Details
-
Bug
-
Resolution: Unresolved
-
Critical
-
None
-
4.14.z, 4.15.z, 4.16
-
Critical
-
Yes
-
Proposed
-
ppc64le
-
False
-
Description
Description of problem:
1. OCP Cluster deployment on Power - CoreOS nodes fail to boot with multipath enabled at install time . Kernel parameters for enabling multipath : rhcos_pre_kernel_options = ["rd.multipath=default","root=/dev/disk/by-label/dm-mpath-root"]
Version-Release number of selected component (if applicable):
How reproducible:
We could reproduce this issue with 4.14.20 and 4.16.0-ec.5 builds on IBM Power platform with 'Storage Connectivity Group' set to "vSCSI". Issue is not reproducing with "NPIV" storage.
Steps to Reproduce:
1.Follwed the RH doc for deploying an OCP cluster with multipath enabled as day1(at install time) Kernel parameters for enabling multipath : rhcos_pre_kernel_options = ["rd.multipath=default","root=/dev/disk/by-label/dm-mpath-root"] 2. deployment failed with node boot issue.
Actual results:
Deployment failed - coreos node is failing to boot. Error logs are attached.
Expected results:
Cluster should deploy successfully with multipath enabled as day1.
Additional info:
This looks like regression issue. Multipath enablement was functioning properly with earlier builds 4.14 and 4.15 versions. We could reproduce this issue with 4.14.20 and 4.16.0-ec.5 builds on IBM Power platform with 'Storage Connectivity Group' set to "vSCSI". Issue is not reproducing with "NPIV" storage.
Below is the link for logs:
