I'm working towards creating an ML job through the Kibana UI as well as using the ML wrapper API that we developed internally.
Using the wrapper, first I'm indexing all the test data and then starting the job as well as the datafeed. However, this is not letting to process all the records. If I use the same data index and create a job using the UI with same configuration parameters, the whole data index set is being processed.
This image shows the difference in number of records processed for the same data set- particularly, last two ML jobs are configured with the same parameters (including the query delay and frequencey) but still there is a difference in number of records processed as well as the latest date stamp.
The job named test is created using UI and processed all the 322 indexed records but the other jobs, which are created through the wrapper API only processed partial data.
PS: There is no invalid data as well.
Please let me know where the issue is and what could be the possible solution. I will be happy to provide more information as needed.