narrow this down a bit further as it still isn't making sense.

 

Regards

 

Wayne

 

 

 

On 7 December 2011 08:27, Martin Pala <address@hidden> wrote:

Hi Lawrence,

the test which triggers the alert is "system" cpu => it's the time the system spend in kernel mode. The cpu usage could be triggered by some background kernel task, to verify the monit report matches the system cpu usage, you should use either "vmstat" or "top" instead of "ps".

Best regards,

Martin 

On Dec 6, 2011, at 1:19 PM, Lawrence, Wayne wrote:

Hi Igor,

 

the operating system is RHEL6 and monit version is 5.3.1

 

this is what i have in my config

 

    if cpu usage (user) > 70% then alert
    if cpu usage (system) > 30% then alert
    if cpu usage (wait) > 20% then alert

this is one of the errors

Description: cpu system usage of 50.0% matches resource limit [cpu system usage>30.0%]

 

this is what i get in /var/log/messages

Dec  6 12:01:29 <hostname-removed> monit[864]: <hostname-removed> cpu system usage of 50.0% matches resource limit [cpu system usage>30.0%]

Dec  6 12:02:29 <hostname-removed> monit[864]: <hostname-removed><hostname-removed>' cpu system usage check succeeded [current cpu system usage=0.9%]

 

this is the output of ps --no-headers -A -o "%cpu sz ucomm" | sort -k1nr | head -20

 

 12:01:29 up 4 days, 20:24,  2 users,  load average: 0.04, 0.01, 0.00
             total       used       free     shared    buffers     cached
Mem:       2055108    1092176     962932          0      53156     811864
-/+ buffers/cache:     227156    1827952
Swap:      4128760          0    4128760
 1.2 44308 perl
 0.0     0 aio/0
 0.0     0 async/mgr
 0.0     0 ata/0
 0.0     0 ata_aux
 0.0     0 bdi-default
 0.0     0 cpuset
 0.0     0 crypto/0
 0.0     0 events/0
 0.0     0 ext4-dio-unwrit
 0.0     0 flush-253:0
 0.0     0 jbd2/dm-0-8
 0.0     0 kacpi_hotplug
 0.0     0 kacpi_notify
 0.0     0 kacpid
 0.0     0 kauditd
 0.0     0 kblockd/0
 0.0     0 kdmflush
 0.0     0 khelper
 0.0     0 khubd

Have to say i am at a total loss as there is no way the usage figures are accurate.

If there is any other info i can supply that will be useful please let me know.

 

Regards

 

Wayne

On 6 December 2011 12:03, Igor Homyakov <address@hidden> wrote:

Hi Lawrence,

Could you be a little bit more specific ?  Please provide information
about you operation system, monit version on which the problem
occurred and so on.

Regards
Igor Homyakov

On Tue, Dec 6, 2011 at 15:35, Lawrence, Wayne
<address@hidden> wrote:
> Hi,
>
> I have a few CPU usage checks in my monitrc but it seems monit is
> misreporting the usage.
>
> I have run several tests and it seems that monit is multiplying the actual
> usage by 10.
>
> I ran a process with top running in another shell and CPU usage for the user
> was never above 10% yet monit informed me that there was 100% cpu usage.
>
> I have tried various configurations including the one that came with the
> default config for system cpu monitoring and all seem to demonstrate the
> same issue.
>
> Any advice welcomed on this
>
> Regards
>
> Wayne Lawrence
>
>
>

> --
> 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