Getting stuck bootstrapping LXD on Ubuntu 20.04 LTS

juju
#1

Just now tried experimenting with LXD on my new installation of Ubuntu 20.04… but seeing it hang at the bootstrap step

(base) emcp@2600k:~$ juju bootstrap
Clouds
aws
aws-china
aws-gov
azure
azure-china
cloudsigma
google
joyent
localhost
oracle
oracle-classic
rackspace

Select a cloud [localhost]: localhost

Enter a name for the Controller [localhost-localhost]: 

Creating Juju controller "localhost-localhost" on localhost/localhost
Looking for packaged Juju agent version 2.7.6 for amd64
To configure your system to better support LXD containers, please see: https://github.com/lxc/lxd/blob/master/doc/production-setup.md
Launching controller instance(s) on localhost/localhost...
  -- creating container

then after a while it breaks my ssh session with

client_loop: send disconnect: Broken pipe

EDIT: Seems it totally crashed mid-bootstrap and now I need to wipe whatever it started and start over.

any tips to debug it? thinking of reinstalling 20.04 anyway since I am having VNC troubles as well (Tiger VNC wiped tight vnc)

When I try to get back in, my PC is not reachable… forcing me to reboot and login with keyboard and mouse… once inside I get this

(base) emcp@2600k:~$ juju clouds --debug
20:50:19 INFO  juju.cmd supercommand.go:83 running juju [2.7.6  gc go1.10.4]
20:50:19 DEBUG juju.cmd supercommand.go:84   args: []string{"/snap/juju/11454/bin/juju", "clouds", "--debug"}
20:50:19 INFO  juju.util.exec exec.go:209 run result: exit status 1
20:50:19 DEBUG juju.kubernetes.provider detectcloud.go:24 failed to query local microk8s: microk8s is not installed: microk8s not found
20:50:19 WARN  cmd list.go:140 no API addresses
ERROR cmd: error out silently
20:50:19 DEBUG cmd supercommand.go:519 error stack: 
cmd: error out silently
/build/juju/parts/juju/go/src/github.com/juju/juju/cmd/juju/cloud/list.go:182:
#2

I tried deleting and re-installing juju… now it complains on another step

(base) emcp@2600k:~$ snap install juju --classic
juju 2.7.6 from Canonical✓ installed
(base) emcp@2600k:~$ juju bootstrap
Clouds
aws
aws-china
aws-gov
azure
azure-china
cloudsigma
google
joyent
localhost
oracle
oracle-classic
rackspace

Select a cloud [localhost]: 

Enter a name for the Controller [localhost-localhost]: lxd-host

Creating Juju controller "lxd-host" on localhost/localhost
Looking for packaged Juju agent version 2.7.6 for amd64
To configure your system to better support LXD containers, please see: https://github.com/lxc/lxd/blob/master/doc/production-setup.md
Launching controller instance(s) on localhost/localhost...
 - Create instance from image: Failed to run: zfs clone default/images/93b9eeb85479af2029203b4a56a2f1fdca6a0e1bf23cdc26b567790bf0f3f3bd@readonly default/containerERROR failed to bootstrap model: cannot start bootstrap instance in availability zone "2600k": Create instance from image: Failed to run: zfs clone default/images/93b9eeb85479af2029203b4a56a2f1fdca6a0e1bf23cdc26b567790bf0f3f3bd@readonly default/containers/juju-3134f4-0: cannot open 'default/images/93b9eeb85479af2029203b4a56a2f1fdca6a0e1bf23cdc26b567790bf0f3f3bd@readonly': dataset does not exist

i even wiped my entire .local/share/juju in an effort to get it unstuck but same error

#3

Can you launch lxd containers on this machine? And if so, can you ping created containers, and exec into them from the host?

I’d try doing something like:

lxc launch ubuntu:20.04 
lxc list
ping <the ip of the container you just made>
lxc shell <the name of the container you just launched>

Since you’ve having other networking related issues on the host, and since the second attempt at launching a container failed with an interesting zfs error, I think that the first step is to verify that you can launch and connect to containers. Then you can move on to troubleshooting Juju, if necessary :slight_smile:

#4

seems to freeze on this step

image

I’ll take this up with the LXD forum then

https://discuss.linuxcontainers.org/t/lxc-launch-freezes-on-ubuntu-20-04/7880

thanks for the tips

1 Like