Welcome to the Linux Foundation Forum!

Lab 2.1 Ubuntu k8sMaster.sh script completed with 'localhost:8080 connection refused' ?

Hi there, am just setting up course on AWS Ubuntu 16.04 node, following instructions have got to running the k8sMaster.sh which looked ok to completion (in master.out) by successfully downloading calico.yaml & rbac-kdd.yaml, but then script output loads of "connection localhost:8080 refused" messages. Now all kubectl commands just universally output "The connection to the server localhost:8080 was refused - did you specify the right host or port?" ?
Dont think i missed as step so there must be some local config to do on spun-up Ubunuti 16.04 network config ??
Do i need to run script as root ?
Also the AWS Free Tier Ubuntu 16.04 offers only 1 vCPU rather than recommended 2 vCPU (see errors and downgrades) - is two a mandatory requirement (so need to upspec image choice) ?
All help appreciated !!


  • k8ckad
    k8ckad Posts: 3
    edited September 2019

    Ahh just read the issue raised and answer in crixo's issue before mine here (error setting up august release in aws) by crixo - i see the issue is around the 1vCPU impl as discussed... many thanks!

  • I am facing the same issue . Can you please provide the link where you found the answer for this issue?

  • serewicz
    serewicz Posts: 1,000


    Are you also using a single vCPU node? You will find it is not enough to get all the pods running which need to run. I think the start of lab 2 mentions using 2vCPU/7.5G nodes.


  • seetha33tmr
    seetha33tmr Posts: 17
    edited December 2019

    @serewicz Yes I am getting the same issue as well. While 2 vCPU and 8G of memory allows for quick labs you could use much smaller VMs. Other Linux distributions should work in a very similar manner, but have not been tested. This is what mentioned in the Lab 2.1 doc. And since the ready_for.sh gave me PASS on all checks I assumed t2.micro instance from AWS would work. I am not sure AWS comes with 2CPU VM on free trail. Would you recommend using Google cloud instead?

  • serewicz
    serewicz Posts: 1,000

    If you can get use their free credits Google may be a good option. Over time the master pods have become more resource intensive. I will probably have to edit the intro to explain this.


  • Alright. Appreciate the timely response. Thank you


Upcoming Training