A bare-metal database server (64-core, 512GB RAM, 8x NVMe in RAID-10) is experiencing write latency spikes every 20 minutes that last ~30 seconds. Application logs show query timeouts during these windows. You SSH into the box. What specific commands do you run to determine whether the issue is filesystem journaling, a RAID rebuild, swap activity, or a runaway background process flushing dirty pages? For each potential cause, state the one-liner that confirms or rules it out.