dead model not removed, triggering issues with duplicate names
Bug #1740815 reported by
Paul Gear
This bug affects 1 person
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
Canonical Juju |
Expired
|
High
|
Unassigned | ||
2.3 |
Won't Fix
|
High
|
Unassigned |
Bug Description
'juju status -m modelname --format=json' occasionally reports the message:
ERROR model "username/
This occurs when running 'juju models', finding all models with valid life and state, then running 'juju status' on each of them per above. Running the command again shortly afterwards works fine.
This is occurring on 2.2.6 client with 2.2.8 agents, but was also seen before the upgrade to 2.2.8.
Changed in juju: | |
status: | New → Triaged |
importance: | Undecided → High |
milestone: | none → 2.4-beta1 |
description: | updated |
tags: | added: canonical-is |
Changed in juju: | |
status: | New → Triaged |
Changed in juju: | |
assignee: | Andrew Wilkins (axwalk) → nobody |
Changed in juju: | |
milestone: | 2.4-beta1 → none |
tags: | added: teardown |
To post a comment you must log in.
As far as I can tell, the only way we would get that error message is if:
(1) the model has been removed, or
(2) there was an error on the backend accessing the database for that model
Based on the bug description, we can rule out (1), because you state that a later call to "juju status" for the same model succeeds.
So (2) seems likely, but we would need much more detailed information to diagnose the error. That, and we're not being specific enough in our error checking, and considering any error to fetch model data from mongo as a "model not found" error: https:/ /github. com/juju/ juju/pull/ 8252.
For more detailed info, you can use TRACE level logging on the client side for the "juju status" commands:
juju status --logging- config= '<root> =TRACE' --show-log -m admin/foo