Topic: 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.
-
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.
-
A 60-second first look at a slow Linux server: the command order and what each line answers
A fixed sequence of ten standard commands — uptime through top, several from the sysstat package that may need installing first — each answering one question about CPU, memory, disk and network, run before opening a single log file or restarting anything.
-
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.
Machine-readable: JSON