Jump to content

Kubernetes SIG/Meetings/2024-09-24

From mediawiki.org

Agenda:

  • Introductions for new members (if any):
  • SIG administrivia:
    • mediawiki.org SIG page updated with links to read only notes and Gmeet. Still missing a proper way to publish the gcal event
  • Misc
  • Topics:
    • Context at https://phabricator.wikimedia.org/T373526, 2 questions:
      • Could we use the SIG's email as Maintainer field for “core” images?
      • Could we assign the duty to the SIG to update/assign the right maintainer fields periodically
      • [CD] How about if we have a team as maintainer and perhaps 1-2 individuals recorded as "expert of last resort"?
      • [JM] Or the sig as the expert of last resort for core stuff, maybe without promising maintenance?
      • Probably okay to have the SIG as maintainer for some images, but not for everything
      • How would we design a round-robin core image update duty?
    • Race condition in iptables rules during puppet runs on k8s nodes
      • Situation/Number of ferm reloads should improve with the outstanding patch
      • Moving to nftables (which kube-proxy does not yet support) will probably fix this
    • Reverse DNS for k8s pods IPs
      • Running an additional CoreDNS daemonset on k8s apiservers and using their nodeport is probably the easiest option to integrate - if DNS servers can pull the apiservers APIs automatically via puppet
    • What would be the challenges involved in a rolling upgrade of the dse-k8s cluster, as opposed to a full reimage and reinitialisation?
Retrieved from "https://www.mediawiki.org/w/index.php?title=Kubernetes_SIG/Meetings/2024-09-24&oldid=6772152"