页缓存 / 脏页回写 / 内存回收诊断 Skill

To diagnose Linux kernel dirty page writeback and memory reclamation bottlenecks, this skill correlates real-time system metrics from /proc/vmstat and /proc/meminfo with kernel tracepoint events to perform automated root cause analysis.

Intermittent latency spikes and D-state process hangs often stem from I/O backpressure or memory pressure, which this skill identifies by categorizing issues into failure modes like dirty page writeback anomalies or kswapd CPU spikes.

Correlating real-time system metrics with kernel tracepoints identifies I/O backpressure root causes by automatically matching symptoms like global I/O throttling to specific configurations such as an improperly configured vfs_cache_pressure or slow disk devices.

To troubleshoot memory pressure and page cache thrashing, you need command-line dependencies including bash, awk, grep, and bc to execute the scripts and process system metrics for comprehensive analysis.

Yes, you can use this automated diagnosis skill for production servers experiencing kswapd CPU spikes, as it automatically categorizes such anomalies and identifies whether they originate from memory reclamation or writeback faults.