| *** yadnesh|away is now known as yadnesh | 05:01 | |
| *** yadnesh is now known as yadnesh|afk | 08:49 | |
| *** yadnesh|afk is now known as yadnesh | 10:00 | |
| skraynev | Hi folks. I need some advice about two unexpected behaviours for Octavia in yoga. 1. when I create LB - it stuck in OFFLINE operating state. provisioning is ACTIVE. On some previous release it goes to ONLINE directly after creation. (without creation other things like listener and etc.) is it expected? if yes - what should I do to make it ONLINE? 2. when I tries to create LB in one Fully Populated request - all resources go to OFFLINE and | 13:41 |
|---|---|---|
| skraynev | stuck in PENDING CREATE. The loadbalancer itself goes to ACTIVE/OFFLINE. and as result I could not remove it without hacking DB. what I could do here incorrect? and how could I debug it? | 13:41 |
| gthiemonge | skraynev: hi | 13:42 |
| gthiemonge | for 1. I need to check the code | 13:43 |
| skraynev | gthiemonge: hey. btw - thank you for the help with previous issue :) | 13:43 |
| gthiemonge | skraynev: for your 2nd pbl, there's a open patch that fixes a similar behavior: https://review.opendev.org/c/openstack/octavia/+/864204 | 13:44 |
| gthiemonge | skraynev: hey, np ;-) | 13:44 |
| gthiemonge | skraynev: could you share the payload you send to the API for your 2nd issue? | 13:46 |
| gthiemonge | skraynev: in my env, a LB is ONLINE after creation (no listener, etc...) | 13:51 |
| gthiemonge | skraynev: resources in OFFLINE status might been the result of an issue with the heartbeat packets that are sent by the amphorae to the health-manager service | 13:52 |
| skraynev | gthiemonge. please see payload here: https://pastebin.com/3aaXpiyT | 13:52 |
| gthiemonge | skraynev: which resources are in PENDING_CREATE? all? do you see errors in the logs? | 13:54 |
| skraynev | gthiemonge: it's a good information, that on your env it works. It means, that I really have issue in my env and now I know what is expected result! could it be an issue, if I use amphora image from previous release? should I create new one according my current octavia release an use it? is any approach how to debug missed heartbeat packets? do I need read logs in amphora or in health-manager? sorry for the naive questions :) I am not enough | 13:57 |
| skraynev | familiar with octavia (hope, yet) | 13:57 |
| skraynev | gthiemonge: regarding PENDING CREATE. I see, that all resources except LB stay in such state. I did not find any tracebacks in logs... but maybe it's because I have log filtering and I already thought about removing filter and reading ALL logs | 13:59 |
| tweining | one way of debugging heartbeats could be to run "tcpdump -nn -i o-hm0 udp" on the controller I think | 14:04 |
| gthiemonge | FYI I'm on master, but I don't remember that we had an issue with the op status since yoga | 14:04 |
| gthiemonge | an old amphora would be fine, but you will skip some bugfixes from yoga | 14:04 |
| gthiemonge | when debug is enabled, the hm logs print a line for each received heartbeat message | 14:07 |
| skraynev | gthiemonge: I have a yoga release :) | 14:09 |
| skraynev | or you mean, that it should be presented on yoga? I probably misunderstood your answer | 14:10 |
| skraynev | tweining: do you mean directly on the node, where I have octavia daemons? right? | 14:11 |
| gthiemonge | skraynev: sorry, I mean that yoga should be fine, no known issue regarding the op status | 14:13 |
| tweining | Yes, where the hm service runs | 14:14 |
| skraynev | gthiemonge, tweining: thank you for the help! now I know the direction for debugging! ;) | 14:18 |
| *** yadnesh is now known as yadnesh|away | 16:26 | |
| opendevreview | Merged openstack/octavia-tempest-plugin master: Add scenario test for Prometheus over TLS https://review.opendev.org/c/openstack/octavia-tempest-plugin/+/830015 | 19:05 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!