Welcome to the Linux Foundation Forum!

Lab 3-4: Cluster IP not accessible from master node


Hi there, I have run lab 3.4 without any issue but there's one point which I can't understand. I agree that the endpoint IP is not reachable from the master node if the nginx pod is running on the worker node. But how about the cluster IP ? Irrespective of whether I am on the master node or on the worker node, I should be able to get a response when doing curl :80 ...
The above is true in the case when I have only one endpoint running and when I have 3 endpoints running. In both cases, the curl :80 times out. What is that so ?
Thanks for your explanation.


  • teggy
    teggy Posts: 6

    I mean curl cluster-ip:80 ... (angled braces and its contents got removed...)

  • chrispokorni
    chrispokorni Posts: 2,190

    Hi @teggy,

    The issue you encountered is frequently reported in the forum. It is typical for an environment where the cluster nodes are not able to fully talk to each other. Take a look at the lab setup videos in the course resources, as they address known possible setup issues. In summary, these issues are caused by firewalls blocking traffic to specific ports or port ranges of the cluster nodes.


  • danielclough

    I don't see any videos in the resources section, only one PDF.

    Did you figure it out?

  • fcioanca
    fcioanca Posts: 1,935
    edited September 2020

    @danielclough The lab setup videos are in the introductory chapter, and the title of the pages starts with IMPORTANT...

  • danielclough

    If anyone else has this problem, and is trying to setup a bare metal cluster, try configuring calico from default to IP_AUTODETECTION_METHOD=can-reach=<DESTINATION>

    In my case kubectl -n kube-system set env daemonset/calico-node IP_AUTODETECTION_METHOD=can-reach=

    docs here: https://docs.projectcalico.org/networking/ip-autodetection

    Thanks. I thought by "resources" he was referring to the tab named "resources" with files to be downloaded.


Upcoming Training