| opendevreview | Merged openstack/octavia-tempest-plugin master: Fix legacy admin in RBAC tests https://review.opendev.org/c/openstack/octavia-tempest-plugin/+/872648 | 09:39 |
|---|---|---|
| opendevreview | Tom Weininger proposed openstack/python-octaviaclient stable/yoga: Fix new pylint warnings https://review.opendev.org/c/openstack/python-octaviaclient/+/876263 | 09:44 |
| opendevreview | Tom Weininger proposed openstack/python-octaviaclient stable/xena: Fix new pylint warnings https://review.opendev.org/c/openstack/python-octaviaclient/+/875504 | 09:45 |
| opendevreview | Tom Weininger proposed openstack/python-octaviaclient stable/wallaby: Fix new pylint warnings https://review.opendev.org/c/openstack/python-octaviaclient/+/875505 | 09:46 |
| opendevreview | OpenStack Release Bot proposed openstack/octavia stable/2023.1: Update .gitreview for stable/2023.1 https://review.opendev.org/c/openstack/octavia/+/876299 | 11:15 |
| opendevreview | OpenStack Release Bot proposed openstack/octavia stable/2023.1: Update TOX_CONSTRAINTS_FILE for stable/2023.1 https://review.opendev.org/c/openstack/octavia/+/876300 | 11:15 |
| opendevreview | OpenStack Release Bot proposed openstack/octavia master: Update master for stable/2023.1 https://review.opendev.org/c/openstack/octavia/+/876301 | 11:15 |
| opendevreview | Merged openstack/octavia-tempest-plugin master: Reduce the number of load balancers in PoolAPITest https://review.opendev.org/c/openstack/octavia-tempest-plugin/+/771166 | 13:13 |
| opendevreview | Merged openstack/octavia stable/2023.1: Update .gitreview for stable/2023.1 https://review.opendev.org/c/openstack/octavia/+/876299 | 13:25 |
| opendevreview | Merged openstack/octavia master: Update master for stable/2023.1 https://review.opendev.org/c/openstack/octavia/+/876301 | 13:34 |
| opendevreview | Omer Schwartz proposed openstack/octavia master: Add h2 section to Octavia cookbook https://review.opendev.org/c/openstack/octavia/+/874652 | 14:55 |
| opendevreview | Omer Schwartz proposed openstack/octavia master: Add --wait to Octavia cookbook https://review.opendev.org/c/openstack/octavia/+/876392 | 14:55 |
| servagem | Hello | 18:02 |
| servagem | I am facing some performance issues with loadbalancer. My current cloud has only one flavor for octavia with 1 vCPU, 1 GB RAM 8 GB disk and no "openstack loadbalancer flavor". I would like to have other flavors to support some workloads that needs additional performance. If I understand correctly, I should create an loadbalancer flavorprofile for that, like: | 18:02 |
| servagem | openstack flavor create .... large_flavor_id | 18:02 |
| servagem | openstack loadbalancer flavorprofile create --name large --provider amphora --flavor-data '{"compute_flavor": "large_flavor_id"}' | 18:02 |
| servagem | openstack loadbalancer flavor create --flavorprofile large --description '2 vcpu, 2048 ram' --enable --name large | 18:02 |
| servagem | Then, I can choose which flavor I would like to use during openstack loadbalancer create command. Is that correct? | 18:03 |
| johnsom | Correct | 18:03 |
| servagem | great, thank you johnsom! | 18:04 |
| johnsom | We would also recommend that your amphora image have haproxy 2.x in it as the older versions don't use multiple CPUs well | 18:04 |
| servagem | yes, I am using ubuntu focal for amphora images. It comes with version 2.0.xx | 18:05 |
| johnsom | What is your performance goal? Mostly curious | 18:05 |
| servagem | api that handles 2000 requests per second | 18:08 |
| johnsom | You should easily be able to hit that with 1 vCPU | 18:08 |
| johnsom | Keep an eye on the CPU and RAM usage inside the amphora, if it's not close to 100% usage, the bottleneck is not the load balancer, it's something else in the deployment. | 18:10 |
| servagem | right. we are still investigating to find the bottleneck. We tested using an regular instance with more cpu and ram. We configured the haproxy manually. It had a better performance. That is why I would like to test using a lb flavor with addicionar resource. | 18:12 |
| johnsom | Make sure your VMs are using kvm and not tcg (software VM). | 18:15 |
| servagem | yep | 18:15 |
| johnsom | Just trying to think of what would have that big of an impact on performance. You should be able to handle a lot more RPS than that. | 18:15 |
| servagem | ok. It's good to know. We are investigating other components too. | 18:19 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!