Red Hat Bugzilla – Bug 1578996
[RHV] When Graph refresh is ON, RHV provider refresh time is longer
Last modified: 2018-07-12 09:15:34 EDT
New commits detected on ManageIQ/manageiq-providers-ovirt/gaprindashvili: https://github.com/ManageIQ/manageiq-providers-ovirt/commit/64e7cd3bdbd67f5c83e86abcb64eb092cc6e15ef commit 64e7cd3bdbd67f5c83e86abcb64eb092cc6e15ef Author: Piotr Kliczewski <piotr.kliczewski@gmail.com> AuthorDate: Mon May 14 04:37:30 2018 -0400 Commit: Piotr Kliczewski <piotr.kliczewski@gmail.com> CommitDate: Mon May 14 04:37:30 2018 -0400 Merge pull request #237 from borod108/refresh_huge Performance improvements for graph refresh (cherry picked from commit d631734debc7a84e8aa8b8f32fd0f28c10e9adaf) Fixes https://bugzilla.redhat.com/show_bug.cgi?id=1578996 app/models/manageiq/providers/redhat/inventory_collection_default/infra_manager.rb | 6 +- spec/models/manageiq/providers/redhat/infra_manager/refresh/refresh_recording_modifier_spec.rb | 58 + spec/models/manageiq/providers/redhat/infra_manager/refresh/refresher_huge_async_spec.rb | 141 + spec/support/modify_refresh_yml_recording.rb | 214 + spec/vcr_cassettes/manageiq/providers/redhat/infra_manager/refresh/ovirt_sdk_refresh_recording_for_mod.yml | 2819 + 5 files changed, 3236 insertions(+), 2 deletions(-) https://github.com/ManageIQ/manageiq-providers-ovirt/commit/9f425673f033459f51761df80cd17ace5cedc188 commit 9f425673f033459f51761df80cd17ace5cedc188 Author: Boris Od <boris.od@gmail.com> AuthorDate: Wed May 16 02:12:08 2018 -0400 Commit: Boris Od <boris.od@gmail.com> CommitDate: Wed May 16 02:12:08 2018 -0400 Merge pull request #240 from agrare/folder_bulk_connect Improve performance of EmsFolders and EmsClusters child save blocks (cherry picked from commit 4c764a9af5b8339ca0aa2c50490605e29c0260d2) https://bugzilla.redhat.com/show_bug.cgi?id=1578996 app/models/manageiq/providers/redhat/inventory_collection_default/infra_manager.rb | 25 +- 1 file changed, 7 insertions(+), 18 deletions(-)
Moving to verified based on the following test: Here are testing details, towards turning RHV Graph refresh ON, by default. Test purpose: Time measurement of full refresh for Graph refresh turned ON, to examine the RHV Graph refresh optimizations, added in CFME-5.9.3. Tested parts: 2 CFME-5.9.3 machines (Memory 8G, 4 cores) (in US):: 1. With Graph refresh OFF (default) 2. With Graph refresh ON. Scale RHV-4.2.3.5-0.1.el7 (in RDU): b01-h21-r620.rhev.openstack.engineering.redhat.com Clusters 4 Hosts 403 Datastores 6 Virtual Machines 4059 Templates 13 Test results: Graph refresh OFF: 1st refresh: 19 min; 1st removal 17 min 2nd refresh: 16 min; 2nd removal 1 min Graph refresh ON: 1st refresh: 15 min; 1st removal 41 min 2nd refresh: 15 min; 2nd removal 1 min Conclusion: Graph refresh seem to provide refresh time improvement. Issues: The 1st removal time of the RHV provider is quite large: GR OFF: 17 min, and for GR ON it's even larger GR ON: 41 min. Notes: * Both RHV & CFME located in the US, so these tests are with no latency. * Refresh time is taken from evm.log (messages: Refreshing targets for EMS... & Refreshing targets for EMS...complete) * Removal time is taken from the CFME UI tasks page, task: "Destroying ManageIQ::Providers::Redhat::InfraManager with id: " * The overall time measurement is larger than those measured in previous testing, using RHV in the same VMs/Hosts scale. Testing recommendation: Repeat this test with high latency.
Created attachment 1440964 [details] gr_off_evm_log.tgz
Created attachment 1440979 [details] gr_off_automation_log.tgz
Created attachment 1440980 [details] gr_on_evm_log.tgz
Created attachment 1440982 [details] gr_on_automation_log.tgz
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHSA-2018:2184