[release-4.21] Fix periodic task processing a node via the wrong interface - #491
Conversation
node_periodic matched a node's interface with isinstance(), so a periodic defined on a base interface also matched nodes whose interface is a subclass. When both a base and a subclass that inherits the periodic are enabled (e.g. redfish and idrac-redfish BIOS or RAID), each interface is collected as its own task and the base task also matched the subclass's nodes. The same node was then processed by two tasks, issuing duplicate resume RPCs that advanced past the next clean, service, or deploy step (silently skipping it). Fix: match the node's interface by exact type instead of isinstance. The two tasks now cover disjoint node sets -- the redfish task only matches RedfishBIOS nodes and the idrac-redfish task only matches DracRedfishBIOS nodes -- so no node is processed twice, and each node is handled by its own interface instance so subclass overrides apply. This keeps the existing one-task-per-interface-instance collection model untouched; the whole functional change is the match predicate. Assisted-By: Claude Opus 4.8 Change-Id: I26e66fce55cb013e66b6da3174d4ba6bb64a675e Signed-off-by: Jacob Anders <jacob-anders-dev@proton.me> (cherry picked from commit 5fe5a9b) (cherry picked from commit de95d98) (cherry picked from commit f785d64)
|
[APPROVALNOTIFIER] This PR is NOT APPROVED This pull-request has been approved by: jacob-anders The full list of commands accepted by this bot can be found here. DetailsNeeds approval from an approver in each of these files:Approvers can indicate their approval by writing |
|
@jacob-anders: all tests passed! Full PR test history. Your PR dashboard. DetailsInstructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes-sigs/prow repository. I understand the commands that are listed here. |
node_periodic matched a node's interface with isinstance(), so a periodic defined on a base interface also matched nodes whose interface is a subclass. When both a base and a subclass that inherits the periodic are enabled (e.g. redfish and idrac-redfish BIOS or RAID), each interface is collected as its own task and the base task also matched the subclass's nodes. The same node was then processed by two tasks, issuing duplicate resume RPCs that advanced past the next clean, service, or deploy step (silently skipping it).
Fix: match the node's interface by exact type instead of isinstance. The two tasks now cover disjoint node sets -- the redfish task only matches RedfishBIOS nodes and the idrac-redfish task only matches DracRedfishBIOS nodes -- so no node is processed twice, and each node is handled by its own interface instance so subclass overrides apply.
This keeps the existing one-task-per-interface-instance collection model untouched; the whole functional change is the match predicate.
Assisted-By: Claude Opus 4.8
Change-Id: I26e66fce55cb013e66b6da3174d4ba6bb64a675e
(cherry picked from commit 5fe5a9b) (cherry picked from commit de95d98) (cherry picked from commit f785d64)