Marking as a regression - https://bugzilla.redhat.com/show_bug.cgi?id=1754892

This has zero user impact, to the best of my knowledge, thus, lowering severity. Please correct if you see any impact.

The Rook-Ceph Pods use a different image than the Ceph Pods, registry.redhat.io/ocs4/rook-ceph-rhel8-operator vs registry.redhat.io/rhceph/rhceph-4-rhel8. Any discrepancy in Ceph versions between the two are down to the DS build process and what dependencies the Rook-Ceph project pulls in.

That said, if there is no technical problem then I don't think this matters. If there was a significant difference then we would probably have already made sure they were at advanced enough versions to remove the problem.

Moving this to the rook component and OCS 4.7 in case there's further need for discussion.

Just like Jose said, the mismatch is on the DS build.
Nothing Rook can do at this point.

Also, I don't know how the build can always guarantee the exact same version between the Ceph image and Operator image Ceph packages.
If you update the operator image, you might just get newer Ceph packages and the running cluster might have "older" pin-point release packages.

Typically, higher Ceph packages is not an issue since they are compatible with earlier versions.

@Boris: We are suspecting a minor issue with the downstream builds: The rook-ceph-operator shows a different ceph build version (14.2.8-111) than the ceph component containers (osd/mon/mgr...) (14.2.8-91).

Can you comment?

We are running dnf update -y in rook Dockerfile to download any security fixes. However, we are consuming rhceph-4 repos in rook build which also updates ceph packages in that container. We can remove that line but we won't get any security fixes for the other packages anymore. Maybe, we could add some --exclude (or --disablerepo) flags to the command to avoid updating the ceph packages.

I pushed a fix for this issue to ocs-4.4 (and onwards) dist-git branches. Do we want to do another RC of OCS 4.5.2 for this? It would be a relatively simple rebuild (we would just have to rebuild rook and the operator bundle).

btw: We are still doing security updates in rook, I just disabled the rhceph repos to prevent the ceph packages from being updated.

This should already be fixed in 4.6.0 in the latest build (it did rebuild rook).

Do we know if we want to target 4.5.2, too? Anyway, this would be fixed by any 4.5 rebuild (i.e. even in 4.5.3 if we ever release).

If the fix is in 4.6 already can I retarget this to 4.6?
Unless another RC is needed? If so, this will be in only if one more RC will be required

Bug Fixed, all the pods display the same ceph version.


SetUp:
Provider: Vmware_Dynamic
OCP Version:4.6.0-0.nightly-2020-11-21-194817
OCS Version:ocs-operator.v4.6.0-160.ci


Test Process:
1.Check Ceph version on all relevnat components.
============================================================================================
$ oc rsh rook-ceph-mon-a-74b5fcb97d-s972f
sh-4.4# ceph -v
ceph version 14.2.8-111.el8cp (2e6029d57bc594eceba4751373da6505028c2650) nautilus (stable)
============================================================================================
$ oc rsh rook-ceph-mgr-a-5f8695cc48-7j7l7
sh-4.4# ceph -v
ceph version 14.2.8-111.el8cp (2e6029d57bc594eceba4751373da6505028c2650) nautilus (stable)
============================================================================================
$ oc rsh rook-ceph-mon-b-54fdc9c889-wj47l
sh-4.4# ceph -v
ceph version 14.2.8-111.el8cp (2e6029d57bc594eceba4751373da6505028c2650) nautilus (stable)
============================================================================================
$ oc rsh rook-ceph-mon-c-8bbd8f9d4-rdp7l
sh-4.4# ceph -v
ceph version 14.2.8-111.el8cp (2e6029d57bc594eceba4751373da6505028c2650) nautilus (stable)
============================================================================================
$ oc rsh rook-ceph-osd-0-75cb6d6fc8-6gvnk
sh-4.4# ceph -v
ceph version 14.2.8-111.el8cp (2e6029d57bc594eceba4751373da6505028c2650) nautilus (stable)
============================================================================================
$ oc rsh rook-ceph-osd-1-55545b4fcc-4m6t2
sh-4.4# ceph -v
ceph version 14.2.8-111.el8cp (2e6029d57bc594eceba4751373da6505028c2650) nautilus (stable)
============================================================================================
$ oc rsh rook-ceph-osd-2-5bbd9d949b-7fvzj
sh-4.4# ceph -v
ceph version 14.2.8-111.el8cp (2e6029d57bc594eceba4751373da6505028c2650) nautilus (stable)
============================================================================================
$ oc rsh rook-ceph-tools-85dc5f7bc8-tnlqj
sh-4.4# ceph -v
ceph version 14.2.8-111.el8cp (2e6029d57bc594eceba4751373da6505028c2650) nautilus (stable)
============================================================================================
$ oc rsh rook-ceph-operator-54f449df55-bzwrq
sh-4.4$ ceph -v
ceph version 14.2.8-111.el8cp (2e6029d57bc594eceba4751373da6505028c2650) nautilus (stable)
============================================================================================

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 (Moderate: Red Hat OpenShift Container Storage 4.6.0 security, bug fix, enhancement 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/RHSA-2020:5605