[jira] [Commented] (YUNIKORN-1253) PVCs won't get past WaitForFirstConsumer with Apache Yunikorn
[ https://issues.apache.org/jira/browse/YUNIKORN-1253?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17649605#comment-17649605 ] Chen Yu Teng commented on YUNIKORN-1253: Update. Hi [~ccondit] , sorry of late update. I apply the pod adopting ephemeral and the yunikorn would correctly schedule the pod after yunikorn bind pvc created by template and pv. The version of yunikorn and kubernetes is following: 1. Kubernetes version is 1.21.1 2. latest Yunikorn-k8shim 3. latest Yunikorn-core > PVCs won't get past WaitForFirstConsumer with Apache Yunikorn > - > > Key: YUNIKORN-1253 > URL: https://issues.apache.org/jira/browse/YUNIKORN-1253 > Project: Apache YuniKorn > Issue Type: Bug > Components: shim - kubernetes >Reporter: Alex Kalenyuk >Assignee: Chen Yu Teng >Priority: Major > Attachments: case01_schedulingWhenPVisNotCreated.log, > case01_yunikorn_scheduling_decision.log, pv.yaml, storageclass.yaml > > > It seems that with Apache Yunikorn, WaitForFirstConsumer volume binding > storage classes are not supported (not sure if this is intended or not). > This makes it problematic to use storage that is not globally accessible from > all nodes: > [https://kubernetes.io/docs/concepts/storage/storage-classes/#volume-binding-mode] > To reproduce a simple failure to use a WaitForFirstConsumer binding-backed > PVC: > ``` > kind: Pod > apiVersion: v1 > metadata: > name: test > namespace: default > labels: > app: sleep > applicationId: "sleep0001-node-selector-test" > queue: "root.sandbox" > spec: > schedulerName: yunikorn > nodeSelector: > storage/ssd: 'true' > containers: > - name: test > resources: > limits: > cpu: 1 > memory: 1G > requests: > cpu: 1 > memory: 1G > image: busybox > command: > - sleep > - '100' > volumeMounts: > - name: scratch-volume > mountPath: /data > volumes: > - name: scratch-volume > ephemeral: > volumeClaimTemplate: > spec: > accessModes: > - ReadWriteOnce > resources: > requests: > storage: 1Gi > storageClassName: hostpath-provisioner > volumeMode: Filesystem > ``` > Storage used: > https://github.com/kubevirt/hostpath-provisioner-operator > A similar issue was spotted in: > [https://github.com/kubernetes/kubernetes/issues/86262] > And this PR seems to introduce the VolumeBinding filter but comments it out: > [https://github.com/apache/yunikorn-k8shim/pull/313] > > I might be off with the "Bug" type here so feel free to correct me; > My thinking was that if introducing support for WFFC is trivial, this may > make sense to exist in older versions too. -- This message was sent by Atlassian Jira (v8.20.10#820010) - To unsubscribe, e-mail: issues-unsubscr...@yunikorn.apache.org For additional commands, e-mail: issues-h...@yunikorn.apache.org
[jira] [Commented] (YUNIKORN-1253) PVCs won't get past WaitForFirstConsumer with Apache Yunikorn
[ https://issues.apache.org/jira/browse/YUNIKORN-1253?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17649431#comment-17649431 ] Craig Condit commented on YUNIKORN-1253: [~Yukali] were you able to verify that this is fixed after YUNIKORN-1196? > PVCs won't get past WaitForFirstConsumer with Apache Yunikorn > - > > Key: YUNIKORN-1253 > URL: https://issues.apache.org/jira/browse/YUNIKORN-1253 > Project: Apache YuniKorn > Issue Type: Bug > Components: shim - kubernetes >Reporter: Alex Kalenyuk >Assignee: Chen Yu Teng >Priority: Major > Attachments: case01_schedulingWhenPVisNotCreated.log, > case01_yunikorn_scheduling_decision.log, pv.yaml, storageclass.yaml > > > It seems that with Apache Yunikorn, WaitForFirstConsumer volume binding > storage classes are not supported (not sure if this is intended or not). > This makes it problematic to use storage that is not globally accessible from > all nodes: > [https://kubernetes.io/docs/concepts/storage/storage-classes/#volume-binding-mode] > To reproduce a simple failure to use a WaitForFirstConsumer binding-backed > PVC: > ``` > kind: Pod > apiVersion: v1 > metadata: > name: test > namespace: default > labels: > app: sleep > applicationId: "sleep0001-node-selector-test" > queue: "root.sandbox" > spec: > schedulerName: yunikorn > nodeSelector: > storage/ssd: 'true' > containers: > - name: test > resources: > limits: > cpu: 1 > memory: 1G > requests: > cpu: 1 > memory: 1G > image: busybox > command: > - sleep > - '100' > volumeMounts: > - name: scratch-volume > mountPath: /data > volumes: > - name: scratch-volume > ephemeral: > volumeClaimTemplate: > spec: > accessModes: > - ReadWriteOnce > resources: > requests: > storage: 1Gi > storageClassName: hostpath-provisioner > volumeMode: Filesystem > ``` > Storage used: > https://github.com/kubevirt/hostpath-provisioner-operator > A similar issue was spotted in: > [https://github.com/kubernetes/kubernetes/issues/86262] > And this PR seems to introduce the VolumeBinding filter but comments it out: > [https://github.com/apache/yunikorn-k8shim/pull/313] > > I might be off with the "Bug" type here so feel free to correct me; > My thinking was that if introducing support for WFFC is trivial, this may > make sense to exist in older versions too. -- This message was sent by Atlassian Jira (v8.20.10#820010) - To unsubscribe, e-mail: issues-unsubscr...@yunikorn.apache.org For additional commands, e-mail: issues-h...@yunikorn.apache.org
[jira] [Commented] (YUNIKORN-1253) PVCs won't get past WaitForFirstConsumer with Apache Yunikorn
[ https://issues.apache.org/jira/browse/YUNIKORN-1253?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17625143#comment-17625143 ] Chen Yu Teng commented on YUNIKORN-1253: Hi [~vovaz] , thanks for the information. I traced the comments in YUNIKORN-1196 PR. I will recheck whether this issue could be reproduced in my environment after updating the dependencies. > PVCs won't get past WaitForFirstConsumer with Apache Yunikorn > - > > Key: YUNIKORN-1253 > URL: https://issues.apache.org/jira/browse/YUNIKORN-1253 > Project: Apache YuniKorn > Issue Type: Bug > Components: shim - kubernetes >Reporter: Alex Kalenyuk >Assignee: Chen Yu Teng >Priority: Major > Attachments: case01_schedulingWhenPVisNotCreated.log, > case01_yunikorn_scheduling_decision.log, pv.yaml, storageclass.yaml > > > It seems that with Apache Yunikorn, WaitForFirstConsumer volume binding > storage classes are not supported (not sure if this is intended or not). > This makes it problematic to use storage that is not globally accessible from > all nodes: > [https://kubernetes.io/docs/concepts/storage/storage-classes/#volume-binding-mode] > To reproduce a simple failure to use a WaitForFirstConsumer binding-backed > PVC: > ``` > kind: Pod > apiVersion: v1 > metadata: > name: test > namespace: default > labels: > app: sleep > applicationId: "sleep0001-node-selector-test" > queue: "root.sandbox" > spec: > schedulerName: yunikorn > nodeSelector: > storage/ssd: 'true' > containers: > - name: test > resources: > limits: > cpu: 1 > memory: 1G > requests: > cpu: 1 > memory: 1G > image: busybox > command: > - sleep > - '100' > volumeMounts: > - name: scratch-volume > mountPath: /data > volumes: > - name: scratch-volume > ephemeral: > volumeClaimTemplate: > spec: > accessModes: > - ReadWriteOnce > resources: > requests: > storage: 1Gi > storageClassName: hostpath-provisioner > volumeMode: Filesystem > ``` > Storage used: > https://github.com/kubevirt/hostpath-provisioner-operator > A similar issue was spotted in: > [https://github.com/kubernetes/kubernetes/issues/86262] > And this PR seems to introduce the VolumeBinding filter but comments it out: > [https://github.com/apache/yunikorn-k8shim/pull/313] > > I might be off with the "Bug" type here so feel free to correct me; > My thinking was that if introducing support for WFFC is trivial, this may > make sense to exist in older versions too. -- This message was sent by Atlassian Jira (v8.20.10#820010) - To unsubscribe, e-mail: issues-unsubscr...@yunikorn.apache.org For additional commands, e-mail: issues-h...@yunikorn.apache.org
[jira] [Commented] (YUNIKORN-1253) PVCs won't get past WaitForFirstConsumer with Apache Yunikorn
[ https://issues.apache.org/jira/browse/YUNIKORN-1253?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17624218#comment-17624218 ] Vladimir Zheltov commented on YUNIKORN-1253: Hi, wanted to point out that the issue was solved in this pr [YUNIKORN-1196] [https://github.com/apache/yunikorn-k8shim/pull/466] Generic ephemeral volumes were added in Kubernetes 1.23 and building yunikorn-k8shim against version 1.23.12 seems to have solved this issue. I have tested it on openshift 4.10.15 (Which is using v1.23.5+9ce5071) and it works without any apparent issue > PVCs won't get past WaitForFirstConsumer with Apache Yunikorn > - > > Key: YUNIKORN-1253 > URL: https://issues.apache.org/jira/browse/YUNIKORN-1253 > Project: Apache YuniKorn > Issue Type: Bug > Components: shim - kubernetes >Reporter: Alex Kalenyuk >Assignee: Chen Yu Teng >Priority: Major > Attachments: case01_schedulingWhenPVisNotCreated.log, > case01_yunikorn_scheduling_decision.log, pv.yaml, storageclass.yaml > > > It seems that with Apache Yunikorn, WaitForFirstConsumer volume binding > storage classes are not supported (not sure if this is intended or not). > This makes it problematic to use storage that is not globally accessible from > all nodes: > [https://kubernetes.io/docs/concepts/storage/storage-classes/#volume-binding-mode] > To reproduce a simple failure to use a WaitForFirstConsumer binding-backed > PVC: > ``` > kind: Pod > apiVersion: v1 > metadata: > name: test > namespace: default > labels: > app: sleep > applicationId: "sleep0001-node-selector-test" > queue: "root.sandbox" > spec: > schedulerName: yunikorn > nodeSelector: > storage/ssd: 'true' > containers: > - name: test > resources: > limits: > cpu: 1 > memory: 1G > requests: > cpu: 1 > memory: 1G > image: busybox > command: > - sleep > - '100' > volumeMounts: > - name: scratch-volume > mountPath: /data > volumes: > - name: scratch-volume > ephemeral: > volumeClaimTemplate: > spec: > accessModes: > - ReadWriteOnce > resources: > requests: > storage: 1Gi > storageClassName: hostpath-provisioner > volumeMode: Filesystem > ``` > Storage used: > https://github.com/kubevirt/hostpath-provisioner-operator > A similar issue was spotted in: > [https://github.com/kubernetes/kubernetes/issues/86262] > And this PR seems to introduce the VolumeBinding filter but comments it out: > [https://github.com/apache/yunikorn-k8shim/pull/313] > > I might be off with the "Bug" type here so feel free to correct me; > My thinking was that if introducing support for WFFC is trivial, this may > make sense to exist in older versions too. -- This message was sent by Atlassian Jira (v8.20.10#820010) - To unsubscribe, e-mail: issues-unsubscr...@yunikorn.apache.org For additional commands, e-mail: issues-h...@yunikorn.apache.org
[jira] [Commented] (YUNIKORN-1253) PVCs won't get past WaitForFirstConsumer with Apache Yunikorn
[ https://issues.apache.org/jira/browse/YUNIKORN-1253?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17624207#comment-17624207 ] Chen Yu Teng commented on YUNIKORN-1253: Update: I finished the [sanityCheckBeforeScheduling]([https://github.com/apache/yunikorn-k8shim/blob/master/pkg/cache/task.go#L508-L529|https://github.com/apache/yunikorn-k8shim/blob/master/pkg/cache/task.go#L508-L529)])part, so yunikorn should be waiting for the generic pvc creation. My implementation is [here](https://github.com/0yukali0/incubator-yunikorn-k8shim/tree/YUNIKORN-1253) However, [AssumePod](https://github.com/apache/yunikorn-k8shim/blob/master/pkg/cache/context.go#L452-L466[)|https://github.com/apache/yunikorn-k8shim/blob/master/pkg/cache/context.go#L452-L466] did not find the unbound pvc controlled by the pod and cache recorded the pod allVolumeBound. Is there any sugesstions? > PVCs won't get past WaitForFirstConsumer with Apache Yunikorn > - > > Key: YUNIKORN-1253 > URL: https://issues.apache.org/jira/browse/YUNIKORN-1253 > Project: Apache YuniKorn > Issue Type: Bug > Components: shim - kubernetes >Reporter: Alex Kalenyuk >Assignee: Chen Yu Teng >Priority: Major > Attachments: case01_schedulingWhenPVisNotCreated.log, > case01_yunikorn_scheduling_decision.log, pv.yaml, storageclass.yaml > > > It seems that with Apache Yunikorn, WaitForFirstConsumer volume binding > storage classes are not supported (not sure if this is intended or not). > This makes it problematic to use storage that is not globally accessible from > all nodes: > [https://kubernetes.io/docs/concepts/storage/storage-classes/#volume-binding-mode] > To reproduce a simple failure to use a WaitForFirstConsumer binding-backed > PVC: > ``` > kind: Pod > apiVersion: v1 > metadata: > name: test > namespace: default > labels: > app: sleep > applicationId: "sleep0001-node-selector-test" > queue: "root.sandbox" > spec: > schedulerName: yunikorn > nodeSelector: > storage/ssd: 'true' > containers: > - name: test > resources: > limits: > cpu: 1 > memory: 1G > requests: > cpu: 1 > memory: 1G > image: busybox > command: > - sleep > - '100' > volumeMounts: > - name: scratch-volume > mountPath: /data > volumes: > - name: scratch-volume > ephemeral: > volumeClaimTemplate: > spec: > accessModes: > - ReadWriteOnce > resources: > requests: > storage: 1Gi > storageClassName: hostpath-provisioner > volumeMode: Filesystem > ``` > Storage used: > https://github.com/kubevirt/hostpath-provisioner-operator > A similar issue was spotted in: > [https://github.com/kubernetes/kubernetes/issues/86262] > And this PR seems to introduce the VolumeBinding filter but comments it out: > [https://github.com/apache/yunikorn-k8shim/pull/313] > > I might be off with the "Bug" type here so feel free to correct me; > My thinking was that if introducing support for WFFC is trivial, this may > make sense to exist in older versions too. -- This message was sent by Atlassian Jira (v8.20.10#820010) - To unsubscribe, e-mail: issues-unsubscr...@yunikorn.apache.org For additional commands, e-mail: issues-h...@yunikorn.apache.org
[jira] [Commented] (YUNIKORN-1253) PVCs won't get past WaitForFirstConsumer with Apache Yunikorn
[ https://issues.apache.org/jira/browse/YUNIKORN-1253?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17568682#comment-17568682 ] Chen Yu Teng commented on YUNIKORN-1253: To handle the ephemeral case, i think we need to follow the idea in [https://github.com/kubernetes/kubernetes/blob/master/pkg/scheduler/framework/plugins/volumebinding/volume_binding.go#L121-L163] I will modify the https://github.com/apache/yunikorn-k8shim/blob/master/pkg/cache/task.go#L508-L529. volumeClaimTemplate https://github.com/kubernetes/component-helpers/blob/master/storage/ephemeral/ephemeral.go > PVCs won't get past WaitForFirstConsumer with Apache Yunikorn > - > > Key: YUNIKORN-1253 > URL: https://issues.apache.org/jira/browse/YUNIKORN-1253 > Project: Apache YuniKorn > Issue Type: Bug > Components: shim - kubernetes >Reporter: Alex Kalenyuk >Assignee: Chen Yu Teng >Priority: Major > Attachments: case01_schedulingWhenPVisNotCreated.log, > case01_yunikorn_scheduling_decision.log, pv.yaml, storageclass.yaml > > > It seems that with Apache Yunikorn, WaitForFirstConsumer volume binding > storage classes are not supported (not sure if this is intended or not). > This makes it problematic to use storage that is not globally accessible from > all nodes: > [https://kubernetes.io/docs/concepts/storage/storage-classes/#volume-binding-mode] > To reproduce a simple failure to use a WaitForFirstConsumer binding-backed > PVC: > ``` > kind: Pod > apiVersion: v1 > metadata: > name: test > namespace: default > labels: > app: sleep > applicationId: "sleep0001-node-selector-test" > queue: "root.sandbox" > spec: > schedulerName: yunikorn > nodeSelector: > storage/ssd: 'true' > containers: > - name: test > resources: > limits: > cpu: 1 > memory: 1G > requests: > cpu: 1 > memory: 1G > image: busybox > command: > - sleep > - '100' > volumeMounts: > - name: scratch-volume > mountPath: /data > volumes: > - name: scratch-volume > ephemeral: > volumeClaimTemplate: > spec: > accessModes: > - ReadWriteOnce > resources: > requests: > storage: 1Gi > storageClassName: hostpath-provisioner > volumeMode: Filesystem > ``` > Storage used: > https://github.com/kubevirt/hostpath-provisioner-operator > A similar issue was spotted in: > [https://github.com/kubernetes/kubernetes/issues/86262] > And this PR seems to introduce the VolumeBinding filter but comments it out: > [https://github.com/apache/yunikorn-k8shim/pull/313] > > I might be off with the "Bug" type here so feel free to correct me; > My thinking was that if introducing support for WFFC is trivial, this may > make sense to exist in older versions too. -- This message was sent by Atlassian Jira (v8.20.10#820010) - To unsubscribe, e-mail: issues-unsubscr...@yunikorn.apache.org For additional commands, e-mail: issues-h...@yunikorn.apache.org
[jira] [Commented] (YUNIKORN-1253) PVCs won't get past WaitForFirstConsumer with Apache Yunikorn
[ https://issues.apache.org/jira/browse/YUNIKORN-1253?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17565930#comment-17565930 ] Weiwei Yang commented on YUNIKORN-1253: --- Hi [~Yukali] thanks a lot. Pls keep us posted on updates. > PVCs won't get past WaitForFirstConsumer with Apache Yunikorn > - > > Key: YUNIKORN-1253 > URL: https://issues.apache.org/jira/browse/YUNIKORN-1253 > Project: Apache YuniKorn > Issue Type: Bug > Components: shim - kubernetes >Reporter: Alex Kalenyuk >Assignee: Chen Yu Teng >Priority: Major > Attachments: pv.yaml, storageclass.yaml > > > It seems that with Apache Yunikorn, WaitForFirstConsumer volume binding > storage classes are not supported (not sure if this is intended or not). > This makes it problematic to use storage that is not globally accessible from > all nodes: > [https://kubernetes.io/docs/concepts/storage/storage-classes/#volume-binding-mode] > To reproduce a simple failure to use a WaitForFirstConsumer binding-backed > PVC: > ``` > kind: Pod > apiVersion: v1 > metadata: > name: test > namespace: default > labels: > app: sleep > applicationId: "sleep0001-node-selector-test" > queue: "root.sandbox" > spec: > schedulerName: yunikorn > nodeSelector: > storage/ssd: 'true' > containers: > - name: test > resources: > limits: > cpu: 1 > memory: 1G > requests: > cpu: 1 > memory: 1G > image: busybox > command: > - sleep > - '100' > volumeMounts: > - name: scratch-volume > mountPath: /data > volumes: > - name: scratch-volume > ephemeral: > volumeClaimTemplate: > spec: > accessModes: > - ReadWriteOnce > resources: > requests: > storage: 1Gi > storageClassName: hostpath-provisioner > volumeMode: Filesystem > ``` > Storage used: > https://github.com/kubevirt/hostpath-provisioner-operator > A similar issue was spotted in: > [https://github.com/kubernetes/kubernetes/issues/86262] > And this PR seems to introduce the VolumeBinding filter but comments it out: > [https://github.com/apache/yunikorn-k8shim/pull/313] > > I might be off with the "Bug" type here so feel free to correct me; > My thinking was that if introducing support for WFFC is trivial, this may > make sense to exist in older versions too. -- This message was sent by Atlassian Jira (v8.20.10#820010) - To unsubscribe, e-mail: issues-unsubscr...@yunikorn.apache.org For additional commands, e-mail: issues-h...@yunikorn.apache.org
[jira] [Commented] (YUNIKORN-1253) PVCs won't get past WaitForFirstConsumer with Apache Yunikorn
[ https://issues.apache.org/jira/browse/YUNIKORN-1253?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17565284#comment-17565284 ] Chen Yu Teng commented on YUNIKORN-1253: Update: I tried WaitForFirstConsumer case with default scheduler and yunikorn. This case with default scheduler is work, but yunikorn is not. > PVCs won't get past WaitForFirstConsumer with Apache Yunikorn > - > > Key: YUNIKORN-1253 > URL: https://issues.apache.org/jira/browse/YUNIKORN-1253 > Project: Apache YuniKorn > Issue Type: Bug > Components: shim - kubernetes >Reporter: Alex Kalenyuk >Assignee: Chen Yu Teng >Priority: Major > Attachments: pv.yaml, storageclass.yaml > > > It seems that with Apache Yunikorn, WaitForFirstConsumer volume binding > storage classes are not supported (not sure if this is intended or not). > This makes it problematic to use storage that is not globally accessible from > all nodes: > [https://kubernetes.io/docs/concepts/storage/storage-classes/#volume-binding-mode] > To reproduce a simple failure to use a WaitForFirstConsumer binding-backed > PVC: > ``` > kind: Pod > apiVersion: v1 > metadata: > name: test > namespace: default > labels: > app: sleep > applicationId: "sleep0001-node-selector-test" > queue: "root.sandbox" > spec: > schedulerName: yunikorn > nodeSelector: > storage/ssd: 'true' > containers: > - name: test > resources: > limits: > cpu: 1 > memory: 1G > requests: > cpu: 1 > memory: 1G > image: busybox > command: > - sleep > - '100' > volumeMounts: > - name: scratch-volume > mountPath: /data > volumes: > - name: scratch-volume > ephemeral: > volumeClaimTemplate: > spec: > accessModes: > - ReadWriteOnce > resources: > requests: > storage: 1Gi > storageClassName: hostpath-provisioner > volumeMode: Filesystem > ``` > Storage used: > https://github.com/kubevirt/hostpath-provisioner-operator > A similar issue was spotted in: > [https://github.com/kubernetes/kubernetes/issues/86262] > And this PR seems to introduce the VolumeBinding filter but comments it out: > [https://github.com/apache/yunikorn-k8shim/pull/313] > > I might be off with the "Bug" type here so feel free to correct me; > My thinking was that if introducing support for WFFC is trivial, this may > make sense to exist in older versions too. -- This message was sent by Atlassian Jira (v8.20.10#820010) - To unsubscribe, e-mail: issues-unsubscr...@yunikorn.apache.org For additional commands, e-mail: issues-h...@yunikorn.apache.org
[jira] [Commented] (YUNIKORN-1253) PVCs won't get past WaitForFirstConsumer with Apache Yunikorn
[ https://issues.apache.org/jira/browse/YUNIKORN-1253?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17562955#comment-17562955 ] Weiwei Yang commented on YUNIKORN-1253: --- Discussed this issue with [~Yukali], please help to investigate, thanks! > PVCs won't get past WaitForFirstConsumer with Apache Yunikorn > - > > Key: YUNIKORN-1253 > URL: https://issues.apache.org/jira/browse/YUNIKORN-1253 > Project: Apache YuniKorn > Issue Type: Bug > Components: shim - kubernetes >Reporter: Alex Kalenyuk >Assignee: Chen Yu Teng >Priority: Major > > It seems that with Apache Yunikorn, WaitForFirstConsumer volume binding > storage classes are not supported (not sure if this is intended or not). > This makes it problematic to use storage that is not globally accessible from > all nodes: > [https://kubernetes.io/docs/concepts/storage/storage-classes/#volume-binding-mode] > To reproduce a simple failure to use a WaitForFirstConsumer binding-backed > PVC: > ``` > kind: Pod > apiVersion: v1 > metadata: > name: test > namespace: default > labels: > app: sleep > applicationId: "sleep0001-node-selector-test" > queue: "root.sandbox" > spec: > schedulerName: yunikorn > nodeSelector: > storage/ssd: 'true' > containers: > - name: test > resources: > limits: > cpu: 1 > memory: 1G > requests: > cpu: 1 > memory: 1G > image: busybox > command: > - sleep > - '100' > volumeMounts: > - name: scratch-volume > mountPath: /data > volumes: > - name: scratch-volume > ephemeral: > volumeClaimTemplate: > spec: > accessModes: > - ReadWriteOnce > resources: > requests: > storage: 1Gi > storageClassName: hostpath-provisioner > volumeMode: Filesystem > ``` > Storage used: > https://github.com/kubevirt/hostpath-provisioner-operator > A similar issue was spotted in: > [https://github.com/kubernetes/kubernetes/issues/86262] > And this PR seems to introduce the VolumeBinding filter but comments it out: > [https://github.com/apache/yunikorn-k8shim/pull/313] > > I might be off with the "Bug" type here so feel free to correct me; > My thinking was that if introducing support for WFFC is trivial, this may > make sense to exist in older versions too. -- This message was sent by Atlassian Jira (v8.20.10#820010) - To unsubscribe, e-mail: issues-unsubscr...@yunikorn.apache.org For additional commands, e-mail: issues-h...@yunikorn.apache.org
[jira] [Commented] (YUNIKORN-1253) PVCs won't get past WaitForFirstConsumer with Apache Yunikorn
[ https://issues.apache.org/jira/browse/YUNIKORN-1253?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17562783#comment-17562783 ] Weiwei Yang commented on YUNIKORN-1253: --- Thanks [~akalenyu], the description helps a lot. As far as I know, we haven't tested the "WaitForFirstConsumer" mode before, this is probably not supported today. If somebody has some bandwidth to take a look at this, that will be great. Cc [~wilfreds], [~ccondit], [~yuteng], [~tingyao] > PVCs won't get past WaitForFirstConsumer with Apache Yunikorn > - > > Key: YUNIKORN-1253 > URL: https://issues.apache.org/jira/browse/YUNIKORN-1253 > Project: Apache YuniKorn > Issue Type: Bug > Components: shim - kubernetes >Reporter: Alex Kalenyuk >Priority: Major > > It seems that with Apache Yunikorn, WaitForFirstConsumer volume binding > storage classes are not supported (not sure if this is intended or not). > This makes it problematic to use storage that is not globally accessible from > all nodes: > [https://kubernetes.io/docs/concepts/storage/storage-classes/#volume-binding-mode] > To reproduce a simple failure to use a WaitForFirstConsumer binding-backed > PVC: > ``` > kind: Pod > apiVersion: v1 > metadata: > name: test > namespace: default > labels: > app: sleep > applicationId: "sleep0001-node-selector-test" > queue: "root.sandbox" > spec: > schedulerName: yunikorn > nodeSelector: > storage/ssd: 'true' > containers: > - name: test > resources: > limits: > cpu: 1 > memory: 1G > requests: > cpu: 1 > memory: 1G > image: busybox > command: > - sleep > - '100' > volumeMounts: > - name: scratch-volume > mountPath: /data > volumes: > - name: scratch-volume > ephemeral: > volumeClaimTemplate: > spec: > accessModes: > - ReadWriteOnce > resources: > requests: > storage: 1Gi > storageClassName: hostpath-provisioner > volumeMode: Filesystem > ``` > Storage used: > https://github.com/kubevirt/hostpath-provisioner-operator > A similar issue was spotted in: > [https://github.com/kubernetes/kubernetes/issues/86262] > And this PR seems to introduce the VolumeBinding filter but comments it out: > [https://github.com/apache/yunikorn-k8shim/pull/313] > > I might be off with the "Bug" type here so feel free to correct me; > My thinking was that if introducing support for WFFC is trivial, this may > make sense to exist in older versions too. -- This message was sent by Atlassian Jira (v8.20.10#820010) - To unsubscribe, e-mail: issues-unsubscr...@yunikorn.apache.org For additional commands, e-mail: issues-h...@yunikorn.apache.org