| openstackgerrit | Duc Truong proposed openstack/senlin master: [WIP] Clean up devstack tempest jobs https://review.openstack.org/635638 | 00:35 |
|---|---|---|
| openstackgerrit | Duc Truong proposed openstack/senlin master: [WIP] Clean up devstack tempest jobs https://review.openstack.org/635638 | 01:26 |
| *** sapd1_ has quit IRC | 02:13 | |
| openstackgerrit | Duc Truong proposed openstack/senlin master: [WIP] Clean up devstack tempest jobs https://review.openstack.org/635638 | 02:17 |
| *** crownwin has joined #senlin | 02:26 | |
| dtruong | crownwin Unfortunately, Senlin code prevents the user from adding more than 1 lb policy to the same cluster. | 02:27 |
| dtruong | But I do understand your use case | 02:28 |
| dtruong | We would have to change the Senlin code and test it to see how it behaves with 2 lb policies attached to the same cluster. | 02:28 |
| dtruong | crownwin Are you using neutron-lbaas or octavia? | 02:29 |
| crownwin | octavia | 02:32 |
| crownwin | month ago, I talked with you in launchpad. Usecase is when high traffic is comming, my backends (Autoscaling cluster) still fine, but my LB (lb policy) is bottleneck and can't serve traffic | 02:34 |
| *** sapd1 has joined #senlin | 02:38 | |
| crownwin | I know that having many change Senlin code. I tried and couldn't do it. It seems to exceed my ability. | 02:38 |
| dtruong | When you create an instance without senlin to support your use case, you add the instance as a member to two different load balancers? | 02:54 |
| *** sapd1 has quit IRC | 03:34 | |
| crownwin | but that's very manualy | 03:42 |
| crownwin | I want everything is automatic because number of nodes is very large | 03:43 |
| crownwin | How can I contribute to that features? | 03:44 |
| crownwin | I will willing to do that | 03:47 |
| *** sapd1 has joined #senlin | 03:51 | |
| *** irclogbot_3 has quit IRC | 04:55 | |
| dtruong | Sure, I can help you. You can start by adding this line (https://github.com/openstack/senlin/blob/master/senlin/policies/scaling_policy.py#L133) to the __init__ of the load balancer policy here: https://github.com/openstack/senlin/blob/master/senlin/policies/lb_policy.py#L290 | 05:10 |
| dtruong | This will allow you to add two lb policies to the same clusters. Let me know what happens after you test that with your use case. | 05:11 |
| openstackgerrit | Duc Truong proposed openstack/senlin master: Clean up devstack tempest jobs https://review.openstack.org/635638 | 05:15 |
| *** crownwin has quit IRC | 05:45 | |
| *** crownwin has joined #senlin | 06:52 | |
| crownwin | @Truong, I changed it before and I had a problem. | 07:19 |
| openstackgerrit | Merged openstack/senlin-tempest-plugin master: Check required health policy version https://review.openstack.org/635621 | 07:20 |
| crownwin | When I scale out cluster, new created node only be added to one lb, another lb have no action. | 07:22 |
| openstackgerrit | Merged openstack/senlin-tempest-plugin master: Check required microversion https://review.openstack.org/635634 | 07:33 |
| openstackgerrit | Merged openstack/senlin-tempest-plugin master: Generate profile spec baed on tempest config https://review.openstack.org/635648 | 07:33 |
| openstackgerrit | Merged openstack/senlin master: Fix cluster recovery and node recovery params https://review.openstack.org/636811 | 07:33 |
| *** crownwin has quit IRC | 07:54 | |
| *** crownwin has joined #senlin | 07:54 | |
| crownwin | @dtruong, I discovered these lines caused issue with my use case (https://github.com/openstack/senlin/blob/master/senlin/policies/lb_policy.py#L529-L535) | 08:37 |
| crownwin | my use case, node must contain all its lb, which means that node_data.get('lb_member', None) return a array contain list of lbs instead of one lb_member. | 08:48 |
| openstackgerrit | zhongshengping proposed openstack/senlin master: add python 3.7 unit test job https://review.openstack.org/637747 | 09:06 |
| openstackgerrit | zhongshengping proposed openstack/python-senlinclient master: add python 3.7 unit test job https://review.openstack.org/637755 | 09:06 |
| *** sapd1 has quit IRC | 09:09 | |
| *** sapd1 has joined #senlin | 09:22 | |
| *** crownwin has quit IRC | 10:57 | |
| *** crownwin has joined #senlin | 12:18 | |
| *** jmlowe has quit IRC | 13:41 | |
| *** jmlowe has joined #senlin | 14:07 | |
| *** crownwin has quit IRC | 14:21 | |
| *** jmlowe has quit IRC | 15:23 | |
| *** jmlowe has joined #senlin | 15:25 | |
| *** jmlowe has quit IRC | 15:34 | |
| *** jmlowe has joined #senlin | 15:45 | |
| *** jmlowe has quit IRC | 19:37 | |
| *** jmlowe has joined #senlin | 19:38 | |
| *** openstackgerrit has quit IRC | 20:09 | |
| *** openstackgerrit has joined #senlin | 20:45 | |
| openstackgerrit | Duc Truong proposed openstack/senlin master: [DNM] Trying out multiple lb policies https://review.openstack.org/638001 | 20:45 |
| *** ChanServ sets mode: +v openstackgerrit | 20:46 | |
| dtruong | crownwin ^ Try out this patch. It changes the node data to contain a map of lb_id to member_id. | 20:46 |
| dtruong | I did not test this patch set, so feel free to fix it if it does not work. | 20:47 |
| *** jmlowe has quit IRC | 20:57 | |
| openstackgerrit | Duc Truong proposed openstack/senlin master: Fix node delete with lifecycle hook bug https://review.openstack.org/638035 | 22:47 |
| *** jmlowe has joined #senlin | 23:40 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!