[Users] Monitoring instances

[Simon Boulet]

Hi Edward,

On Wed, Nov 13, 2013 at 10:59 AM, Edward Konetzko <konetzed at gmail.com>wrote:

> Anyone have that they have setup to monitor instances?  I am looking for
> something that can give me metrics on cpu, disk (io,space), memory and
> network.  Ideally the script would be smart enough to get the limits of the
> vz along with the utilization values.  I have seen a few plugins for
> monitoring systems such as collectd but none of them gather all the
> information needed so I thought I would see if anyone here had what I was
> looking for.
>
>
Here we use Host sFlow [1]. It collects various hypervisor metrics and
publish them over the network. Ganglia can be used as a metric collector
for Host sFlow [2].

The KVM-enabled version of Host sFlow uses libvirt to gather the VM
metrics. You can use my unofficial Libvirt-Openvz driver [3] it will gather
some OpenVZ instances metrics for you (container disk usage, disk I/O, and
vSwap containers memory usage).

This solution is not for the faint of heart, it does requires some hacking
and patching, if you're looking for an out of the box solution, this is
certainly NOT a solution for you. However, I have this working to monitor
our environment and it works great, scales very well, and has proven to be
very reliable so far.

Simon

[1] http://host-sflow.sourceforge.net/
[2] http://blog.sflow.com/2010/10/ganglia.html
[3] https://github.com/siboulet/libvirt-openvz
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openvz.org/pipermail/users/attachments/20131114/667cb75e/attachment.html>