-
Bug
-
Resolution: Cannot Reproduce
-
Normal
-
4.13.z
-
?
-
Low
-
No
-
ShiftStack Sprint 235, ShiftStack Sprint 236, ShiftStack Sprint 237, ShiftStack Sprint 238
-
4
-
False
-
-
-
Bug Fix
Description of problem:
Reported upstream in https://github.com/kubernetes/cloud-provider-openstack/issues/2217 Not specifically reproduced in OpenShift, but I have no reason to think we would not be affected, and I know we have users with strict proxy requirements. The user's configuration requires all OpenStack API requests from the tenant network to go through a proxy. They have configured a proxy 'globally' in their cluster in a manner which also affects the CSI driver. Attempting to attach a volume to a pod fails. Inspecting the logs we see that cinder attempted to attach the volume to the proxy server, not the node hosting the pod. The reason for this is that the metadata request was also proxied, meaning the returned values relate to the proxy server, not the local server.
Version-Release number of selected component (if applicable):
4.13, but likely all versions since we enabled CSI
How reproducible:
Steps to Reproduce:
1. 2. 3.
Actual results:
Expected results:
Additional info: