can not deploy flannel - regression of snapd

[E. MAS]

The last release of snapd bring another a new bug .

The flannel charm use /usr/local/bin/ to deploy some binary

Because /usr/local/bin/ is not anymore in system-d PATH , all daemons does not have it .

so juju can not access it .

2019-02-02 04:21:27 DEBUG install File "/var/lib/juju/agents/unit-flannel-3/charm/reactive/flannel.py", line 224, in set_flannel_version
2019-02-02 04:21:27 DEBUG install version = check_output(split(cmd), stderr=STDOUT).decode('utf-8')
2019-02-02 04:21:27 DEBUG install File "/usr/lib/python3.6/subprocess.py", line 336, in check_output
2019-02-02 04:21:27 DEBUG install **kwargs).stdout
2019-02-02 04:21:27 DEBUG install File "/usr/lib/python3.6/subprocess.py", line 403, in run
2019-02-02 04:21:27 DEBUG install with Popen(*popenargs, **kwargs) as process:
2019-02-02 04:21:27 DEBUG install File "/usr/lib/python3.6/subprocess.py", line 709, in __init__
2019-02-02 04:21:27 DEBUG install restore_signals, start_new_session)
2019-02-02 04:21:27 DEBUG install File "/usr/lib/python3.6/subprocess.py", line 1344, in _execute_child
2019-02-02 04:21:27 DEBUG install raise child_exception_type(errno_num, err_msg, err_filename)
2019-02-02 04:21:27 DEBUG install FileNotFoundError: [Errno 2] No such file or directory: 'flanneld': 'flanneld'

I reported the bug in snapd :

https://bugs.launchpad.net/ubuntu/+source/snapd/+bug/1814355

to fix:

wget http://archive.ubuntu.com/ubuntu/pool/main/s/snapd/snapd_2.34.2_amd64.deb
dpkg -i snapd_2.34.2_amd64.deb
init 6

wait a few minutes and juju can continue the deployment

[Alexander Turek (turekaj)]

confirmed -- this has broken juju canonical Kubernetes charm. Cannot deploy new Kubernetes workers with flannel.

[George Kraft (cynerva)]

I can't reproduce this. Per https://bugs.launchpad.net/ubuntu/+source/snapd/+bug/1814355, it looks like it was fixed over the weekend with the release of snapd 2.37.1.1 to bionic-updates. Can you confirm you are no longer experiencing the issue?

[E. MAS (erwan-mas)]

I dont have anymore the issue .

In my context , i use maas , and in this context , at this time , i need to reboot the server .

The server is deployed with the bogus package ( 2.37.1+18.04 ) , and after when juju install his agent , juju do the upgrade to 2.37.1.1 .
This already too late because the systemd path is already corrupted .

On the next reboot , snapd will not corrupt the systemd path .

I dont know when my maas server will install directly with the right version .

You can close , the problem is solved .