sealos安装集群，镜像是通过containerd管理，如何修改containerd的rootdir目录，有环境变量支持吗？ || sealos installs the cluster, and the image is managed through containerd. How to modify the rootdir directory of containerd? Is there any environment variable support?

[fenghanwei]

我先安装containerd，sealos会提示已经安装containerd，请先卸载！我希望能够在安装集群的时候就修改containerd的rootdir（默认是/var/lib）,而不是集群安装成功了，再去修改，这可能会导致集群组件服务异常（端口被占用，无法启动），必须要手动停止组件服务，让其自动重启才行（这就比较麻烦了）。

kubectl get pod -A|grep -v Run
NAMESPACE NAME READY STATUS RESTARTS AGE
calico-system calico-typha-df68869c-lt59j 0/1 CrashLoopBackOff 9 (4m31s ago) 30m
calico-system calico-typha-df68869c-qp96d 0/1 CrashLoopBackOff 9 (4m44s ago) 30m
kube-system etcd-host-172-16-200-140 0/1 CrashLoopBackOff 23 (2m18s ago) 31m
kube-system kube-apiserver-host-172-16-200-140 0/1 CrashLoopBackOff 24 (99s ago) 31m
kube-system kube-controller-manager-host-172-16-200-140 0/1 CrashLoopBackOff 23 (115s ago) 31m
kube-system kube-scheduler-host-172-16-200-140 0/1 CrashLoopBackOff 23 (118s ago) 31m

kubectl logs -f -n kube-system etcd-host-172-16-200-140

{"level":"error","ts":"2025-12-12T07:39:47.509410Z","caller":"embed/etcd.go:538","msg":"creating peer listener failed","error":"listen tcp 172.16.200.140:2380: bind: address already in

kubectl logs -f -n kube-system kube-scheduler-host-172-16-200-140
I1212 15:40:07.780626 1 serving.go:380] Generated self-signed cert in-memory
E1212 15:40:07.783294 1 run.go:74] "command failed" err="failed to create listener: failed to listen on 0.0.0.0:10259: listen tcp 0.0.0.0:10259: bind: address already in use

---

I installed containerd first, sealos will prompt that containerd has been installed, please uninstall it first! I hope to be able to modify the rootdir of containerd (the default is /var/lib) when installing the cluster. If I modify it after the cluster is installed successfully, this may cause the cluster component service to be abnormal (the port is occupied and cannot be started). The component service must be manually stopped and allowed to restart automatically (this is more troublesome).

kubectl get pod -A|grep -v Run
NAMESPACE NAME READY STATUS RESTARTS AGE
calico-system calico-typha-df68869c-lt59j 0/1 CrashLoopBackOff 9 (4m31s ago) 30m
calico-system calico-typha-df68869c-qp96d 0/1 CrashLoopBackOff 9 (4m44s ago) 30m
kube-system etcd-host-172-16-200-140 0/1 CrashLoopBackOff 23 (2m18s ago) 31m
kube-system kube-apiserver-host-172-16-200-140 0/1 CrashLoopBackOff 24 (99s ago) 31m
kube-system kube-controller-manager-host-172-16-200-140 0/1 CrashLoopBackOff 23 (115s ago) 31m
kube-system kube-scheduler-host-172-16-200-140 0/1 CrashLoopBackOff 23 (118s ago) 31m

kubectl logs -f -n kube-system etcd-host-172-16-200-140

{"level":"error","ts":"2025-12-12T07:39:47.509410Z","caller":"embed/etcd.go:538","msg":"creating peer listener failed","error":"listen tcp 172.16.200.140:2380: bind: address already in

kubectl logs -f -n kube-system kube-scheduler-host-172-16-200-140
I1212 15:40:07.780626 1 serving.go:380] Generated self-signed cert in-memory
E1212 15:40:07.783294 1 run.go:74] "command failed" err="failed to create listener: failed to listen on 0.0.0.0:10259: listen tcp 0.0.0.0:10259: bind: address already in use

[fenghanwei]

sealos：5.0.1
k8s：1.30.9