Hide Forgot
- What is the nature and description of the request? As an admin I require the ability to free up my PV's if they are supposed to be auto-recycled and were not for some reason. - Why does the customer need this? (List the business requirements here) If PV's are supposed to be recycled but are stuck in bound state for some reason there needs to be a way to clear it using the pods. - How would the customer like to achieve this? (List the functional requirements here) Kick start a recycler pod to clear a single PV - For each functional requirement listed, specify how Red Hat and the customer can test to confirm the requirement is successfully implemented. Bind a PV and kill the default recycler pod before it succeeds (leaving it still bound but with no PVC) and then manually start that recycler pod. - Is there already an existing RFE upstream or in Red Hat Bugzilla? Unclear - Who is the customer behind the request? Support - Does the customer have any specific timeline dependencies and which release would they like to target (i.e. RHEL5, RHEL6)? Soon - Is the sales team involved in this request and do they have any additional input? No - Would the customer be able to assist in testing this functionality if implemented? Yes
In OpenShift 3.2, we re-tried recycling only few times and then we marked the PV as Failed. In OpenShift 3.3 and 3.4, we retry recycling forever, until an admin does something (e.g. manually deletes the PV). OpenShift trigger for recycling in 3.3 and above is: - PV.Spec.ClaimRef points to a PVC, i.e. full PV.Spec.ClaimRef must be filled, incl. its UID - AND the referenced PVC does not exist PV.Status is not considered at all in 3.3! It can be Bound, Released, Failed or even Available. All that matters is PV.Spec.ClaimRef. In 3.2 the behavior was a bit different and I'd rather not touch that code, we've rewritten it from scratch in 3.3. Is it a valid solution for this bug?
I'm closing this bug, original support ticket is closed.