Closed ecliptik closed 6 years ago
@StefanScherer I think cgroup_enable=cpuset
can stay, I didn't update it on my rpi nodes, just replaced cgroup_enable=memory
with cgroup_memory=1
and kubeadm pre-flight checks all passed after that.
@ecliptik Yes thanks. Just after typing my question I found it myself after looking closer to the keywords :-) I think @DieterReuter will merge your PR asap and publish a fixed release. Thank you!
Re-created my rpi cluster tonight using hypriotos-rpi64-v20180429-184538.img
and when running kubeadm
it failed with,
CGROUPS_MEMORY: missing
[WARNING SystemVerification]: docker version is greater than the most recently validated version. Docker version: 18.04.0-ce. Max validated version: 17.03
[WARNING FileExisting-crictl]: crictl not found in system path
Suggestion: go get github.com/kubernetes-incubator/cri-tools/cmd/crictl
[preflight] Some fatal errors occurred:
[ERROR SystemVerification]: missing cgroups: memory
[preflight] If you know what you are doing, you can make a check non-fatal with `--ignore-preflight-errors=...`
Looking at the comments on the original kernel thread it seems the cgroup_memory=1
was kept for the 4.9.x kernels, but for newer kernels it went back to cgroup_enable=memory
.
I updated /boot/cmdline.txt
and changed cgroup_memory=1
to cgroup_enable=memory
, rebooted and kubeadm
passed all checks. Should probably revert #41 in order to use Kubernetes on images using 4.14 kernels
v20171130-113423 sets
/boot/cmdline.txt
withcgroup_enable=memory
, which does not setup the memory cgroup properly and causes kubeadm to fail pre-flight checks.This has changed in recent kernels according to https://archlinuxarm.org/forum/viewtopic.php?f=15&t=12086#p57035.
To fix this,
/boot/cmdline.txt
should usecgroup_memory=1
instead.Updated this on a v20171130-113423 Raspberry Pi and rebooted, and kubeadm passes pre-flight checks and setups up k8s properly.