Open Stack

I am running into issues with deploying Openstack using Kayobe. Is this
list group the best place to raise? If not, my apologies - please could you
share where I need to go?

1. kayobe hangs during deployment, does not time out, does not error out
when has previously been successful and without configuration changes.

2. deployment fails due to breaking the network relating to the bridge.
Also changes login password which locks out of console.

Details:
1.
Environment deployment: Train
all-in-one host: centos7
Ansible control host: Ubuntu 18

With the first issue, I've seen this multiple times but have not been able
to find the root cause. I searched online and came across other ansible
users that state their playbooks were hanging and were able to resolve by
clearing cache. I've tried clearing out ~/.ansible/* and /tmp/* on the
control host. Also tried doing the same on the all-in-one host without
success. This issue came about after doing a full destroy of the
environment and then redeploying, making a minor config change and then
redeploying.

2.
Environment deployment: Victoria
all-in-one host: centos8
Ansible control host: Ubuntu 20

Because I couldnt resolve the issue above and centos 8.4 is available, I
decided to try and go to centos 8 and deploy Victoria. I hit 2 issues:

1. I am unable to login with "kayobe_ansible_user" using after "kayobe
overcloud host configure" with "wrong password" message. Resetting the
password resolves but the password seems changed again with a host
configure.

2. deployment fails when "TASK [openvswitch : Ensuring OVS bridge is
properly setup]"
Looking at the ovs container, it's unable to add the physical interface to
the bridge bond0 interface, complaining that the device is busy or is
already up. I saw some log messages relating to ipv6 so I tried disabling
ipv6 and redeploying but the same issue.
I then rebuilt the host again and opted not to use a bond0 interface
however the same loss of network occurs. If I log into the openvswitch_db
container then there are CLI commands where I can delete the bridge to
restore the network.

So at this point, after turning off ipv6 again and running without bond0
bond, I tried another deploy wile tailing
`/var/log/kolla/openvswitch/ovs-vswitchd.log` and now I do not see any
errors but the network is lost to the host and the script fails to finish
deployment. I've attached the logs that appear at the point the network
dies: https://pasteboard.co/K65qwzR.png


Are these known issues and does anyone have any information as to how I can
work through or around them?

Regards,

Tony Pearce
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack-discuss/attachments/20210611/41176021/attachment-0001.html>