Bug 1989958

Summary: Unable to setup Stream 8 with md device at install/kickstart time (but working fine with CentOS Linux 8
Product: Red Hat Enterprise Linux 8 Reporter: farrotin
Component: python-blivetAssignee: Blivet Maintenance Team <blivet-maint-list>
Status: CLOSED DUPLICATE QA Contact: Release Test Team <release-test-team-automation>
Severity: high Docs Contact:
Priority: high    
Version: CentOS StreamCC: bstinson, dledford, dnehring, jdonohue, jwboyer, ncroxon, vtrefny
Target Milestone: betaFlags: pm-rhel: mirror+
Target Release: ---   
Hardware: x86_64   
OS: Unspecified   
Whiteboard:
Fixed In Version: Doc Type: If docs needed, set a value
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2021-09-16 14:42:23 UTC Type: Bug
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Embargoed:
Attachments:
Description Flags
anaconda log screenshot none

Description farrotin 2021-08-04 12:43:42 UTC
Created attachment 1810852 [details]
anaconda log screenshot

Description of problem:

We (CentOS Infra team) are using the same kickstart to deploy CentOS Stream 8 on bare-metal nodes and when we tried to reinstall nodes (from CentOS 7 to CentOS Stream 8), it tracebacks, but same kickstart works fine with CentOS Linux 8 (that is going EOL soon)

Version-Release number of selected component (if applicable):

CentOS Stream 8

How reproducible:

create a kickstart that will configuring raid1 md devices and kick the install


Additional info:

Part of the common kickstart.j2 template rendered by ansible to deploy the node : 

<snip>
# Partition clearing information
clearpart --all --initlabel
zerombr

# Disk partitioning information
# Adding first reqpart to automatically add /boot/efi or prepboot for aarch64, uefi, or IBM Power architectures
reqpart

# Using software raid 1
part raid.603 --fstype="mdmember" --ondisk=sdb --size=20000 --grow
part raid.469 --fstype="mdmember" --ondisk=sda --size=1024
part raid.597 --fstype="mdmember" --ondisk=sda --size=20000 --grow
part raid.475 --fstype="mdmember" --ondisk=sdb --size=1024
raid pv.609 --device=pv00 --fstype="lvmpv" --level=RAID1 raid.597 raid.603
raid /boot --device=boot --fstype="ext4" --level=RAID1 raid.469 raid.475
volgroup vg_kempty-n28 --pesize=4096 pv.609
</snip>

Adding screenshot from supermicro java kvm viewer for the log/traceback from anaconda, while it deploys fine with current CentOS Linux 8.4

The "mdadm: specifying chunk size is forbidden for this level" seems to be the pointer

Comment 1 farrotin 2021-08-10 12:21:21 UTC
seems to be related to https://bugzilla.redhat.com/show_bug.cgi?id=1984335 , so let's keep fingers crossed, as it's still not possible to deploy stream 8 at this stage, when using md device that is

Comment 2 Dirk Nehring 2021-08-11 15:59:39 UTC
I can confirm the problem, SW RAID is not working anymore in CentOS8 Stream and kickstart!

Perhaps we should upgrade to the latest version: 

https://mirrors.edge.kernel.org/pub/linux/utils/raid/mdadm/mdadm-4.2-rc2.tar.gz

I think the problem is related to the following commit:

https://git.kernel.org/pub/scm/utils/mdadm/mdadm.git/commit/?id=5b30a34aa4b5ea7a8202314c1d737ec4a481c127

Comment 3 Dirk Nehring 2021-08-12 06:34:55 UTC
Fixed in blivet:

https://github.com/storaged-project/blivet/commit/dc1e2fe7783748528cac2f7aa516c89d1959b052

Setting chunk size for RAID 1 doesn't make sense and latest
mdadm started returning error instead of ignoring the --chunk
option when creating an array.

So the problem seems to be related to blivet.

Comment 4 farrotin 2021-08-12 08:15:59 UTC
I see that a new python-blivet correcting this was built : https://koji.mbox.centos.org/koji/buildinfo?buildID=18685
With changelog : 

* Mon Aug 02 2021 Vojtech Trefny <vtrefny> - 3.4.0-4
- Do not set chunk size for RAID 1
  Resolves: rhbz#1987170

I'll retry a redeploy with 8-stream but I can say that by looking at the QA stream logs, it passed this time, so should be "out".

Comment 6 Vojtech Trefny 2021-09-16 14:29:19 UTC
I did a quick installation with the latest C8S (20210907) and I was able to finish the installation with RAID 1 (with python3-blivet-3.4.0-5.el8) so I think we can close this now. @farrotin I'm not sure how this works with CentOS Stream, can we just close this as duplicate of the RHEL bug (https://bugzilla.redhat.com/show_bug.cgi?id=1987170) or is there some special process for CentOS?

Comment 7 farrotin 2021-09-16 14:33:05 UTC
Well, yes it can be closed as it was fixed in stream-8 as soon as mentioned pkgs were rebuilt/pushed through centos infra

Comment 8 Vojtech Trefny 2021-09-16 14:42:23 UTC

*** This bug has been marked as a duplicate of bug 1987170 ***