echo “=== Hadoop Fault Diagnosis ===”
echo “Date: $(date)”

# 1. HDFS诊断
echo “=== HDFS Diagnosis ===”
LIVE=$(hdfs dfsadmin -report | grep “Live datanodes” | awk ‘{print $3}’)
DEAD=$(hdfs dfsadmin -report | grep “Dead datanodes” | awk ‘{print $3}’)
MISSING=$(hdfs dfsadmin -report | grep “Missing blocks” | awk ‘{print $3}’)
CORRUPT=$(hdfs fsck / -list-corruptfileblocks 2>/dev/null | grep “CORRUPT” | wc -l)

echo “Live Nodes: ${LIVE}”
echo “Dead Nodes: ${DEAD}”
echo “Missing Blocks: ${MISSING}”
echo “Corrupt Files: ${CORRUPT}”

# 2. YARN诊断
echo “=== YARN Diagnosis ===”
RUNNING=$(yarn node -list | grep “RUNNING” | wc -l)
UNHEALTHY=$(yarn node -list -states UNHEALTHY | grep “UNHEALTHY” | wc -l)

echo “Running Nodes: ${RUNNING}”
echo “Unhealthy Nodes: ${UNHEALTHY}”

# 3. 系统资源诊断
echo “=== System Resource Diagnosis ===”
CPU=$(top -bn1 | grep “Cpu(s)” | awk ‘{print $2}’)
MEM=$(free | grep Mem | awk ‘{printf “%.1f”, $3/$2 * 100.0}’)
DISK=$(df -h /bigdata | tail -1 | awk ‘{print $5}’)

echo “CPU Usage: ${CPU}%”
echo “Memory Usage: ${MEM}%”
echo “Disk Usage: ${DISK}”

# 4. 故障判断
if [ ${DEAD} -gt 0 ] || [ ${MISSING} -gt 0 ] || [ ${CORRUPT} -gt 0 ]; then
echo “ALERT: HDFS has issues!”
fi

if [ ${UNHEALTHY} -gt 0 ]; then
echo “ALERT: YARN has unhealthy nodes!”
fi