#12 nmon -s not working

[Mohamed LAKRAA]

Hello,

I run monitoring with NMOM, and i want to make snapshots every 3 seconds but it does not working.
I run this command:
~/TMC/monitor/nmon_x86_64_rhel54 -f -t -s 3 -c 5000 &

The output it's like:
CPU Total (Every 2 minutes)
15:33:36
15:35:32
15:37:28
15:39:20
15:41:15
15:43:11
15:45:06
15:47:02

Thanks for help.

Regards,

[Nigel Griffiths]

Hi,
Not seen this issue . . . ever!
Are you really running RHEL 5.4 that is pretty ancient?

Can you tell me the nmon version, please?
Type: nmon -?
It should be version 16f or higher.

Then can you check the that yourcomputer is not totally CPU bound so badly that nmon can' get CPU time. PerhapsType: nmon and hit c
q to quit.

On the output file actually return the snap shot times with: grep ^ZZZ yourfile.nmon | head
and return the output here.

Also the output from: grep ^AAA yourfile.nmon

If it not anythng obvious I would like a copy of the nmon output file next.

Cheers, Nigel

[Mohamed LAKRAA]

Hello,

Just for information, I run the same command on many IS, and it works well except one. so I suspect a configuration of IS and not related to NMON.

what can I check ?

Thanks and regards,

[Mohamed LAKRAA]

Hello,

I installed last version of NMON.
Please find in attach the output.

Thanks and regards,

[Nigel Griffiths]

Thanks for the information - that looks OK.

Current ideas:
a) Lack of CPU time: Can you share the nmon file so I can check the how busy the CPU is, if this is Disk bound and is there any free memory?
b) Problem reading a file: if we have hardware errors some files might beVERY slow to read.

Something must be holding nmon back.
The 3 second snapshots are taking 4 to 8 seconds short of two minutes.
There is nothing in nmon to do that.
1) Quick test as a non-root user (root has a special high priority): date; sleep 3; date
and compare the dates for a 3 second gap or more.

2) Also check the Linux error logs - perhaps we have a faultly hardware or disk taking 90 seconds to return stats?

3) If you are willing to share the nmon file either attach it here (I can delete it once I have it) or compress it with gzip or zip and email it to me nigelargriffiths at hotmail dot com
If you have security concerns edit out or replace the serial number, hostname and IP addresses as everything else is not really a security concern.

4) The VM has been running 9 months - is a rebot possible. But do check the error logs first or i may fail to boot - but that would fix the nmon issues :-)
Cheers, Nigel

[Mohamed LAKRAA]

Hello,

an email has been sent.

Thank you.

[Nigel Griffiths]

The nmon file you sent me says:

1) command = /home/user/TMC/monitor/nmon_x86_64_rhel54 -f -t -s 3 -c 5000
Which implies this is the nmon for RHEL 5.4 which is 10 years old now.
but you are Running
etc/release PRETTY_NAME=Red Hat Enterprise Linux Server 7.4 (Maipo)
RHEL 7.4

2) it says: version = 14i of nmon
Which is 7 years out of date!

3) There is only 8 data point - I would like 100 snapshots please.
If that takes 200 minutes that is OK with me.

4) The /proc/modules implies this is a VMware virtual machine.
Can you check if the VM is using the maximum CPU it is allowed on the VM host?

Please try again but using the up to date nmon version, thanks Nigel

[Mohamed LAKRAA]

Ok , so for 100 snapshots, i run this command:
~/TMC/monitor/nmon16g_x86_rhel72 -f -t -s 3 -c 100 ?

Or i must change flag to make time ?
I'll send you NMON file tomorrow.

Thanks and regards,

[Mohamed LAKRAA]

Hello,

an new email has been sent (NMON file 100 snapshots) .

Thanks and regards,

[Nigel Griffiths]

Hi,
Thanks for the data.
The virtual machine is not doing much.
There is some paging out to paging space.
Sorry, but I have no ideas left.
nmon is just using the sleep() system call to pause.
If that sleep() is taking nearly 2 minutes then the OS is very sick.
If the collecting of the data is taking 2 minutes then other system calls are having problems.

It is a mystery.
You could try a reboot and perhaps a service pack updated to the latest available.
The reboot miht free up the OS and messed up resources.
Best of luck, Nigel

[Mohamed LAKRAA]

Hi,

Thank you for your answer.
I will try to do what you asked.

Thank you very much.
Regards,
Mohamed LAKRAA

[Nigel Griffiths]

Going to close this request.
If more information becomes available we can continue.