Operating Systems - Medium

Troubleshooting Workflow: "Why Is This Server Slow"

Put the tools from this tier together and a standard diagnostic order emerges, roughly cheapest and broadest first, narrowing down to the most invasive:

uptime                        # is load actually elevated, and is it rising or falling?
vmstat 1 5                       # CPU breakdown (us/sy/id/wa) and any swap activity (si/so)
top                                 # which specific process(es) are actually consuming resources?
iostat -x 1 5                          # is a specific disk device saturated (%util near 100)?
strace -c -p <pid>                        # only once you've narrowed it to one suspect process

The wa (I/O wait) column from vmstat is usually the fork in the road: high wa alongside high load points you toward iostat and storage, while low wa with high us (user CPU time) points you toward top and application-level CPU consumption instead. Resisting the urge to jump straight to strace on a hunch, and instead working outward-in from these broad signals, is what separates a five-minute diagnosis from an hour of chasing the wrong process.