主题： sysstat

• Per-process CPU and I/O with pidstat and /proc/<pid>/io: finding what causes I/O wait

  pidstat reports per-task CPU, memory and disk I/O without grepping a name out of top; reading /proc/<pid>/io directly separates logical I/O (rchar/wchar, including cache) from the physical read_bytes/write_bytes that actually reach the device.

  methodology · en · reviewed · 知识截至 2026-09-24

• Disk I/O latency with iostat -x: r_await, w_await, aqu-sz, and why %util misleads on SSD and RAID

  iostat -x adds per-device latency and queue fields to the plain throughput view. r_await and w_await separate read from write latency, aqu-sz shows how many requests are outstanding (named avgqu-sz in older sysstat releases), and the sysstat manual itself warns that %util stops meaning saturation on devices that serve requests in parallel.

  article · en · reviewed · 知识截至 2026-09-24

• 用 60 秒初看缓慢的 Linux 服务器：命令顺序与每行回答的问题

  按固定顺序运行十条标准命令，从 uptime 到 top，其中若干来自可能需要先安装的 sysstat 包；每条分别回答一个有关 CPU、内存、磁盘或网络的问题。在打开任何日志文件或重启任何东西之前先执行这一轮。

  methodology · zh · reviewed · 知识截至 2026-09-24

• Historical performance data with sysstat: enabling collection and reading past days with sar -f

  sysstat's sadc collector, run periodically by sa1/sa2 or a systemd timer, writes a day's counters to the standard system activity daily data file; sar -f replays a chosen day's file for any past interval instead of only the live counters.

  methodology · en · reviewed · 知识截至 2026-09-24

机器可读： JSON