Cluster API is reporting that only 19 nodes out of 24 have registered.

[Vitaly Glazkov]

Hi,

We have a project named "en-development" that contains a cluster "essence-stage-small-central-cluster" with the following configuration: 24 nodes, n1-standard-16. After I resized cluster to 3 nodes and then back to 24, I've got working only 19 out of 24 nodes. I've cleared up cluster (delete ingresses, services, pods), tried again the experiment - the problem was still persisting. After that I've decided to upgrade all nodes including master to the next version of kubernetes: from 1.4.5 to 1.4.6. The problem still persists and in the google console I see the following message:

Cluster updated to 1.4.6 but failed health check: All cluster resources were brought up, but the cluster API is reporting that only 19 nodes out of 24 have registered. Cluster may be unhealthy.

[George (Google Cloud Support)]

Hello Vitaly,

Is it possible that you might be hitting one of the Quotas in your project (CPU, Disk...), did you have the chance to look at it?

More information about this matter can be found in this help center article.

I hope this helps.

Sincerely,

George

[Jingpeng Wu]

I got a similar problem: 

All cluster resources were brought up, but the cluster API is reporting that: only 2 nodes out of 4 have registered; cluster may be unhealthy.

my quota should be large enough.

On Tuesday, November 29, 2016 at 9:06:22 AM UTC-5, Vitaly Glazkov wrote:

[Rahi (Google Cloud Support)]

Hi Jingpeng,

The root cause of the error could be many things such as autoscaler, health check, terraform (if used) configuration and many more. An in-depth investigation requires to see why the expected number of nodes is not spinning up. I would recommend filing a defect report at this public issue tracker link [1]. This URL will take you to the public issue tracker link where the component is set to private. Please provide your project ID, cluster name, affected node names, details of the issue and possible logs/screenshot with a timestamp (Date-Time-Timezone). As the component of this thread is set to private thus your project ID and relevant information won't be visible to the public. One of the support team members from the Google Cloud Platform team will deep dive into the issue.

[1] https://issuetracker.google.com/issues/new?component=491576&template=1161385

[cabe...@gmail.com]

I'm seeing a similar issue 

All cluster resources were brought up, but the cluster API is reporting that: only 0 nodes out of 6 have registered; cluster may be unhealthy.

I'm trying to follow the steps here to create a shared VPC cluster in an existing network/subnet:

https://cloud.google.com/kubernetes-engine/docs/how-to/cluster-shared-vpc

I was able to create the cluster in a new network. But I'm having trouble identifying what in the existing network (firewall rules, routes, static IPs) could cause this.

I would appreciate any guidance.

[Sajjad Ur Rehman]

I am having same issue while getting Jenkins X with terraform on GKE, was there any solution/work around for this? Nodes are up and running but they do not get back to cluster with health status.

 All cluster resources were brought up, but: only 0 nodes out of 1 have registered

iProov Ltd. Company Registration No.07866563 incorporated in England and Wales.  Registered Office: 14 Bank Chambers, 25 Jermyn Street, London, SW1Y 6HR

[Alexandre Duval-Cid]

Hi,

There can be multiple causes to this, some outlined in the thread, most are errors in configuration. I would encourage you to open a tech case [1] though your cloud console so we can look deeper into the specifics of what is causing this issue on your end. If you believe this issue is platform wide you can also open an issue tracker [2] so we can validate your reproduction and fix the issue.

[1] https://cloud.google.com/support/docs/

[2] https://issuetracker.google.com/issues/new?component=491576&template=1161385

[Sajjad Ur Rehman]

Hi, 

Thanks in my case it was default IAM account disabled in GCP which was causing issues..