Bug 2009019

Summary: [4.8.z] legacy RHCOS boot image metadata is not in sync with stream metadata
Product: OpenShift Container Platform Reporter: Micah Abbott <miabbott>
Component: InstallerAssignee: Micah Abbott <miabbott>
Installer sub component: openshift-installer QA Contact: Pedro Amoedo <pamoedom>
Status: CLOSED ERRATA Docs Contact:
Severity: medium    
Priority: medium CC: bgilbert, gpei, miabbott, pamoedom
Version: 4.8   
Target Milestone: ---   
Target Release: 4.8.z   
Hardware: Unspecified   
OS: Unspecified   
Whiteboard:
Fixed In Version: Doc Type: If docs needed, set a value
Doc Text:
Story Points: ---
Clone Of: 2009018 Environment:
Last Closed: 2021-10-19 20:35:31 UTC Type: ---
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Embargoed:
Bug Depends On: 2009000, 2009018    
Bug Blocks:    

Description Micah Abbott 2021-09-29 17:31:13 UTC 
+++ This bug was initially created as a clone of Bug #2009018 +++

This BZ is created to facilitate the fix in 4.8.z; this problem does not actually affect 4.9.z

+++ This bug was initially created as a clone of Bug #2009000 +++

When the most recent RHCOS boot image bump for 4.8 landed, only the stream metadata was changed. This was done since `openshift-installer` itself was updated to use only the stream metadata.

https://github.com/openshift/installer/pull/5227

However, it appears that UPI jobs still rely on the "legacy" metadata:

https://github.com/openshift/installer/blob/release-4.8/images/installer/Dockerfile.upi.ci#L17


While the drift in the different metadata sources shouldn't break the UPI jobs, it creates a different workflow as the IPI jobs will start from a different set of RHCOS artifacts than the UPI jobs.

--- Additional comment from Micah Abbott on 2021-09-29 16:34:34 UTC ---

@gpei noticed this discrepancy first - https://github.com/openshift/installer/pull/5227#issuecomment-930010185

--- Additional comment from Micah Abbott on 2021-09-29 17:25:46 UTC ---

This issue only affects 4.8 and will be used as the source for the necessary backport BZs.

Closing as CURRENTRELEASE
Comment 4 Pedro Amoedo 2021-10-10 13:44:28 UTC 
Hi Micah, my apologies but I'm temporarily reverting back this BZ to "ON_QA" status because I've detected a possible issue that we need to discard first:

Deploying OCP 4.8.0-0.nightly-2021-10-09-130601 "Baremetal UPI dual-stack" via iPXE UEFI using "rhcos-48.84.202109241901-0-live-{kernel,rootfs}" images, the manual IPv6 configuration passed via kernel arguments is not having effect and all nodes are falling back to internal IPv6 ranges instead (fe80::3443:22c:4e3a:), which causes bootstrap failure, this was not happening with previous "rhcos-48.84.202106301921-0".

Can you please corroborate/report this?

BR
Comment 6 Pedro Amoedo 2021-10-11 13:26:54 UTC 
Hi again Micah, I've raised BZ#2012855 to investigate and/or corroborate the IPv4v6 behavior that I'm seeing on "rhcos-48.84.202109241901-0-live-{kernel,rootfs}" images, but regarding this BZ it makes no sense to block it because it was raised to correct the image inconsistency, which was already fixed, therefore I'm moving it to VERIFIED again and we'll track the other issue in the aforementioned BZ, sorry for the inconvenience.

Best Regards.
Comment 9 errata-xmlrpc 2021-10-19 20:35:31 UTC 
Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.

For information on the advisory (OpenShift Container Platform 4.8.15 bug fix update), and where to find the updated
files, follow the link below.

If the solution does not work for you, open a new bug report.

https://access.redhat.com/errata/RHBA-2021:3821