autopkgtest lxd provider tests fail for 2.0
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
Canonical Juju |
Fix Released
|
Critical
|
Unassigned | ||
juju-core (Ubuntu) |
Fix Released
|
Critical
|
Unassigned |
Bug Description
The new autopkgtests for the lxd provider are still not passing yet. The bridge setup looks like it works, but when juju comes to try and connect to lxd it is rejected on an http url.
<http://
+ juju bootstrap my-controller lxd --upload-tools --debug --config default-
2016-04-15 07:59:35 INFO juju.cmd supercommand.go:60 running juju [2.0-beta4 gc go1.6.1]
2016-04-15 07:59:35 INFO cmd cmd.go:141 cloud "lxd" not found, trying as a provider name
2016-04-15 07:59:35 INFO cmd cmd.go:141 no credentials found, checking environment
2016-04-15 07:59:35 DEBUG juju.cmd.
2016-04-15 07:59:39 DEBUG juju.tools.
2016-04-15 07:59:46 DEBUG juju.tools.
2016-04-15 07:59:47 ERROR cmd supercommand.go:448 Get https:/
adt-run [07:59:48]: test current-
adt-run [07:59:48]: test current-
current-
This is particularly confusing, as the exact same test passes if run on a canonistack machine using the ssh runner:
<http://
Test scripts are in the debian/tests directory of this packaging branch:
<https:/
From discussion on irc earlier, the only way Tycho thought we could hit the Forbidden here:
[16:18] <tych0> mgz: so juju configures LXD to listen over HTTPS
[16:18] <tych0> and if it accesses LXD via https, it needs to also configure LXD to recognize juju's cert
[16:18] <tych0> it strikes me that that cert configuration has failed
[16:19] <tych0> and that's why you get a 403
[16:19] <tych0> i'm not sure why you'd get it for any other reason
Changed in juju-core: | |
importance: | Undecided → Critical |
milestone: | none → 2.0-rc1 |
status: | New → Triaged |
tags: | added: lxd-provider packaging |
tags: | added: jujuqa |
Changed in juju-core: | |
status: | Triaged → Fix Released |
affects: | juju-core → juju |
Changed in juju: | |
milestone: | 2.0-beta5 → none |
milestone: | none → 2.0-beta5 |
With less than a week to go to 16.04 release date, the status is:
- juju-core-1 is in xenial
- juju-core 1.25 is also in xenial
- juju-core 2 is blocked in xenial-proposed because of test failures.
This situation is clearly not releasable; we need to get juju-core 2 into xenial for release. This is also blocking several other related packages from migrating to xenial for release (closure, bigdata, openstack).
I have attempted to get a clean baseline locally with the autopkgtests, using the adt-virt-ssh backend (to a xenial amd64 VM) to avoid any problems with nested lxd or otherwise. The current- lxd-provider test failed for me as follows:
2016-04-16 21:51:33 DEBUG juju.provider.lxd environ_ broker. go:126 LXD requires https://, using: https:/ /cloud- images. ubuntu. com/releases/ lxdclient client.go:73 connecting to LXD remote "default cloud images": "https:/ /streams. canonical. com/juju/ images/ releases/" lxdclient client_image.go:129 no image for ubuntu-xenial found in https:/ /streams. canonical. com/juju/ images/ releases/ lxdclient client.go:73 connecting to LXD remote "default ubuntu cloud images": "https:/ /cloud- images. ubuntu. com/releases/" lxdclient client_image.go:135 found image from https:/ /cloud- images. ubuntu. com/releases/ for xenial = 6cb0ba80a5fe323 57568a473cbaf69 f14d26da0ba6b08 f5b1bcde7053fc7 3757 lxdclient client_image.go:147 copying image for ubuntu-xenial from https:/ /cloud- images. ubuntu. com/releases/: 1% /cloud- images. ubuntu. com/releases/: 2016-04-16 21:51:37 INFO juju.tools. lxdclient client_image.go:147 copying image for ubuntu-xenial from https:/ /cloud- images. ubuntu. com/releases/: 100% /cloud- images. ubuntu. com/releases/: 2016-04-16 21:51:38 DEBUG juju.tools. lxdclient client_image.go:139 dropped 0 progress messages broker. go:223 LXD user data; 1342 bytes broker. go:201 starting instance "juju-b43656cd- a9d3-4c92- 80ac-599997219a d7-machine- 0" (image "ubuntu-xenial")... common destroy.go:22 destroying model "admin" common destroy.go:33 destroying instances common destroy.go:53 destroying storage lxd-provider: ------- ------- ------- --] lxd-provider: - - - - - - - - - - results - - - - - - - - - - lxd-provider FAIL non-zero exit status 1
2016-04-16 21:51:33 DEBUG juju.tools.
2016-04-16 21:51:33 INFO juju.tools.
2016-04-16 21:51:33 DEBUG juju.tools.
2016-04-16 21:51:34 INFO juju.tools.
2016-04-16 21:51:34 INFO juju.tools.
[...]
copying image for ubuntu-xenial from https:/
copying image for ubuntu-xenial from https:/
2016-04-16 21:51:38 DEBUG juju.service discovery.go:59 discovered init system "systemd" from series "xenial"
2016-04-16 21:51:38 DEBUG juju.provider.lxd environ_
2016-04-16 21:51:38 INFO juju.provider.lxd environ_
2016-04-16 21:52:04 INFO juju.provider.
2016-04-16 21:52:04 INFO juju.provider.
2016-04-16 21:52:04 INFO juju.provider.
2016-04-16 21:52:04 ERROR cmd supercommand.go:448 failed to bootstrap model: cannot start bootstrap instance: exit status 2
adt-run [14:52:05]: test current-
adt-run [14:52:06]: test current-
current-
If I could get this test to pass locally, I would go ahead with letting juju-core 2.0 in as-is and trust that the test incompatibilities with the autopkgtest infrastructure would be sor...