All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
OpenVZ mis-reporting load average?
Hey, so I'm trying to figure out why my container was stopped on an OpenVZ provider. It was running Nginx with a handful of sites and about a dozen users, maybe 22MB of ram usage (with a few hundred MB of cache) and a load average of 0.13. I have a gig of ram provisioned, and 1 vCPU, I don't see any load spikes besides the load average going up (no notable disk I/O or network usage, CPU hovering around 1 to 4%, ram running right about average), but the load average shot up to 14.
Nothing seems out of order, is there something I'm missing? The provider just says the load average shot up, and took some corrective action and called it a bug in OpenVZ, but how can I be sure that erroneous load averages won't screw me over again?
Comments
Could be other users on the node, that's what it sounds like!
Most likely someone spiking the load
Is this a known OpenVZ issue?
If the node load is very high, then your vps load will spike randomly yes!
Huh, that was probably the case, any recommendations @linuxthefish for a West Coast Xen/KVM VPS provider that offers HA?
It could have been caused by disk IO. If the IO on the node is being abused then any little amount of disk IO you use will cause your load to spike up. This could also mean your VPS was hogging the disk IO, without being logged into the node I can't tell you if it was your VPS or another client's VPS. A good rule of thumb though is if the loads on the node were spiking and powering off/rebooting your VPS fixed the problem then your VPS was probably he cause. Have you considered adding any monitoring to your VPS so you can track the resources and so you have something to show your provider?
@kujoe I already have munin set up, it shows the load average spike, but no spike in disk I/O, ram, cpu or network usage, this matches the builtin statistics graph the provider has in their control panel. I'd be a tad suprised if it was an I/O spike that took it down, supposedly I'm on 100% SSD based storage.