New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Pod volume mounting failing even after PV is bound and attached to pod #49926
Comments
/sig storage |
@amolsh you can get the pv? you storage is use storage-class, if you create pvc ,it's will auto create pv. you can see the pv exist or not |
@amolsh did you created a storage-class was named default? cany you show the storage-class? l think the reason is miss storage-class of named default. do you have any other yaml? |
@huangjiasingle This is my storage class
|
@amolsh do you have the master log we can take a look? You can email us if your prefer. The kubelet log shows that the volume is not attached to the node yet. |
master kube-controll-manager is not generating any logs related to above statefulset(it did not generate any logs when I created above statefulset). Also I couldn't find anything in api-server logs also. One more thing I forgot to mention my cluster is on CoreOs machines |
@huangjiasingle Failed to attach volume "pvc-fff84c66-7b35-11e7-b125-02f3f42ec6aa" on node "ip-100-x-x-x.us-west-2.compute.internal" with: Error attaching EBS volume "vol-00cbb836374d1b37b" to instance "i-03d4ba6b17ab9cf5f": UnauthorizedOperation: You are not authorized to perform this operation. Encoded authorization failure message: 4nedpChQKhsxXs...... |
I updated IAM policy and added
|
Now volume attachement is successfull, but still getting error master kube-controller logs:
|
@amolsh are you still seeing this issue? |
Yes, it did not resolve, so I stopped trying.
On Nov 9, 2017 10:26 AM, "Michelle Au" <notifications@github.com> wrote:
@amolsh <https://github.com/amolsh> are you still seeing this issue?
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#49926 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AKdrDSbpWZ_E6iP4X1yw0B3d4U9XKu78ks5s0oX3gaJpZM4OpRyq>
.
|
This seems like a possible issue related to EBS volumes. Can @kubernetes/sig-aws-misc help out? |
I'm also having this issue: "AttachVolume.Attach failed for volume "pvc-5aa1db99-d04b-11e7-96e1-0a328f684a08" : Error attaching EBS volume "vol-026ef055a9d715c14" to instance "i-0e6f2c5d8edd415e4": IncorrectState: vol-026ef055a9d715c14 is not 'available'. status code: 400" Using a single master, 3 node cluster in AWS (deployed via KOPS)... trying to deploy influxdb via helm. It worked the first time I deployed it but after deleting the deployment (to change collectd config) I can't get it to work... it doesn't seem to matter which node, none of them work. Grafana is deployed on the same cluster just fine (but have only deployed it once).
|
The @mattcamp are you seeing "Not available" error more than once in controller logs? Looking at kubectl output I see following message:
Which indicates attach succeeded and volume was attached to the node. I think something else happened when container was started:
You may want to look into it. |
Ran into "IncorrectState: ... is not 'available'. status code: 400" when trying a stateful set in 1.8 that works great in 1.7. It does retry but keeps failing and the pod stays in a CreateContainerConfigError state (something that's new to me). |
Hi, i got the same (?) error when trying 'helm install'
|
@feffi the mount error you got is pretty self-explanatory:
This is different from the OP's error, which is related to EBS volumes. |
It's worth noting that a message such |
Same issue here! |
Same problem when deploying a Stateful Set on kops on AWS.
versions:
|
Same error here... I'm create an storage class, pv, pvc and a statefulset in one go and I get the same error. The ebs volume it's actually attached to the node so I'm not sure what's going on. |
I have tried to remove the pod and it still fails, once recreated. |
I'm using Kubernetes 1.8.8, deployed using kops 1.8.1, with RBAC enabled.
|
Some additional information. If I remove everything using
|
I am also having issues with nvme instances, running Kubernetes 1.10.x in this case, I tried using m5.large with pvc.
this is the storage classes:
The EBS is created and attached to the instance. but Kubelet fails to mount the disk into pod
Cluster provisioned using Kops 1.10 kubectl version
When checking on m5.large the mounts on the node there is not disk mount on the nvme drive. when replacing to m4.large the mounts has: The node image in Kops is: And on the same note, different use case, when launching a new cluster using Kops and masters are nvme instances like m5.large, the host startup fails to mount the etcd volume and hangs with protokube:1.10.0 in a loop for
After installing nvme-cli I can see that the volume exists
But this /dev/disk/by-id/nvme-Amazon_Elastic_Block_Store_vol0fdeaa59d34bd2ab1 not exists. the disk mapping /dev/disk/by-id doesn't exists only /dev/disk/by-uuid/ So basically I can not use any nvme base instance for masters or nodes that has a EBS pvc |
Cloud: AWS Having a similar problem with storage myself. When draining a node that has an EBS volume which is attached to a Pod and/or Deployment, the storage doesn't move. It releases from original node but never makes it to the new/next node. Like others, switching to |
Same problem here, AWS, k8s version 1.13.0 |
Same problem as well, AWS, eks. k8s version 1.11.0 I have a feeling that it might be due to the maximum number of ebs attachments to an ec2 node. |
I've been seeing a similar issue on EKS with k8s version 1.11. Our support agent suggested the following: It's very likely that the Kubernetes scheduler was choosing worker nodes in Availability Zones (AZ) with no volumes available. This can happen when the node selected for placement of the pod by the scheduler is not in the availability zone in which the Persistent volume(s) claim are available i.e. EBS volume exists. For example, when there isn't sufficient CPU and/or memory resources available on Nodes in which the persistent volume exists. That would lead to the scheduler choosing a Node in another zone and failing to schedule pod with this error. This was a known issue in Kubernetes[1][2] and has been fixed by having "VolumeScheduling"[2] feature enabled in scheduler. Another workaround could include creating the volumes manually and updating the PVCs but both options would turn into a less available/dynamic cluster. References: |
/assign @leakingtapan |
Unfortunately this has nothing to do with AWS Availability Zones or VolumeScheduling. AZ related problems are kinda hot nowadays, so people like to mix up that problem with this one, but a quick look at the Availability Zones makes clear that there is no connection. Today's testing results:
|
After some debugging I found my problem is this, all the symptoms are matching: |
@gabordk I got the same issue. Any solution or progress on this? |
same issue here. on microk8s, enabled default storage addon, then install helm-consul got this issue, all pv and pvc are bound. but consul-server pods still got "pod has unbound immediate PersistentVolumeClaims", recreate these pods didn't help |
I believe I'm seeing the same issue on aws w/ eks 1.3.8 (1.3 "eks.2")
|
@jhoblitt Did you tried with EKS 1.2.x? I faced some issues with statefulsets with PVC on EKS 1.3.x but everything ran just fine on EKS 1.2.x. |
@jhoblitt I was facing the same issue until 10 min ago, I realised there is a problem/bug with the kubernetes-plugin I was using. I solved upgrading to 1.18.3 the kubernetes-plugin for Jenkins. |
@ishantanu I don't believe I was seeing this problem with 1.2 but it's been awhile since I've tested with that version. @mogaal This problem is present outside of pods managed by jenkins. |
@mogaal |
Issues go stale after 90d of inactivity. If this issue is safe to close now please do so with Send feedback to sig-testing, kubernetes/test-infra and/or fejta. |
Stale issues rot after 30d of inactivity. If this issue is safe to close now please do so with Send feedback to sig-testing, kubernetes/test-infra and/or fejta. |
Rotten issues close after 30d of inactivity. Send feedback to sig-testing, kubernetes/test-infra and/or fejta. |
@fejta-bot: Closing this issue. In response to this:
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. |
/reopen |
@sudip-moengage: You can't reopen an issue/PR unless you authored it or you are a collaborator. In response to this:
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. |
kubectl version:
Client Version: version.Info{Major:"1", Minor:"5", GitVersion:"v1.5.1", GitCommit:"82450d03cb057bab0950214ef122b67c83fb11df", GitTreeState:"clean", BuildDate:"2016-12-14T00:57:05Z", GoVersion:"go1.7.4", Compiler:"gc", Platform:"darwin/amd64"}
Server Version: version.Info{Major:"1", Minor:"5", GitVersion:"v1.5.4+coreos.0", GitCommit:"97c11b097b1a2b194f1eddca8ce5468fcc83331c", GitTreeState:"clean", BuildDate:"2017-03-08T23:54:21Z", GoVersion:"go1.7.4", Compiler:"gc", Platform:"linux/amd64"}
yml file:
List of bound PVs:
ERROR:
relevant kubelet logs:
The text was updated successfully, but these errors were encountered: