Kubernetes
6 posts
·7 min read
Running LLM inference on K8s: auto-injecting RDMA permissions and GPU-NIC topology affinity
KubernetesRDMAInfiniBand
·10 min read
MySQL freezes for 2s every few hours: runc freeze timing out on a D-state process
Kubernetesruncsystemd
·8 min read
PLEG NotReady on Kubelet: from one-off fixes to a systemic cure
KubernetescontainerdKubelet
·9 min read
containerd init container stuck in Running: an rootfs-umount timeout corrupts the state machine
KubernetescontainerdCRI
·13 min read
PLEG NotReady on short-task workloads: an overlooked lock contention inside containerd
KubernetescontainerdPLEG
·12 min read
/dev/urandom intermittently returns EPERM: device-permission jitter under runc + systemd cgroup driver
Kubernetescontainerdrunc