New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
How to find abusive VPS?
My node specs
Quad Core Intel XEON E3-1230 @ 3.2 Ghz (8 cores)
12 GB Ram
1TB Drive Sata3 (6 Gbits)
@ DimeNoc datacenter
I have KVM virtualization (with SolusVM) running on it. I have 4 VPS created and running smoothly. The only issue that I see is, that all these VMs (two 256MB boxes and one 512MB box) have a loadavg of either 0.0 or 0.1 but overall the entire node is showing a loadavg of 1.0 which cannot be right since the node specs are way too muchf or what I am using right now. So somewhere something is causing a load for the node.
My question (and especially to the providers), how do you find abusive vm?
Comments
With a single (sata I suppose) drive I think that the load is I/O related. Do you see some io waiting running top?
htop -d 2 or top in any of the VMs of the host shows no activity or no timed activity (crontabs etc)
I rebooted all VMs and the host load dropped down to 0.01 so there was some process which was stuck or something
My question remains, how to identify a abusing or resource-hog vm without rebooting all vms or the host node itself?
Run top on the node, see which process is stuck running "R" or waiting for io - "D" state.
@rds100 well I am no expert in VPS industry, running a node to keep my tens of vms in one place. Right now I rebooted all VMs (three total at the moment) and the loadavg dropped to '0' so top is not giving me anything
I thought I am missing out on some tools or scripts to find abusers, if there are any, please share it with me (even in PM if you dont want to share in public)
@Asim,
I'd think virt-top (http://people.redhat.com/~rjones/virt-top/) might be useful.
@eLohkCalb
Wow, This is exactly the kind of utilities I was missing. Thanks
@Asim sorry, i don't run KVM VPSes, only OpenVZ, and there it is easy, the SolusVM panel shows you graphs of the load of the node, and which VPS generates how much load.
@rds100 using KVM slaves, SolusVM only shows traffic usage per VM but the load graphs are collective
Now I know when @Francisco used to say that SolusVM has limited KVM support
Still if you ssh on the node and run "top", every VPS should show there as a single KVM process and you should be able to see which process uses much CPU time.
Also iostat should show which lvm volume gets how much reads/writes.
I just found the problem, the VMs running CentOS 6.2 have issues consuming too much CPU. I installed it from CentOS 6.2 minimal ISO and there is nothing else installed over it.
Probably another CentOS 6.x bug? If I turn this VM off, the load drops to 0, if its on, it gradually starts going up until max 1 loadavg
iostat shows no credible io activity
Is it bad to have a load of 1.0 when you have 8 cores?
No not really.
Is this an installed VPS or one that is booting to an ISO still? I do notice I'll get 100% CPU on a core while a VPS idles at the install screen of a boot CD
@miTgiB correct. or when it's stuck on grub boot loader.