...
In order to handle larger requests the get samples processes its response in smaller steps. This means that it will first read 100,000 time series headers and all samples they are part of, then those full samples are read and written to the xml response. Then the next 100,00 000 headers are processed, this continues until all headers are processed. If there are less than 100,000 headers all samples will be processed at the same time. All samples that are fully written to the response are kept track of so that subsequent sets of 100,000 headers will not include the previously written samples again.
Because the time series headers are processed step by step it is not known what the headers and samples for the next step are going to be. Therefore the samples will not appear in a predictable order.
...