Asking for help? Comment out what you need so we can get more information to help you!
Cluster information:
Kubernetes version: 1.27
Cloud being used: bare-metal
Hello,
I followed the instruction from Manage HugePages | Kubernetes to make huge pages available on a node by setting it in the grub file.
In my yaml file, I set the limit as follows
containers:
- name: foo2
image: myrepo.comfoo2:latest
volumeMounts:
- mountPath: /hugepages
name: hugepage
resources:
limits:
hugepages-2Mi: 20Mi
memory: 4000Mi
requests:
memory: 4000Mi
- name: foo
image: myrepo.com/foo:latest
volumes:
- name: hugepage
emptyDir:
medium: HugePages
However, my container is able to exceed the huge page limit of 20Mi set and use up 70 pages:
grep -i hugepages /proc/meminfo
AnonHugePages: 200704 kB
ShmemHugePages: 0 kB
HugePages_Total: 3000
HugePages_Free: 2930
HugePages_Rsvd: 0
HugePages_Surp: 0
Hugepagesize: 2048 kB
The pages are loaded to shared memory using huge pages:
------ Shared Memory Segments --------
key shmid owner perms bytes nattch status
0x0000200 17 root 600 145354312 2
How come the pod does not forbid the container from exceeding this limit?