| wncslln | morning o/ I sent a problem that I'm having with loadbalancer on friday. i will send it again, if anyone could help me, i appreciate it | 12:47 |
|---|---|---|
| wncslln | i have an environment deployed with kolla-ansible and i has encountered an error which amphore stuck in PENDING_CREATE and after a while status changes to ERROR | 12:47 |
| wncslln | also amphora VMs stuck in BOOTING status | 12:47 |
| wncslln | bug is similar to this https://bugs.launchpad.net/kolla-ansible/+bug/1847905 | 12:47 |
| gthiemonge | wncslln: Hi, I think it's directly related to the way kolla-ansible creates the Octavia management network (https://docs.openstack.org/kolla-ansible/latest/reference/networking/octavia.html#networking) | 13:23 |
| gthiemonge | wncslln: maybe you should ask the kolla Folks directly | 13:24 |
| wncslln | gthiemonge: i first asked here because in the end of the bugreport Dincer said is not a kolla related, but i will send there | 13:31 |
| wncslln | thanks for the help!! | 13:31 |
| gthiemonge | wncslln: do you have the logs of the octavia worker? | 13:32 |
| wncslln | i doing another try before any configuration changes, im building a new amphora image | 13:32 |
| wncslln | gthiemonge: yeah, one moment | 13:32 |
| gthiemonge | wncslln: do you see "ERROR.*ConnectTimeout: HTTPSConnectionPool" (not WARNING)? | 13:32 |
| kevko | gthiemonge: btw, if you have a octavia configured for amphora driver (which is now v2 path ) ...and you don't have octavia_persistence database ... what will be the result of LB create ? | 13:34 |
| kevko | i mean "octavia driver" for amphora (octavia changed this from v1 to v2 in xena ) | 13:34 |
| gthiemonge | kevko: octavia_persistence (jobboard) is optional, v2 can work without it | 13:34 |
| kevko | hmm | 13:34 |
| kevko | okay, thanks | 13:34 |
| gthiemonge | kevko: this is the default config (like in the CI), amphorav2 without jobboard | 13:35 |
| kevko | gthiemonge: thanks .. just asking | 13:35 |
| kevko | wncslln: ensure that you can reach amphora instance from your octavia-worker ... ( octavia worker should have IP from management network ....) | 13:36 |
| wncslln | i have lost the error logs but iirc octavia worker cant connect to amphora instance | 13:38 |
| wncslln | gthiemonge: Could not connect to instance. Retrying.: requests.exceptions.ConnectTimeout: HTTPSConnectionPool(host='10.5.2.53', port=9443): Max retries exceeded with url: // (Caused by ConnectTimeoutError(<urllib3.connection.HTTPSConnection object at 0x7f889c9bc7c0>, 'Connection to 10.5.2.53 timed out. (connect timeout=10.0)')) | 13:40 |
| gthiemonge | wncslln: if they are WARNING-level messages, it's fine, we expect to see that, if it's an ERROR-level message, it's a real issue | 13:42 |
| wncslln | its a warning | 13:42 |
| gthiemonge | when your LB goes to provisioning_status ERROR, you should get ERROR-level messages with the reason in the logs | 13:43 |
| wncslln | i sent a request to create another LB, im waiting to change status to ERROR to see any log message | 13:44 |
| gthiemonge | ack | 13:45 |
| kevko | wncslln: point is that you *have to* reach amphora instance from worker ... did you use octavia_auto_configure in kolla_ansible ? or you've created network resources by a hand ? | 13:45 |
| wncslln | kevko: i use auto_configure | 13:47 |
| wncslln | i will test if i can reach instance | 13:47 |
| kevko | wncslln: systemctl list-units | grep o-hm | 13:47 |
| wncslln | not found | 13:47 |
| kevko | ip a | grep o-hm0 | 13:48 |
| kevko | wncslln: ^^ you should have this interface working | 13:48 |
| wncslln | oh | 13:49 |
| wncslln | not found yoo | 13:49 |
| wncslln | too | 13:49 |
| wncslln | :/ | 13:49 |
| kevko | wncslln: that's your problem | 13:49 |
| wncslln | this interface that makes comm with loadbalancer and instances? | 13:49 |
| kevko | wncslln: in production environment your should have management network on which the amphoras are started - some external network ..and physical interfaces on nodes ... | 13:50 |
| kevko | wncslln: kolla has octavia_auto_configure ...so kolla create tenant network in service project ...and create ip interface on physical node which is actually created port in ovs | 13:51 |
| kevko | wncslln: https://github.com/openstack/kolla-ansible/blob/master/ansible/roles/octavia/tasks/hm-interface.yml | 13:52 |
| wncslln | i have the management network on service project | 13:54 |
| wncslln | but doesnt have the interface on infra nodes | 13:55 |
| kevko | wncslln: so check conditionals | 13:55 |
| opendevreview | Tom Weininger proposed openstack/octavia-tempest-plugin master: WIP test rebase of below patch https://review.opendev.org/c/openstack/octavia-tempest-plugin/+/905683 | 15:28 |
| servagem | Hello, I was wondering if there is a planned release date for a new version of Octavia Yoga? I've noticed some fixes have been made in the stable/yoga branch since the last release (version 10.1.0). | 18:28 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!