Thanks Lahiru for looking into this..yah..+1 for the above fix..
Thanks,
Reka
On Sun, Dec 21, 2014 at 4:30 PM, Lahiru Sandaruwan wrote:
> I think we can solve it better by adding an "else" part to the logic,
>
> We can increase the required number instance by one in the case of
> "max=min". The
I think we can solve it better by adding an "else" part to the logic,
We can increase the required number instance by one in the case of
"max=min". Then it will send the maxOut event in rule evaluation since it
is checked already. I'll change the commit as follows,
if(instanceRange != 0){
fl
Hi Reka,
On Sun, Dec 21, 2014 at 1:18 PM, Reka Thirunavukkarasu
wrote:
> Hi Lahiru,
>
> I saw the following code when calculating required instances. Since it is
> checking the instanceRange > 0, whenever the cluster instance reaches it
> maximum, this instanceRange is 0.
>
We get the instance
Please find the code segment as below:
public int getNumberOfInstancesRequiredBasedOnMemoryConsumption(float
threshold, double predictedValue,
int
max, int min) {
double numberOfAdditionalInstancesRequired = 0;
if(
Hi Lahiru,
I saw the following code when calculating required instances. Since it is
checking the instanceRange > 0, whenever the cluster instance reaches it
maximum, this instanceRange is 0. At that point even though the load is
high no decision is taken as the required instances is not getting
c