-
Bug
-
Resolution: Not a Bug
-
Undefined
-
None
-
4.20
-
Quality / Stability / Reliability
-
False
-
-
None
-
Important
-
None
-
aarch64
-
None
-
None
-
None
-
None
-
None
-
None
-
None
-
None
-
None
-
None
-
None
Description of problem:
The issue is that while attempting to modify the performance profile, an MCP (Machine Config Pool) update is triggered. Upon completion of this update, the corresponding nodes should reflect the changes made in the performance profile. However, it appears that the fields targeted for modification are not taking effect on the nodes. Specifically, kernelPageSize was altered from 64k to 4k, and hugepagesize from 512M to 64K. The problem is that even after the MCP finished updating, the nodes themselves did not reflect these changes.
Version-Release number of selected component (if applicable):
4.20
How reproducible:
Sometimes
Steps to Reproduce:
1. Apply a performance profile
2. Change any field in the performance profile
3. Check inside the nodes if the changes took place
Actual results:
Expected results:
Additional info:
In these logs: The kernelPageSize and hugepagesize values defined in the performance profile (4K and 64K respectively) are not being applied to the nodes. Instead, the nodes are retaining their previous values (64K and 512M), indicating that the intended changes haven't taken effect.
[kni@registry 14_arm]$ oc get no,mcp; echo; oc get pods -n node-inspector-ns; echo; oc get performanceprofile -o json | jq ".items[0].spec.kernelPageSize"; echo; oc get performanceprofile -o json | jq ".items[0].spec.hugepages.pages[0].size" NAME STATUS ROLES AGE VERSION node/master-0.kni-qe-93.telcoqe.eng.rdu2.dc.redhat.com Ready control-plane,master 4d19h v1.33.2 node/master-1.kni-qe-93.telcoqe.eng.rdu2.dc.redhat.com Ready control-plane,master 4d19h v1.33.2 node/master-2.kni-qe-93.telcoqe.eng.rdu2.dc.redhat.com Ready control-plane,master 4d19h v1.33.2 node/worker-0.kni-qe-93.telcoqe.eng.rdu2.dc.redhat.com Ready worker 4d19h v1.33.2 node/worker-1.kni-qe-93.telcoqe.eng.rdu2.dc.redhat.com Ready worker 4d19h v1.33.2 NAME CONFIG UPDATED UPDATING DEGRADED MACHINECOUNT READYMACHINECOUNT UPDATEDMACHINECOUNT DEGRADEDMACHINECOUNT AGE machineconfigpool.machineconfiguration.openshift.io/master rendered-master-00e123d0ca70c700276a6474fe2e2a6e True False False 3 3 3 0 4d19h machineconfigpool.machineconfiguration.openshift.io/worker rendered-worker-3f1f8d886973f4a5ee702fb6b9ee2698 True False False 2 2 2 0 4d19h NAME READY STATUS RESTARTS AGE node-inspector-bljkq 1/1 Running 0 87m node-inspector-ktsbv 1/1 Running 4 87m node-inspector-ns9cz 1/1 Running 0 87m node-inspector-svbc4 1/1 Running 4 87m node-inspector-t64wf 1/1 Running 0 87m "4k" "64k" [kni@registry 14_arm]$ oc debug node/worker-0.kni-qe-93.telcoqe.eng.rdu2.dc.redhat.com Starting pod/worker-0kni-qe-93telcoqeengrdu2dcredhatcom-debug-897wz ... To use host binaries, run `chroot /host`. Instead, if you need to access host namespaces, run `nsenter -a -t 1`. Pod IP: 10.6.159.12 If you don't see a command prompt, try pressing enter. sh-5.1# getconf PAGESIZE 65536 sh-5.1# grep Hugepagesize: /proc/meminfo Hugepagesize: 524288 kB sh-5.1# exit Removing debug pod ... [kni@registry 14_arm]$ oc debug node/worker-1.kni-qe-93.telcoqe.eng.rdu2.dc.redhat.com Starting pod/worker-1kni-qe-93telcoqeengrdu2dcredhatcom-debug-nxnrs ... To use host binaries, run `chroot /host`. Instead, if you need to access host namespaces, run `nsenter -a -t 1`. Pod IP: 10.6.159.13 If you don't see a command prompt, try pressing enter. sh-5.1# chroot /host sh-5.1# getconf PAGESIZE 65536 sh-5.1# grep Hugepagesize: /proc/meminfo Hugepagesize: 524288 kB