| hyperbaba | Hi there, | 07:56 |
|---|---|---|
| hyperbaba | I have a deployment of openstack using kolla-ansible. I lost one of the controller nodes due to hardware failure and created a new server. Now i'm having issues with etcd container which is restarting compleaining that member xxxxxx is already bootstrapped. Can anyone assist me with the issue? | 08:00 |
| yusufgungor | Hi @hyperbaba you may try to remove failed etcd member and join again using etcdctl. There exist k8s etcd docs, i think they can adaptable for openstack too https://itnext.io/breaking-down-and-fixing-etcd-cluster-d81e35b9260d | 08:43 |
| SvenKieske | I think this is mainly a etcd question for the etcd community, not really kolla-ansible. you can of course also discuss/ask here, but you might get more responses in the etcd community I guess? | 08:59 |
| frickler | although it wouldn't hurt if kolla could detect and maybe even resolve this situation. contributions welcomed ;) | 09:16 |
| spatel | mnasiadka Hey! are you there buddy? I have a kolla override question - https://paste.opendev.org/show/bztnBpmwRPxTh7h1PNev/ | 14:55 |
| spatel | I have this tree and i want to push driver = messagingv2 to all nova service but it doesn't work even i have nova.conf file there | 14:56 |
| spatel | I must be doing something wrong | 14:56 |
| rockey | spatel: /etc/kolla/config/nova.conf is your friend, not .../config/nova/nova.conf | 14:58 |
| spatel | When I move nova.conf to ../config/nova.conf then my ceph implementation break and compute start throwing error saying not able to find ceph key :( | 14:59 |
| spatel | I am also shocked with that.. | 15:00 |
| rockey | i have .../config/{nova,neutron}.conf and .../config/nova/ceph.conf + keyrings | 15:01 |
| spatel | what about nova-api.conf and nova-scheduler.conf files? | 15:02 |
| spatel | they should be inside ../config ? | 15:02 |
| rockey | spatel: https://pastebin.com/SeHeiqMi | 15:06 |
| rockey | that's my layout | 15:06 |
| spatel | you don't have nova-api.conf etc.. assuming that config can go to nova.conf | 15:07 |
| rockey | should be in the same place as nova.conf, but untested theory | 15:08 |
| rockey | ie, .../config/nova-api.conf | 15:08 |
| spatel | Okie! let me run some test | 15:09 |
| spatel | I have notice one more thing that i have /etc/kolla/config/nova/ceph.client.nova.keyring file but it not push out to any node | 15:09 |
| spatel | all i can see on compute node /etc/ceph/ceph.client.cinder.keyring file | 15:10 |
| rockey | verify in globals.yml, but keyring for nova perhaps defaults to same as cinder | 15:12 |
| rockey | https://opendev.org/openstack/kolla-ansible/src/branch/master/etc/kolla/globals.yml#L455 | 15:13 |
| rockey | oh sorry, L465 | 15:13 |
| spatel | Yes! I can see that nova = cinder | 15:14 |
| spatel | when I move nova.conf to ../config/nova.conf and run kolla-ansible reconfigure I started seeing this error on all compute nodes - https://paste.opendev.org/show/bNbXtfpPGZknmdRqPGrg/ | 15:14 |
| spatel | even right now I can't see any /etc/ceph/ceph.client.nova.keyring file on compute and it works so trying to hunt it down what is going on | 15:15 |
| spatel | This is yoga but i have same config in Zed and everything working fine there.. so I am little confuse. anyway let me go through in depth and see | 15:16 |
| spatel | ah! I know what is going on, its nova.conf rbd_user = cinder vs rbd_user = nova issue.. | 15:28 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!