Tag: Observability

Linux Performance & Observability: load, latency tails, perf top, eBPF

Working reference for the 10 most common Linux performance problems. High load with low CPU (D-state iowait), p99 latency tails (averages lie), high context switch rate, disk I/O saturation (%util / await), network at line rate, memory pressure without OOM (vmstat si/so), slow boot (systemd-analyze blame), hung app (strace / wchan / stack), kernel CPU vulnerability mitigations, cloud noisy neighbor (%steal). Tools: vmstat / iostat / sar / perf top / bcc-eBPF / biolatency / execsnoop. Cross-linked to monitoring command reference.