About this skill

Problem scope

When TencentOS Server shows high CPU without a clear hotspot, normal ping but slow application responses, unclear OOM root causes, insufficient disk space, or low network throughput, troubleshooting often jumps between logs, kernel state, network stacks, and performance tools. This skill turns scattered operations knowledge into searchable diagnostic paths, helping engineers map natural-language symptoms to relevant modules instead of relying on command memory.

Core capabilities and workflow

Disk and storage: covers disk-space, partitioning, filesystems, LVM, and SMART health checks.
Network: handles connectivity, packet loss, latency, full conntrack tables, and TCP retransmissions.
Performance and memory: includes CPU flame graphs, syscall hotspots, scheduling latency, IRQ balancing, file IO, OOM analysis, and memory leak diagnosis.
System and security: covers service status, time sync, package repositories, security baselines, CVE lookup, and kdump configuration.

It first parses intent, matches 1–3 modules, reads the relevant references/ documents, runs read-only diagnostics, and outputs conclusions in the module-defined report format. Dangerous operations such as reboot, mkfs, rm -rf, and firewall resets are provided as command text only, not executed directly.

Boundaries and caveats

It targets TencentOS differences, such as yum/dnf, iptables/firewalld, and ss/netstat. Performance modules prefer perf-prof and fall back when unavailable. For overlapping symptoms, focus on the primary module first to avoid treating correlated effects as the root cause.