-
Bug
-
Resolution: Done-Errata
-
Critical
-
4.14
-
None
-
No
-
False
-
Description of problem:
The timeout of calls to the csi driver from both the external csi-provisioner and csi-attacher are 15 seconds by default. However hotplugging a volume into the Virtual Machine can take up to a minute (sometimes more). This causes the context timeout to expire, and in some cases causes the bookkeeping of what volumes are attached to become corrupted, and detaching the volumes doesn't always get handled properly afterwards.
Version-Release number of selected component (if applicable):
How reproducible:
Run the standard csi conformance tests against the csi driver. Most of the runs this issue will appear as a random failed test or two. The failed test are because the deletion of the persistent volume never happens. Because of this we cannot get a good signal on the state of the csi driver.
Steps to Reproduce:
1. 2. 3.
Actual results:
Random failed tests of the csi conformance suite.
Expected results:
csi conformance suite passes
Additional info:
Fixed in upstream by increasing the timeouts to 3 minutes instead of 15 seconds.
- clones
-
OCPBUGS-19720 kubevirt csi driver external provisioner/attacher timeout too low
- Closed
- depends on
-
OCPBUGS-19720 kubevirt csi driver external provisioner/attacher timeout too low
- Closed
- links to
-
RHSA-2023:5006 OpenShift Container Platform 4.14.z security update