I was looking for a way to access my nodes' CPU usage, bandwidth, and disk i/o rate and did not see anything in the API that would allow me access to this. The only way I can see this data is in the linode manager and cant really extract any of that data because its only an image and not text. I find myself needing to know things like what the last CPU usage percentage was or the last disk i/o rate was. Since the alerting only kicks in after two hours of exceeding set limits this would help me to sooner realize any issues with CPU,disk i/o,etc.

You can setup something like Munin or others... Just a few of many, including ones you can create and setup yourself.

_________________
Kevin a.k.a. Dweeber

Linode has previously refused to put this information in the API as they don't want people polling it for graphs, and because it's all information that is accessible from inside the Linode itself, where it can be graphed by stuff like Munin.

The problem with relying on data from within the node is if the CPU is maxed out or the disk i/o rate is really high then accessing the node is impossible as the resources have been exhausted. For example if my CPU usage is at 400% and thus using 100% of all 4 CPUs then I cant get any data directly from the node. I cant even login. But if the linode manager is reporting this then it would not beneficial to give access to this data via an API call?

My understanding is that if your linode has four processes running, each at 100 percent CPU, they get 4/4 of a CPU each. When the next cpu-intensive process starts each would receive 4/5 of a CPU, or 80 percent. If there were 8 cpu-intensive tasks running each would receive 4/8 of a CPU or 50 percent.

There is niceness to consider and task switching also, but this is the basic idea as I understand it. So if you tried to start another process it would in fact run, just a bit slower than if CPU were not maxed out.

James

hmm, lost all my text. in a nutshell, if your node stops responding to data requests, then it's an alarm and you need to look at it. It sounds like your node is mis configured if it stops responding all the time, rebooting it isn't the answer.