How do we get a chart the mstats metrics from two indexes?

[englotus]

Hi, all, 

I am looking for a way to chart the the average cpu and memory per process for all the hosts in two indexes?

The scenario that I have been working on is that I have a bunch of machines running roughly 4 or 5 java applications per host.... and I am looking for a way to make a time series chart with the average cpu and memory used by each java processes in a dashboard...  

like 

host 1

1. site 1 average cpu chart and average memory chart
2. site 2 average cpu chart and average memory chart
3. site 3 average cpu chart and average memory chart 
4. site 4 average cpu chart and average memory chart 

host 2

1. site 1 average cpu chart and average memory chart
2. site 2 average cpu chart and average memory chart
3. site 3 average cpu chart and average memory chart 
4. site 4 average cpu chart and average memory chart 
5. site 5 average cpu chart and average memory chart 
6. site 6 average cpu chart and average memory chart

I tried to get the top processes via hosts, via this code 

```

| mstats max("NIX.ps_metric.pctCPU") AS CPU,max("NIX.ps_metric.pctMEM") AS MEM WHERE ("index"="first_index" OR "index"="second_index") AND "host"="test_host" span=5m BY host ARGS COMMAND

| top limit=20 COMMAND BY CPU MEM host

| chart values(CPU) as CPU, values(MEM) as MEM over COMMAND BY host

```

but all i get is the top processes like by run command and not the actual processes, they seem to be grouped under the java process....

java      CPU  ,  MEMORY

crond   cpu, memory

bash  cpu, memory

Is there a way to split the monitoring to get something useable?   Ideally, I would like to alert when a particular site is being over worked, like CPU > 85%...

Ta,

Lane