Eviction of stateful set about descheduler

[aveshagarwal]

@cesartl I dont think so, are they using some sort of local storage or anything that might be preventing them from eviction.

[cesartl]

There is only a mounted empty dir:

---
apiVersion: apps/v1beta1
kind: StatefulSet
metadata:
  name: redis-sentinel
spec:
  serviceName: redis-sentinel
  replicas: 3
  template:
    metadata:
      labels:
        app: redis-sentinel
      annotations:
        pod.beta.kubernetes.io/init-containers: '[{
            "name": "k8s-redis-ha-sentinel",
            "image": "tarosky/k8s-redis-ha:sentinel-3.0.1",
            "volumeMounts": [{"name": "opt", "mountPath": "/opt"}]
          }]'
    spec:
      containers:
        - name: redis-sentinel
          image: redis:3.2
          command:
            - /opt/bin/k8s-redis-ha-sentinel
          args:
            - /opt/sentinel.conf
          env:
            - name: SERVICE
              value: redis-server
            - name: SERVICE_PORT
              value: redis-server
          ports:
            - containerPort: 26379
              name: redis-sentinel
          readinessProbe:
            exec:
              command: ['redis-cli', '-p', '26379', 'info', 'server']
          volumeMounts:
            - name: opt
              mountPath: /opt
        - name: redis-sword
          image: tarosky/k8s-redis-ha:sword-3.0.1
          env:
            - name: SERVICE
              value: redis-server
            - name: SERVICE_PORT
              value: redis-server
            - name: SENTINEL
              value: redis-sentinel
            - name: SENTINEL_PORT
              value: redis-sentinel
          volumeMounts:
            - name: opt
              mountPath: /opt
      volumes:
        - name: opt
          emptyDir: {}

[aveshagarwal]

Yes it is because of empty dir: see here:
https://github.com/kubernetes-incubator/descheduler/blob/master/pkg/descheduler/pod/pods.go#L132

[cesartl]

ok thanks.

do you think it makes sense not to evict pod with empty dir? what's the reason behind it?

[cesartl]

Would it be possible to have an option to enable eviction of pods, even if they have an empty dir.
This is is causing us a lot of trouble.
Because our Redis pods are not evicted they sometimes end up on the same node and our application can go down because of that.

[aveshagarwal]

@cesartl sure having a flag is fine. If you would like to send a PR for it, that would be helpful, otherwise I will send a PR for it soon.

[fejta-bot]

Issues go stale after 90d of inactivity.
Mark the issue as fresh with /remove-lifecycle stale.
Stale issues rot after an additional 30d of inactivity and eventually close.

If this issue is safe to close now please do so with /close.

Send feedback to sig-testing, kubernetes/test-infra and/or fejta.
/lifecycle stale

[fejta-bot]

Stale issues rot after 30d of inactivity.
Mark the issue as fresh with /remove-lifecycle rotten.
Rotten issues close after an additional 30d of inactivity.

If this issue is safe to close now please do so with /close.

Send feedback to sig-testing, kubernetes/test-infra and/or fejta.
/lifecycle rotten

[fejta-bot]

Rotten issues close after 30d of inactivity.
Reopen the issue with /reopen.
Mark the issue as fresh with /remove-lifecycle rotten.

Send feedback to sig-testing, kubernetes/test-infra and/or fejta.
/close

[k8s-ci-robot]

@fejta-bot: Closing this issue.

In response to this:

Rotten issues close after 30d of inactivity.
Reopen the issue with /reopen.
Mark the issue as fresh with /remove-lifecycle rotten.

Send feedback to sig-testing, kubernetes/test-infra and/or fejta.
/close

Instructions 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/test-infra repository.