Description of problem: Compared to 5.5.3.4 (99th Percentile timing for an initial refresh was 737s, refresh on same provider now takes 1600s) (Not including the time for the VIMBroker to load data) Version-Release number of selected component (if applicable): 5.6.0.8 How reproducible: Steps to Reproduce: 1. Add xlarge provider 2. Perform a Refresh and measure timing/memory usage 3. Actual results: ~1600s for the EmsRefresh.refresh message to complete Expected results: Additional info: [----] I, [2016-05-31T21:43:49.642994 #3764:633990] INFO -- : MIQ(ManageIQ::Providers::Vmware::InfraManager::Refresher#refresh) EMS: [vmware-xlarge], id: [1] Refreshing targets for EMS...Complete - Timings {:server_dequeue=>0.004362583160400391, :get_ems_data=>2.9800798892974854, :get_vc_data=>84.45335388183594, :get_vc_data_ems_customization_specs=>4.1711437702178955, :filter_vc_data=>0.00046706199645996094, :get_vc_data_host_scsi=>513.5504930019379, :collect_inventory_for_targets=>605.1574783325195, :parse_vc_data=>44.3983256816864, :parse_targeted_inventory=>44.398534297943115, :db_save_inventory=>931.6641547679901, :save_inventory=>931.6641728878021, :ems_refresh=>1581.2208054065704}
The a large chunk of the increased time appears to be introduced by code that reduces the memory utilization of the broker, https://github.com/ManageIQ/manageiq/commit/4bbad2141b4ec820b56885584abd67ba94cf5569 (note get_vc_data_total is equivalent to collect_inventory_for_targets) 5.5.5.4: :get_vc_data_total=>48.140045166015625 :total_time=>336.6637680530548 5.6.2.1: :collect_inventory_for_targets=>737.8142273426056 :ems_refresh=>1119.3035068511963 5.6.2.1 with 4bbad21 reverted :collect_inventory_for_targets=>44.61478614807129 :ems_refresh=>517.089364528656 Save inventory performance appears to have regressed as well, more investigation is needed into this
https://github.com/ManageIQ/manageiq-gems-pending/pull/17
*** This bug has been marked as a duplicate of bug 1378998 ***