backup

diagnosis/perf.sh

@@ -0,0 +1,99 @@
+#!/bin/bash
+
+# Usage: ./trace_container.sh <container_name> <duration_seconds>
+CONTAINER_NAME=$1
+DURATION=$2
+FLAMEGRAPH_DIR="$HOME/FlameGraph" # Adjust this path
+OUTPUT_DIR="./flamegraphs"
+
+if [ -z "$CONTAINER_NAME" ] || [ -z "$DURATION" ]; then
+    echo "Usage: $0 <container_name> <duration>"
+    exit 1
+fi
+
+# 1. Get the long Container ID
+CONTAINER_ID=$(docker inspect --format='{{.Id}}' "$CONTAINER_NAME")
+if [ $? -ne 0 ]; then echo "Container not found"; exit 1; fi
+
+# 2. Get all PIDs in the container
+PIDS=$(docker top "$CONTAINER_NAME" -eo pid | tail -n +2 | tr '\n' ',' | sed 's/,$//')
+if [ -z "$PIDS" ]; then
+    echo "No processes found in container"
+    exit 1
+fi
+
+# Create output directory if it doesn't exist
+mkdir -p "$OUTPUT_DIR"
+
+echo "--- Tracing Container: $CONTAINER_NAME ---"
+echo "Container ID: $CONTAINER_ID"
+echo "Tracing PIDs: $PIDS"
+echo "Duration: $DURATION seconds"
+echo "Number of PIDs: $(echo $PIDS | tr ',' '\n' | wc -l)"
+
+# 3. Start Off-CPU tracing in the background (using perf sched events)
+echo "Starting Off-CPU trace..."
+sudo /root/linux/tools/perf/perf record -e sched:sched_switch -a -g --pid="$PIDS" -o "$OUTPUT_DIR/offcpu.data" -- sleep "$DURATION" &
+OFF_PID=$!
+
+# 4. Start On-CPU tracing (using Perf) - increase frequency
+echo "Starting On-CPU trace..."
+sudo /root/linux/tools/perf/perf record -F 99 -e cycles -a -g --pid="$PIDS" -o "$OUTPUT_DIR/oncpu.data" -- sleep "$DURATION"
+
+# Wait for Off-CPU to finish
+wait $OFF_PID
+
+echo "--- Processing Results ---"
+
+# Check if data files were created and have content
+if [ ! -f "$OUTPUT_DIR/oncpu.data" ]; then
+    echo "ERROR: oncpu.data not created"
+    exit 1
+fi
+
+if [ ! -f "$OUTPUT_DIR/offcpu.data" ]; then
+    echo "ERROR: offcpu.data not created"
+    exit 1
+fi
+
+echo "Data files created successfully"
+ls -lh "$OUTPUT_DIR"/*.data
+
+# Check if FlameGraph scripts exist
+if [ ! -f "$FLAMEGRAPH_DIR/stackcollapse-perf.pl" ]; then
+    echo "ERROR: stackcollapse-perf.pl not found at $FLAMEGRAPH_DIR"
+    exit 1
+fi
+
+if [ ! -f "$FLAMEGRAPH_DIR/flamegraph.pl" ]; then
+    echo "ERROR: flamegraph.pl not found at $FLAMEGRAPH_DIR"
+    exit 1
+fi
+
+# 5. Generate On-CPU Flamegraph
+echo "Processing On-CPU data..."
+/root/linux/tools/perf/perf script -i "$OUTPUT_DIR/oncpu.data" | "$FLAMEGRAPH_DIR/stackcollapse-perf.pl" > "$OUTPUT_DIR/oncpu.folded"
+ONCPU_LINES=$(wc -l < "$OUTPUT_DIR/oncpu.folded")
+echo "On-CPU folded lines: $ONCPU_LINES"
+
+if [ "$ONCPU_LINES" -gt 0 ]; then
+    "$FLAMEGRAPH_DIR/flamegraph.pl" "$OUTPUT_DIR/oncpu.folded" > "$OUTPUT_DIR/${CONTAINER_NAME}_oncpu.svg"
+    echo "Created $OUTPUT_DIR/${CONTAINER_NAME}_oncpu.svg"
+else
+    echo "WARNING: No On-CPU data captured"
+fi
+
+# 6. Generate Off-CPU Flamegraph
+echo "Processing Off-CPU data..."
+/root/linux/tools/perf/perf script -i "$OUTPUT_DIR/offcpu.data" | "$FLAMEGRAPH_DIR/stackcollapse-perf.pl" > "$OUTPUT_DIR/offcpu.folded"
+OFFCPU_LINES=$(wc -l < "$OUTPUT_DIR/offcpu.folded")
+echo "Off-CPU folded lines: $OFFCPU_LINES"
+
+if [ "$OFFCPU_LINES" -gt 0 ]; then
+    "$FLAMEGRAPH_DIR/flamegraph.pl" --color=io --title="Off-CPU: $CONTAINER_NAME" "$OUTPUT_DIR/offcpu.folded" > "$OUTPUT_DIR/${CONTAINER_NAME}_offcpu.svg"
+    echo "Created $OUTPUT_DIR/${CONTAINER_NAME}_offcpu.svg"
+else
+    echo "WARNING: No Off-CPU data captured"
+fi
+
+echo "Done! Check $OUTPUT_DIR/${CONTAINER_NAME}_oncpu.svg and $OUTPUT_DIR/${CONTAINER_NAME}_offcpu.svg"

diagnosis/strace.sh

@@ -0,0 +1,147 @@
+#!/bin/bash
+
+# Usage: ./strace.sh <container_name> <duration_seconds>
+CONTAINER_NAME=$1
+DURATION=$2
+OUTPUT_DIR="./strace_output"
+
+if [ -z "$CONTAINER_NAME" ] || [ -z "$DURATION" ]; then
+    echo "Usage: $0 <container_name> <duration>"
+    exit 1
+fi
+
+# 1. Get the long Container ID
+CONTAINER_ID=$(docker inspect --format='{{.Id}}' "$CONTAINER_NAME")
+if [ $? -ne 0 ]; then echo "Container not found"; exit 1; fi
+
+# 2. Get all PIDs in the container
+PIDS=$(docker top "$CONTAINER_NAME" -eo pid | tail -n +2)
+if [ -z "$PIDS" ]; then
+    echo "No processes found in container"
+    exit 1
+fi
+
+# Create output directory if it doesn't exist
+mkdir -p "$OUTPUT_DIR"
+
+echo "--- Strace Container: $CONTAINER_NAME ---"
+echo "Container ID: $CONTAINER_ID"
+echo "Duration: $DURATION seconds"
+echo "Number of PIDs: $(echo "$PIDS" | wc -l)"
+echo "PIDs: $(echo "$PIDS" | tr '\n' ' ')"
+
+# Timestamp for output files
+TIMESTAMP=$(date +%Y%m%d_%H%M%S)
+OUTPUT_FILE="$OUTPUT_DIR/${CONTAINER_NAME}_${TIMESTAMP}.strace"
+SUMMARY_FILE="$OUTPUT_DIR/${CONTAINER_NAME}_${TIMESTAMP}_summary.txt"
+
+# 3. Build strace command for all PIDs
+STRACE_PIDS=""
+for pid in $PIDS; do
+    STRACE_PIDS="$STRACE_PIDS -p $pid"
+done
+
+echo "Starting strace..."
+echo "Output file: $OUTPUT_FILE"
+
+# 4. Run strace with timing information and follow forks
+# -f: follow forks
+# -t: print timestamps
+# -T: show time spent in each syscall
+# -y: print paths associated with file descriptors
+# -yy: print protocol specific information for socket file descriptors
+# -s: max string size to print (default 32, increase if needed)
+# -o: output file
+# Use timeout with -s SIGINT to properly stop strace, and -k to SIGKILL if needed
+timeout -s SIGINT -k 5 "$DURATION" sudo strace -f -t -T -y -s 256 $STRACE_PIDS -o "$OUTPUT_FILE" 2>&1
+
+STRACE_EXIT=$?
+
+if [ $STRACE_EXIT -eq 124 ]; then
+    echo "Strace completed (timeout reached)"
+elif [ $STRACE_EXIT -eq 0 ]; then
+    echo "Strace completed successfully"
+else
+    echo "Strace exited with code $STRACE_EXIT"
+fi
+
+echo "--- Processing Results ---"
+
+# Check if output file was created and has content
+if [ ! -f "$OUTPUT_FILE" ]; then
+    echo "ERROR: strace output file not created"
+    exit 1
+fi
+
+FILE_SIZE=$(stat -f%z "$OUTPUT_FILE" 2>/dev/null || stat -c%s "$OUTPUT_FILE" 2>/dev/null)
+echo "Output file size: $FILE_SIZE bytes"
+
+if [ "$FILE_SIZE" -eq 0 ]; then
+    echo "WARNING: No strace data captured"
+    exit 1
+fi
+
+# 5. Generate summary statistics
+echo "Generating summary..."
+
+{
+    echo "=== Strace Summary for $CONTAINER_NAME ==="
+    echo "Timestamp: $TIMESTAMP"
+    echo "Duration: $DURATION seconds"
+    echo "Container ID: $CONTAINER_ID"
+    echo "PIDs traced: $(echo "$PIDS" | tr '\n' ' ')"
+    echo ""
+    echo "=== Top 20 System Calls by Count ==="
+    grep -oP '^\d+\s+\d+:\d+:\d+\s+\K[a-z_0-9]+(?=\()' "$OUTPUT_FILE" | sort | uniq -c | sort -rn | head -20
+    echo ""
+    echo "=== Top 20 System Calls by Time (if -T was used) ==="
+    grep -oP '<[\d.]+>' "$OUTPUT_FILE" | sed 's/[<>]//g' | awk '{sum+=$1; count++} END {if(count>0) print "Average syscall time:", sum/count, "seconds"}'
+    echo ""
+    echo "=== Error Summary (failed syscalls) ==="
+    grep -E 'E[A-Z]+' "$OUTPUT_FILE" | grep -oP 'E[A-Z]+' | sort | uniq -c | sort -rn | head -10
+    echo ""
+    echo "=== File Operations ==="
+    echo "open/openat calls:"
+    grep -c 'open\(at\)\?(' "$OUTPUT_FILE" || echo "0"
+    echo "read/pread calls:"
+    grep -c 'p\?read(' "$OUTPUT_FILE" || echo "0"
+    echo "write/pwrite calls:"
+    grep -c 'p\?write(' "$OUTPUT_FILE" || echo "0"
+    echo "close calls:"
+    grep -c 'close(' "$OUTPUT_FILE" || echo "0"
+    echo ""
+    echo "=== Network Operations ==="
+    echo "socket calls:"
+    grep -c 'socket(' "$OUTPUT_FILE" || echo "0"
+    echo "connect calls:"
+    grep -c 'connect(' "$OUTPUT_FILE" || echo "0"
+    echo "send/write calls:"
+    grep -c 'send\|sendto\|sendmsg' "$OUTPUT_FILE" || echo "0"
+    echo "recv/read calls:"
+    grep -c 'recv\|recvfrom\|recvmsg' "$OUTPUT_FILE" || echo "0"
+    echo ""
+    echo "=== Memory Operations ==="
+    echo "mmap calls:"
+    grep -c 'mmap(' "$OUTPUT_FILE" || echo "0"
+    echo "munmap calls:"
+    grep -c 'munmap(' "$OUTPUT_FILE" || echo "0"
+    echo "brk calls:"
+    grep -c 'brk(' "$OUTPUT_FILE" || echo "0"
+    echo ""
+    echo "=== Process/Thread Operations ==="
+    echo "clone calls:"
+    grep -c 'clone(' "$OUTPUT_FILE" || echo "0"
+    echo "fork calls:"
+    grep -c 'fork(' "$OUTPUT_FILE" || echo "0"
+    echo "execve calls:"
+    grep -c 'execve(' "$OUTPUT_FILE" || echo "0"
+    echo "exit calls:"
+    grep -c 'exit\(_group\)\?(' "$OUTPUT_FILE" || echo "0"
+} > "$SUMMARY_FILE"
+
+echo "Summary saved to: $SUMMARY_FILE"
+echo ""
+cat "$SUMMARY_FILE"
+echo ""
+echo "Done! Full strace output: $OUTPUT_FILE"
+echo "Summary: $SUMMARY_FILE"