Welcome to the Linux Foundation Forum!

Does it work the exercise lab from lab 11.2 to deploy Linkerd?

StefanDiaconu
StefanDiaconu Posts: 6
in LFS258 Class Forum

Hello,

I followed several times the exercise instructions, starting with configuration inside a HA cluster, then converting the HA cluster to a single cp and multiple workers(to have more ressources) and everything I get is a linkerd namespace with several pods on error because of ressources issues or not completely defined configuration(other missing pods that should provide identity, but which are missing because of OOM).

I understand there need to be deployed more ressources, but what are the minimum limits to make the service mesh linkerd available?

Do you have any tips regarding this lab?

Answers

  • chrispokorni
    chrispokorni Posts: 2,629

    Hi @StefanDiaconu,

    The OOM errors are related to insufficient RAM. What are the sizes of your VMs (CPU, RAM, Disk)?

    Regards,
    -Chris

  • StefanDiaconu
    StefanDiaconu Posts: 6

    Hello Chris,

    Thank you for your answer.
    I have 8 VMs with 8GBRAM each.
    My master node is on the 9th VM.(I eliminated the OOM issue)
    Today I stayed more on the subject and I saw there might be some DNS issues also.
    All the VMs are communicating, being referenced by a central DNS server. On my Kubernetes cluster, when I try to do the nslooup from linkerd namespace from a busybox pod, it doesn't reach the kubernetes.default.svc.cluster.local.
    I have to clarify myself how will be used the linkerd, as there are some nginx controller on my cluster, some dns, and it is possible that keeping the configuration by default when deploying linkerd might cause the issue ... I am still struggling with this topic

Categories

Upcoming Training