juju uninstalling itself upon machine creation ?

Bug #1816098 reported by Junien Fridrick on 2019-02-15
8
This bug affects 1 person
Affects Status Importance Assigned to Milestone
juju
Medium
Unassigned

Bug Description

Hi,

I just deployed 4 new juju units (juju 2.4.7 agent on 2.5.0 controller), and 2 of them got stuck in "allocating". When I connected, this is what I found in the machine log :

https://pastebin.canonical.com/p/HNYXNTKPF5/

Most important lines :
2019-02-15 15:29:51 INFO juju.api apiclient.go:603 connection established to "wss://<IP>:17070/model/235f7c2f-b531-46e1-8438-1a663a614a68/api"
2019-02-15 15:29:51 DEBUG juju.worker.apicaller connect.go:155 [235f7c] failed to connect
2019-02-15 15:29:51 DEBUG juju.worker.dependency engine.go:582 "api-caller" manifold worker stopped: [235f7c] "machine-18" cannot open api: connection permanently impossible
2019-02-15 15:29:51 INFO juju.agent uninstall.go:36 marking agent ready for uninstall

Why does juju decide to uninstall itself ? If there's a good reason for that, it should at least notify the controller (or the controller should be able to notice somehow). Also, no connection is ever "permanently impossible", so juju should be able to retry.

Thanks

'permanently impossible' is a case where we were able to successfully
connect to a controller, but when we try to login with the credentials we
have, we are told that those credentials are not valid.
I believe the teardown existed for the manual provider case, if you had a
machine that was down, and the controller was destroyed, it wouldn't clean
up until the machine started again. We can certainly reevaluate the
position.

On Fri, Feb 15, 2019 at 5:01 PM Junien Fridrick <email address hidden>
wrote:

> Public bug reported:
>
> Hi,
>
> I just deployed 4 new juju units (juju 2.4.7 agent on 2.5.0 controller),
> and 2 of them got stuck in "allocating". When I connected, this is what
> I found in the machine log :
>
> https://pastebin.canonical.com/p/HNYXNTKPF5/
>
> Most important lines :
> 2019-02-15 15:29:51 INFO juju.api apiclient.go:603 connection established
> to "wss://<IP>:17070/model/235f7c2f-b531-46e1-8438-1a663a614a68/api"
> 2019-02-15 15:29:51 DEBUG juju.worker.apicaller connect.go:155 [235f7c]
> failed to connect
> 2019-02-15 15:29:51 DEBUG juju.worker.dependency engine.go:582
> "api-caller" manifold worker stopped: [235f7c] "machine-18" cannot open
> api: connection permanently impossible
> 2019-02-15 15:29:51 INFO juju.agent uninstall.go:36 marking agent ready
> for uninstall
>
> Why does juju decide to uninstall itself ? If there's a good reason for
> that, it should at least notify the controller (or the controller should
> be able to notice somehow). Also, no connection is ever "permanently
> impossible", so juju should be able to retry.
>
> Thanks
>
> ** Affects: juju
> Importance: Undecided
> Status: New
>
> --
> You received this bug notification because you are subscribed to juju.
> Matching subscriptions: juju bugs
> https://bugs.launchpad.net/bugs/1816098
>
> Title:
> juju uninstalling itself upon machine creation ?
>
> To manage notifications about this bug go to:
> https://bugs.launchpad.net/juju/+bug/1816098/+subscriptions
>

Changed in juju:
status: New → Triaged
importance: Undecided → Medium
To post a comment you must log in.
This report contains Public information  Edit
Everyone can see this information.

Other bug subscribers