Q: Your team receives PagerDuty alerts that CPU utilization is consistently hitting 100% on a database server. How do you find what's causing it?
First, I'd SSH into the server and run top or htop. I'd check the load average to see the trend (1, 5, 15 minutes).
Want to master this scenario in a live sandbox? The Linux Foundation's Certified System Administrator (LFCS) & SRE Track covers this exact problem with hands-on terminal drills.
🛠️ Production Runbook & Step-by-Step Resolution
Direct Answer: "We encountered this OS-level bottleneck during peak traffic and diagnosed it down to kernel and filesystem metrics. The interviewer is testing: CPU troubleshooting workflow, differentiating between system, user, and wait CPU.. I structure my answer around systematic triage first, root cause analysis second, and permanent remediation third."
Initial Diagnostics & Root Cause Analysis
First, I'd SSH into the server and run top or htop. I'd check the load average to see the trend (1, 5, 15 minutes).
- High
%us(user): The application or DB process is working hard. I'd find the PID, then usestrace -porperf topor check DB-specific tools (likeSHOW PROCESSLISTin MySQL) to see what queries are running. - High
%sy(system): The kernel is doing heavy work (e.g., millions of context switches or network interrupts). - High
%wa(iowait): The CPU is idling waiting for I/O. This means it's a disk bottleneck, not a CPU bottleneck. I'd switch toiotoporiostat -x 1to find the process thrashing the disk.
Remediation & Permanent Safeguards
In top, I'd look at the CPU states:
- High
%st(steal): If it's a VM, the noisy neighbor syndrome. The hypervisor is allocating cycles elsewhere. I would migrate the VM or resize the instance.
- High %us (user): The application or DB process is working hard. I'd find the PID, then use strace...
- High %sy (system): The kernel is doing heavy work (e.g., millions of context switches or network ...
- High %wa (iowait): The CPU is idling waiting for I/O. This means it's a disk bottleneck, not a CP...
Senior hiring managers and bar raisers evaluate how you prioritize during outages. Avoid these common anti-patterns:
- Issuing immediate kill -9 (SIGKILL) commands rather than allowing graceful termination via kill -15 (SIGTERM) when configuring team receives PagerDuty SIGKILL immediately halts execution without allowing processes to flush memory buffers, commit database transactions, or remove stale PID locks.
-
Overlooking file system inode exhaustion (df -i) when diagnosing 'No space left on device' errors when configuring team receives PagerDuty
Millions of small files or unrotated session tokens can exhaust available file system inodes even when standard disk storage (
df -h) shows ample free space. - Exhausting open file descriptor allocations (ulimit -n) under intensive concurrent socket connections when configuring team receives PagerDuty Reaching file descriptor exhaustion throws 'Too many open files' errors, preventing services from accepting new TCP client socket connections.
-
Permitting defunct zombie processes to accumulate without fixing parent process SIGCHLD waitpid reaping when configuring team receives PagerDuty
Zombie processes consume process table entries; accumulating zombies can exhaust the system maximum PID limit (
pid_max).