Issue with counter metrics for large number of keys

[Gaurav Luthra]

I want new counter for every key of my windowed stream, And I want the same
counter to get increment when the same key comes multiple times in incoming
event.

So, I will write below code for every incoming event.
getRuntimeContext().getMetricGroup().counter(myKey).inc();

But above code fails when same value of myKey comes. As this is limitation
of flink metrics. It throws exception, "Name collision: Adding a metric
with the same name as a metric subgroup: "

Note: myKey is of String type and can have different value for every
incoming event.
In my case I expect 100 thousands different values of myKey.

Now, To solve the issue, I have to keep the reference of 100 thousands
values of myKey in some data structure e.g.
 Map<String, Counter> myMetricMap;
 and for every myKey I have to do below.

        Counter counter = myMetricMap.get(myKey);
        if (null == windowMetricGauge)
        {
            Counter counter = new counter();
counter.inc();
            myMetricMap.put(myKey, counter);
            getRuntimeContext().getMetricGroup().counter(myKey,counter);
        }
else
{
counter.inc();
}
Above code suffice my purpose. But I do not want to maintain map of 100
thousands keys values of myKey.

Is there any alternate solution? I am looking for a solution where I
achieve above functionality to maintain approx. 100 thousands counter
metrics without keeping their reference in map (or any other data
structure).


Thanks & Regards
Gaurav Luthra
Mob:- +91-9901945206