cannot provision multiple juju environments
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
MAAS |
New
|
Undecided
|
Unassigned |
Bug Description
According to the documentation,
http://
"One hard and fast rule is that you’ll need to use a different API key for each Juju environment you set up within a single MAAS cluster."
Regardless of whether I use unique oauth keys in the same user or across multiple users, with or without admin
privileges; the outcome is the same
1. juju env A is bootstrapped using key X
2. juju env B is bootstrapped using key Y
3. env A bootstrap node is claimed to not be running even though it plainly is
4. env A becomes unreachable
5. env B keeps on going
All of the machines are on the same network. I get different outcomes depending on the phases
of the moon. At the moment, goju can complete the task but pyju can't, yesterday neither of
them were working.
example:
ubuntu@pluton:~$ juju --version
juju 0.7
ubuntu@pluton:~$ juju bootstrap -e test-alpha
2013-10-09 15:49:11,688 INFO Bootstrapping environment 'test-alpha' (origin: distro type: maas)...
2013-10-09 15:49:12,185 INFO 'bootstrap' command finished successfully
ubuntu@pluton:~$ juju bootstrap -e test-gamma
2013-10-09 15:49:18,263 INFO Bootstrapping environment 'test-gamma' (origin: distro type: maas)...
2013-10-09 15:49:18,640 INFO 'bootstrap' command finished successfully
ubuntu@pluton:~$ juju status -e test-alpha
2013-10-09 15:49:24,514 INFO Connecting to environment...
2013-10-09 15:49:24,581 ERROR juju environment not found: machines are not running (/MAAS/
ubuntu@pluton:~$ maas-cli maas node read node-9570558e-
{
"status": 6,
"macaddress
{
}
],
"netboot": true,
"hostname": "watson.farm",
"power_type": "virsh",
"system_id": "node-9570558e-
"architecture": "amd64/generic",
"tag_names": [],
"resource_uri": "/MAAS/
}
However according to the web interface and the virsh manager, two nodes were allocated.
watson.farm 52:54:00:05:2f:ca Allocated to env2
fringe.farm 52:54:00:05:2f:c0 Allocated to ubuntu
and... maas/juju knows enough to destroy those environments.
ubuntu@pluton:~$ juju destroy-environment -e test-alpha
WARNING: this command will destroy the 'test-alpha' environment (type: maas).
This includes all machines, services, data, and other resources. Continue [y/N] y
2013-10-09 15:54:49,708 INFO Destroying environment 'test-alpha' (type: maas)...
2013-10-09 15:54:52,296 INFO 'destroy_
ubuntu@pluton:~$
ubuntu@pluton:~$ juju destroy-environment -e test-gamma
WARNING: this command will destroy the 'test-gamma' environment (type: maas).
This includes all machines, services, data, and other resources. Continue [y/N] y
2013-10-09 15:54:57,534 INFO Destroying environment 'test-gamma' (type: maas)...
2013-10-09 15:54:57,860 INFO 'destroy_
watson.farm 52:54:00:05:2f:ca Ready
fringe.farm 52:54:00:05:2f:c0 Ready
environment snippet
test-alpha:
type: maas
maas-server: http://
maas-oauth: Pe6nJHpTnDzdEaR
admin-secret: 70c6fd03-
default-series: quantal
authorized-
ssh-rsa AAAAB3NzaC1yc2E
test-gamma:
type: maas
maas-server: http://
maas-oauth: ZF6FFw45hWNAFDe
admin-secret: 2a89dc44-
default-series: quantal
authorized-
ssh-rsa AAAAB3NzaC1yc2E
The doc should say "use a different user account and its API key for each juju environment". This is possibly a bug in juju but I'm duping this with a similar bug anyway.