It would significantly benefit customers and FDF/ODF support to have a mechanism to check for the needed images prior to an upgrade (to include the odf must-gather image), to proactively identify issues such as ( ImagePullBackOff ). regarding the odf-must gather specifically, we should not block an upgrade because odf must-gather image is missing, but it should be a clear warning which is clearly visible to the customer, and maybe even needs to acknowledge to step over.

We have a significant amount of customers who are running disconnected environments which means they are relying on a local image registry. This means that prior to an upgrade all the images must be successfully mirrored to their local image registry so that they are available come time for the upgrade. 

Being able to do this prior to an upgrade, allows customers to proactively resolve images issues. Rather then the current reactive method which can result in an impact which then triggers high severity cases for impacted environments because services are down with ImagePullBackOff for example. This impact is not just felt by our support team but influences other support teams requiring additional coordination with support teams such as OpenShift support, Fusion, IBM cloudpak teams, etc which are engaged in a high severity fashion due to customer impact and closing customer maintenance windows.

-------------------

Additionally, from the support perspective our first step in analysis is usually to request an odf must-gather to review the various detailed reports, logs, status' that are captured. In disconnected environments customers MUST know and seek out the odf must-gather image to mirror to their local registry so that this capture can be generated and provided. We see multiple times a week customers can not provide the odf must-gather and usually that is because it is not in their local image registry.

– Adding odf must-gather as part of that image check will save us and the customer cycles on back and forth to then begin starting to analyze.

-----------------

In Standalone Ceph, cephadm has a mechanism in place that similarly does this already, just prior to an upgrade. Here is an excerpt from IBM Ceph Standalone docs:

Thanks,