Cpu and Disk Usage Spiked without any reason

The behavior indicates a resource bottleneck on the VM host, most likely high CPU and/or disk I/O saturation inside the VM, which then makes SSH and the web app unresponsive.

Use these steps to diagnose and confirm the cause once access is possible again:

1. Check CPU saturation inside the VM (Linux)
   When SSH is responsive again, connect and run:

      top
   

   This shows which processes are consuming CPU. A process can reach near 100% CPU per core; multithreaded processes can show more than 100% total usage across cores.
   Also check load average and CPU count:

      uptime
      nproc
   

   Compare load average to CPU count: if load is higher than the number of CPUs and idle CPU is low, the VM is CPU‑saturated.
2. Check for CPU wait on I/O (wa%)
   Use vmstat to see if the CPU is mostly waiting on disk I/O:

      vmstat 2 5
   

   Focus on:
   • wa: time spent waiting for I/O
   • r: run queue length
     High wa with high r indicates processes are blocked on disk I/O, which matches symptoms of an unresponsive system.
3. Identify disk I/O bottlenecks
   Use iostat to see which disks are saturated and whether latency/queue sizes are high:

      iostat -dxctm 1
   

   The extended stats show throughput, IOPS, queue size, and latency per block device. High utilization and large queue sizes indicate disk saturation, which can explain the monitoring spike and unresponsiveness.
4. Correlate with Azure diagnostics
   If PerfInsights or Azure diagnostics were run, open the PerfInsights report and use:
   • Top CPU Consumers tab to see which processes caused high CPU usage and when.
   • Top Disk Consumers tab to see which processes generated the most disk I/O during the spike. This helps confirm whether Docker or a specific container is responsible.
5. Interpret the monitoring anomaly (111 GB disk bytes)
   The “disk bytes” metric is cumulative I/O throughput, not disk capacity. A value like 111 GB can mean the VM has read/written that much data over the sampling period, even if the disk is smaller. Sustained high I/O from containers or background processes can produce such numbers and cause the VM to stall.
6. Mitigation and next steps
   Once the offending process is identified:
   • Stop or limit the container(s) generating excessive CPU or disk I/O.
   • If the workload is expected to be heavy, consider scaling the VM size up (more CPU/IOPS) or moving to a disk SKU with higher throughput/IOPS.
   • If the spike was unexpected, review recent changes to containers, images, or jobs that might have triggered heavy I/O (e.g., logging, backups, database operations).

If the VM remains inaccessible for a long time, use the Azure portal to capture diagnostics (such as PerfInsights) or restart the VM as a last resort, understanding that this will terminate current processes.

References:

• Troubleshoot CPU performance issues in Linux
• Troubleshoot performance issues and isolate bottlenecks in Linux
• How to use PerfInsights to diagnose high CPU or disk usage issues