creating 100 rules keeps around 60 seconds

Bug #1481761 reported by Leontii Istomin
6
This bug affects 1 person
Affects Status Importance Assigned to Milestone
Mirantis OpenStack
Invalid
Medium
Elena Ezhova

Bug Description

Please look at reports (rally_report.html) for secgroups rally scenarios from here: http://mos-scale-share.mirantis.com/rally_report_and_logs.tar.gz
from link like whis "rally_report.html#/NeutronSecGroupPlugin.boot_and_delete_server_with_secgroups/details"
you can see that neutron.create_100_rules takes around 60 seconds. I believe we need to try to find out where is bottleneck in this action.

cluster configuration:
Baremetal,Ubuntu,IBP,HA,Neutron-vlan,Ceph-all,Nova-debug,Nova-quotas,7.0-98
Controllers:3 Computes:180 Copmutes+Ceph: 20

api: '1.0'
astute_sha: 488db988a1f2e18f99decf417371c50b2a7fb794
auth_required: true
build_id: 2015-07-31_09-31-33
build_number: '764'
feature_groups:
- mirantis
fuel-agent_sha: b27076b8d87719507bd0f7c225858a64799c5be7
fuel-library_sha: 31651882b342e8bcc4991ab76af8342ababed23f
fuel-nailgun-agent_sha: 1512b9af6b41cc95c4d891c593aeebe0faca5a63
fuel-ostf_sha: f73608d0e31d0a02a7fad6c048e931b7deef148e
fuelmain_sha: e29f2061590846e4628aa834e59518268c0bb6c1
nailgun_sha: 5be2c44f0b3e7485d397df0ac816b9860e7f30ff
openstack_version: 2015.1.0-7.0
production: docker
python-fuelclient_sha: 71bb8fa87ee25f0c1bb84317884da7c917902a63
release: '7.0'

Diagnostic Snapshot: http://mos-scale-share.mirantis.com/fuel-snapshot-2015-08-03_10-03-58.tar.xz

Tags: scale
affects: fuel → mos
Changed in mos:
milestone: none → 7.0
assignee: nobody → MOS Neutron (mos-neutron)
importance: Undecided → Medium
status: New → Confirmed
Elena Ezhova (eezhova)
Changed in mos:
assignee: MOS Neutron (mos-neutron) → Elena Ezhova (eezhova)
Revision history for this message
Ilya Shakhat (shakhat) wrote :

Function "neutron.create_100_rules" makes 100 requests to Neutron, the default concurrency defined in the scenario is 20. The total execution time for a single call is 60 seconds, meaning that 1 single query takes 0.6 second. Taking into account concurrency Neutron cluster processes 20 queries in 0.6 seconds, or for one node this gives 6.6 queries in 0.6 seconds which gives in average 0.1 seconds per query. This is quite perfect result!

Results from similar jobs are even better:
** 24.3 seconds (http://mos-scale.vm.mirantis.net/test_results/build_7.0-764/jenkins-11_env_run_rally-84-ubuntu-MSK-2015-08-04-19:12:55-2015-08-05-16:05:25/rally_report.html#/NeutronSecGroupPlugin.boot_and_delete_server_with_secgroups/overview),
** 25 seconds (http://mos-scale.vm.mirantis.net/test_results/build_7.0-764/jenkins-11_env_run_rally-82-ubuntu-MSK-2015-08-02-08:04:30-2015-08-03-07:16:41/rally_report.html#/NeutronSecGroupPlugin.boot_and_delete_server_with_secgroups/overview)
** 18.9 seconds (http://mos-scale.vm.mirantis.net/test_results/build_7.0-98/jenkins-12_env_run_rally-83-ubuntu-MSK-2015-08-05-02:42:34-2015-08-05-15:35:38/rally_report.html#/NeutronSecGroupPlugin.boot_and_delete_server_with_secgroups/overview)

All these numbers look acceptable, thus moving the bug into Invalid

Changed in mos:
status: Confirmed → Invalid
Revision history for this message
Sergey Belous (sbelous) wrote :
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.