)]}'
{
  "log": [
    {
      "commit": "aef87fc607281c13b75b971c29595b2b101807a8",
      "tree": "ebbe195282075b6b3b821f48093319491e7964c8",
      "parents": [
        "a01c390bff097c4edd542e2cb4994ac16ac0a39e"
      ],
      "author": {
        "name": "Matthias Andreas Benkard",
        "email": "code@mail.matthias.benkard.de",
        "time": "Thu Nov 05 11:55:32 2026 +0100"
      },
      "committer": {
        "name": "Matthias Andreas Benkard",
        "email": "code@mail.matthias.benkard.de",
        "time": "Thu Nov 05 11:55:32 2026 +0100"
      },
      "message": "kata-deploy: Keep Kata VMs inside the pod cgroup.\n\nKata\u0027s Go runtime only honours the systemd cgroup paths that kubelet\npasses when sandbox_cgroup_only is true.  Without it, every VM ran in a\nroot-level cgroup named after the raw path, outside kubepods.slice, and\nleaked it after the pod went away.\n\nWith the whole VM charged to the pod, raise the qemu RuntimeClass\noverhead to cover the VM\u0027s 2 GiB base memory.\n\nCo-Authored-By: Claude Opus 5.5 \u003cnoreply@anthropic.com\u003e\nChange-Id: Ia305e4c5df1f67126b9c57c7a638bb096bec82e3\n"
    }
  ]
}