Public bug reported: I saw listing security group is slow and causing cpu spikes unexpectedly,
I run a Rock neutron-server with api worker set to 1, when executing command like ```console root@mgt01:~# time curl -H "x-auth-token: $token" http://neutron-server.openstack.svc.region-stackdev.myinspurcloud.com:9696/v2.0/security_groups -owide -s real 0m2.328s user 0m0.016s sys 0m0.012s root@mgt01:~# curl -H "x-auth-token: $token" http://neutron-server.openstack.svc.region-stackdev.myinspurcloud.com:9696/v2.0/security_groups | jq '.security_groups | length' % Total % Received % Xferd Average Speed Time Time Time Current Dload Upload Total Spent Left Speed 100 497k 100 497k 0 0 219k 0 0:00:02 0:00:02 --:--:-- 219k 225 ``` It will return in around 2 seconds. There are around 200 security groups, so maybe it is not extremely slow, but what is interesting is that calling this rest api seems making cpu spikes for neutorn-server pod, ```console CONTAINER ID NAME CPU % MEM USAGE / LIMIT MEM % NET I/O BLOCK I/O PIDS 8a30733e3932 k8s_neutron-server_neutron-server-787dcd7964-2zxt5_openstack_71cbb9bc-4530-11eb-bcc6-525400d22fc9_0 92.83% 1020MiB / 2.441GiB 40.81% 0B / 0B 0B / 16.4kB 8 ``` I am wondering why security group listing is cpu bound? ** Affects: neutron Importance: Undecided Status: New -- You received this bug notification because you are a member of Yahoo! Engineering Team, which is subscribed to neutron. https://bugs.launchpad.net/bugs/1909160 Title: high cpu usage when listing security groups Status in neutron: New Bug description: I saw listing security group is slow and causing cpu spikes unexpectedly, I run a Rock neutron-server with api worker set to 1, when executing command like ```console root@mgt01:~# time curl -H "x-auth-token: $token" http://neutron-server.openstack.svc.region-stackdev.myinspurcloud.com:9696/v2.0/security_groups -owide -s real 0m2.328s user 0m0.016s sys 0m0.012s root@mgt01:~# curl -H "x-auth-token: $token" http://neutron-server.openstack.svc.region-stackdev.myinspurcloud.com:9696/v2.0/security_groups | jq '.security_groups | length' % Total % Received % Xferd Average Speed Time Time Time Current Dload Upload Total Spent Left Speed 100 497k 100 497k 0 0 219k 0 0:00:02 0:00:02 --:--:-- 219k 225 ``` It will return in around 2 seconds. There are around 200 security groups, so maybe it is not extremely slow, but what is interesting is that calling this rest api seems making cpu spikes for neutorn-server pod, ```console CONTAINER ID NAME CPU % MEM USAGE / LIMIT MEM % NET I/O BLOCK I/O PIDS 8a30733e3932 k8s_neutron-server_neutron-server-787dcd7964-2zxt5_openstack_71cbb9bc-4530-11eb-bcc6-525400d22fc9_0 92.83% 1020MiB / 2.441GiB 40.81% 0B / 0B 0B / 16.4kB 8 ``` I am wondering why security group listing is cpu bound? To manage notifications about this bug go to: https://bugs.launchpad.net/neutron/+bug/1909160/+subscriptions -- Mailing list: https://launchpad.net/~yahoo-eng-team Post to : [email protected] Unsubscribe : https://launchpad.net/~yahoo-eng-team More help : https://help.launchpad.net/ListHelp
