RKE cluster - failed pods

Rancher 2.x
1

Hello everyone! I have a strange issue even though the rke up says it was launched successfully, some pods are crashed.

image
image946×344 36.2 KB

I have 3 node cluster, docker 19, and rke 1.0.14. Please advise!

2

The pod logs can be retrieved using kubectl logs and will most of the time show why it can’t run successfully.

3

@superseb Thanks for trying to assist me! I’ve got farther in the troubleshooting and see this error now

image
image1103×517 58.2 KB
What could it mean… tried disabling selinux, no luck.

4

OK, I can take a look at it. Please supply some more information about your setup:

  • cluster.yml used
  • docker info from one of the nodes (or all if not identical)

If it would be SELinux, there should be logging in /var/log/audit/audit.log

5

too bad now the cluster won’t start

image
image1697×127 19.6 KB
:frowning_face:

6

I don’t know what you did, I requested more info to diagnose the issue. That file needs to be manually removed on the node(s) before re-using the node(s) when provisioning the cluster.

7

Give me one sec trying to clean up the nodes to run the cluster with error again!

8

YEY! somehow it ran, thanks for assisting, I was hopeless :smiley:

image
image993×352 36 KB

9

image is that best practices ? i’ve read it somewhere in the docs.

10

For Rancher HA, 3 nodes with each all 3 roles is the recommended configuration. See Rancher Docs: Setting up a High-availability RKE Kubernetes Cluster