hook failed: "install" when deployed to a non-k8s model
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
CoreDNS Charm |
Fix Released
|
Medium
|
George Kraft |
Bug Description
When deploying the coredns charm to a model backed by a non-k8s cloud (e.g. Azure, vSphere, etc), the charm fails during the install hook due to missing service account credentials.
The real problem here is that the charm doesn't support being deployed in such models, and the charm's metadata is missing the required bits to inform Juju that that is the case.
====== Traceback ======
Traceback (most recent call last):
File "/var/lib/
token = account_
File "/usr/lib/
with self.open(mode='r', encoding=encoding, errors=errors) as f:
File "/usr/lib/
return self._accessor.
FileNotFoundError: [Errno 2] No such file or directory: '/var/run/
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "/var/lib/
return KubeConfig.
File "/var/lib/
raise exceptions.
lightkube.
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "/var/lib/
main(
File "/var/lib/
charm = charm_class(
File "/var/lib/
self.client = Client(
File "/var/lib/
self._client = GenericSyncClie
File "/var/lib/
config = KubeConfig.
File "/var/lib/
return KubeConfig.
File "/var/lib/
raise exceptions.
lightkube.
tags: | added: backport-needed |
Changed in charm-coredns: | |
status: | Fix Committed → Fix Released |
PR: https:/ /github. com/charmed- kubernetes/ charm-coredns/ pull/37