Customer (Moody's Corp) in case 02811513 recently did a series of upgrades. The upgrade process was very slow (~10 hours) due to bandwidth limits or restrictions in their external Proxy.

Initially, it was difficult to identify the slow Image Pulls accounted for most of the slow upgrade. We eventually checked events and noticed the image pull message timestamp differences.

Would it be possible to alert of very slow image pull operations in openshift-* namespaces or would it be difficult to identify what is a slow vs normal image pull speed?