OpenVZ Forum: Support » *SOLVED* SAR and MRTG to monitor the VPSs

Home » General » Support » *SOLVED* SAR and MRTG to monitor the VPSs

Show: Today's Messages :: Show Polls :: Message Navigator
E-mail to friend

*SOLVED* SAR and MRTG to monitor the VPSs [message #4638]

Thu, 20 July 2006 00:16

luismi
Messages: 8
Registered: July 2006
Location: Dublin

Junior Member

Hi all,

I am thinking on create some scripts to create some graphs with mrtg and sar (from sysstat package) in order to view how is the cpu, the memory, the network...

Well,
I have some dudes that I would like to clarify before start to work.

If I install the sysstat package in every VPS.. am I going to see the same values of the host system or evert VPS is going to collect different values (that is what I want)?

I know I could do the experiment but I would like to know if anyone there knows the answer. (I installed sysstat in all my VPSs right now, so I need to wait a while yet to see the results/differences between the metrics from the host system and the VPSs)

If the answer is that sysstat is going to show me the same values for the host system and the VPSs... is there anyway to collect individual values (memory, cpu, network, disk i/o..) of every VPS?
For example, reading /proc? using vmstat, uptime or similar commands?

Any other idea will be appreciated.

Thanks!

[Updated on: Tue, 22 May 2007 08:39] by Moderator

Report message to a moderator

Re: SAR and MRTG to monitor the VPSs [message #4657 is a reply to message #4638]

Thu, 20 July 2006 17:58

luismi
Messages: 8
Registered: July 2006
Location: Dublin

Junior Member

Sysstat version: 5.0.6-5

Is not working in the VPSs I don't know the reason yet.
can anyone there give me a hand?
I run debian sarge as host system and debian sarge as guest system

[Updated on: Thu, 20 July 2006 17:59]

Report message to a moderator

Re: SAR and MRTG to monitor the VPSs [message #4662 is a reply to message #4657]

Fri, 21 July 2006 05:43

Vasily Tarasov
Messages: 1345
Registered: January 2006

Senior Member

What does it output, when it doesn't run?
I there something in logs?

Report message to a moderator

Re: SAR and MRTG to monitor the VPSs [message #4665 is a reply to message #4662]

Fri, 21 July 2006 07:42

luismi
Messages: 8
Registered: July 2006
Location: Dublin

Junior Member

Hi,

It is strange.
If I execute "sadc 1 10" it returns a segmentation fault.

Latest version of sysstat 7.0.0 doesn't fail but I didn't check it yet, so I don't know if the latest version is able to collect the information correctly.

I will do more research.

Any idea or comment will be welcome.

Report message to a moderator

Re: SAR and MRTG to monitor the VPSs [message #4666 is a reply to message #4665]

Fri, 21 July 2006 07:49

Vasily Tarasov
Messages: 1345
Registered: January 2006

Senior Member

Quote:

If I execute "sadc 1 10" it returns a segmentation fault.

Please, check /proc/user_beancounters ouptput after such unsuccessful try.

Quote:

Latest version of sysstat 7.0.0 doesn't fail but I didn't check it yet, so I don't know if the latest version is able to collect the information correctly.

I suppose, that output of sysstat will be not correct in terms of "per VE"...

Report message to a moderator

Re: SAR and MRTG to monitor the VPSs [message #4687 is a reply to message #4666]

Fri, 21 July 2006 16:55

luismi
Messages: 8
Registered: July 2006
Location: Dublin

Junior Member

I will check it when I arrive at home.
Any other way to collect information per VE?

Maybe reading from the /proc of a VE with a cron job and a smaill script?

Report message to a moderator

Re: SAR and MRTG to monitor the VPSs [message #4696 is a reply to message #4687]

Fri, 21 July 2006 20:13

Vasily Tarasov
Messages: 1345
Registered: January 2006

Senior Member

Well the problem is that in OpenVZ some /proc files are virtualized and other aren't. And all statistic programms use /proc to get information.
For example, /proc/meminfo file isn't virtualized (it means, that inside VE you see information about the whole system)
There is however places there you can get something like this information. For instance, /proc/user_beancounters file contains information (in second column) how much resources (kernel memory, private memory pages, sockets, etc.) this VE holds.

HTH, Rolling Eyes