Q: You are setting up an on-call rotation for your SRE team. What are the best practices for a healthy and sustainable on-call process?
On-call is a critical SRE function, but poorly managed on-call leads to burnout, high turnover, and slower incident response.
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: "Never reboot a server blindly; always capture top process telemetry, lsof descriptors, and thread dumps first. The interviewer is testing: On-call practices, alert fatigue, SRE team health.. I structure my answer around systematic triage first, root cause analysis second, and permanent remediation third."
Initial Diagnostics & Root Cause Analysis
On-call is a critical SRE function, but poorly managed on-call leads to burnout, high turnover, and slower incident response.
- Rotation structure:
- Minimum 2-person rotation (primary + secondary) to prevent single points of failure
- 1-week rotations (longer leads to burnout; shorter disrupts sleep patterns)
- Follow-the-sun model for global teams (no one is on-call during sleep hours)
- Alert quality:
- Every alert must be actionable—if the on-call can't do anything about it, it shouldn't page
- Target: fewer than 2 pages per 12-hour shift on average
- Review and eliminate noisy alerts monthly (alert fatigue kills response time)
- Use severity levels: P1 (page immediately), P2 (page during business hours), P3 (ticket only)
- Documentation and tooling:
- Every paging alert must link to a runbook
Remediation & Permanent Safeguards
Best practices:
- On-call engineers must have pre-provisioned access to all production systems
- Incident communication channels (Slack, PagerDuty, status page) must be set up
- Compensation and balance:
- On-call engineers receive compensation (extra pay, time off in lieu)
- After being paged overnight, the engineer gets the next day off or a late start
- Track on-call load per engineer to ensure equitable distribution
- Continuous improvement:
- Hold weekly on-call handoff meetings: outgoing on-call briefs incoming on-call
- Review every page: Was it actionable? Was the runbook sufficient? What can be automated?
- Track metrics: pages per shift, time-to-acknowledge, time-to-resolve
- Rotation structure:
- Minimum 2-person rotation (primary + secondary) to prevent single points of failure
- 1-week rotations (longer leads to burnout; shorter disrupts sleep patterns)
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 setting call rotation 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 setting call rotation
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 setting call rotation 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 setting call rotation
Zombie processes consume process table entries; accumulating zombies can exhaust the system maximum PID limit (
pid_max).