AnsweredAssumed Answered

HCP Indexing task runs for 12+ hours, duration says 1.5 hours

Question asked by Jason vanValkenburgh on Jun 13, 2017
Latest reply on Sep 26, 2017 by Jared Cohen

Hi, we have an existing HCP workflow that uses four HCP data connections as inputs; each connection rolls up an HCP path of about 500K-1M emails. 


We have been running the job for more than a day (with an 8-node HCI system and an 8-node HCP), and the job has only processed 600K objects.   However, the duration on the job status indicates a runtime of only 1.5 hours (and still processing a batch of ~25K documents for about 9 hours) as opposed to the calendar time of ~12 hours.  There aren't any other jobs running and the system is essentially idle. 


How is this duration calculated, and is it indicative of any issues with the environment?


We have other workflows that use the MQE connector, and in all cases they're running fine up to what we believe to be the last batch (e.g. 99.98% of documents are processed), and the job then sits there running permanently and never completes.  Pausing and resuming the job does not change the behavior; if we clear and restart the task then the job runs quickly and then stops again at the 99.98% complete mark.


Any thoughts appreciated,