Describe the bug

When querying a metric, as well as when grouping by node, the list of metrics is of random order. This behavior is not the same as with graphite. This makes the usage of DivideSeriesLists require some form of ordering function on both series list otherwise the result changes at each query.

To Reproduce

This query
divideSeriesLists(groupByNode(group(alias(constantLine(1), 'serie.a'), alias(constantLine(1), 'serie.b')),1),groupByNode(group(alias(constantLine(1), 'serie.a'), alias(constantLine(1), 'serie.b')),1))

should respond always
divideSeries(a,a)
divideSeries(b,b)

but in practice the result are combinations that changes at each query.

Version

latest