Java的CPU性能问题排查的shell脚本以及测试

534 阅读10分钟

🍺 show-busy-java-threads

用于快速排查JavaCPU性能问题(top us值过高),自动查出运行的Java进程中消耗CPU多的线程,并打印出其线程栈,从而确定导致性能问题的方法调用。
目前只支持Linux。原因是MacWindowsps命令不支持列出进程的线程id,更多信息参见 #33,欢迎提供解法。

  1. top命令找出消耗CPU高的Java进程及其线程id

    1. 开启线程显示模式(top -H,或是打开top后按H
    2. CPU使用率排序(top缺省是按CPU使用降序,已经合要求;打开top后按P可以显式指定按CPU使用降序)
    3. 记下Java进程id及其CPU高的线程id
  2. 查看消耗CPU高的线程栈:

    1. 用进程id作为参数,jstack出有问题的Java进程
    2. 手动转换线程id成十六进制(可以用printf %x 1234
    3. jstack输出中查找十六进制的线程id(可以用vim的查找功能/0x1234,或是grep 0x1234 -A 20
  3. 查看对应的线程栈,分析问题

查问题时,会要多次上面的操作以分析确定问题,这个过程太繁琐太慢了
期望整合上面的过程成一个脚本,这样一行命令就可以自动化地搞定。

用法

show-busy-java-threads
# 从所有运行的Java进程中找出最消耗CPU的线程(缺省5个),打印出其线程栈

# 缺省会自动从所有的Java进程中找出最消耗CPU的线程,这样用更方便
# 当然你可以通过 -p 选项 手动指定要分析的Java进程Id,以保证只会显示你关心的那个Java进程的信息
show-busy-java-threads -p <指定的Java进程Id>
show-busy-java-threads -p 42
show-busy-java-threads -p 42,47

show-busy-java-threads -c <要展示示的线程栈个数>

show-busy-java-threads <重复执行的间隔秒数> [<重复执行的次数>]
# 多次执行;这2个参数的使用方式类似vmstat命令

show-busy-java-threads -a <运行输出的记录到的文件>
# 记录到文件以方便回溯查看

show-busy-java-threads -S <存储jstack输出文件的目录>
# 指定jstack输出文件的存储目录,方便记录以后续分析

##############################
# 注意:
##############################
# 如果Java进程的用户 与 执行脚本的当前用户 不同,则jstack不了这个Java进程
# 为了能切换到Java进程的用户,需要加sudo来执行,即可以解决:
sudo show-busy-java-threads

show-busy-java-threads -s <指定jstack命令的全路径>
# 对于sudo方式的运行,JAVA_HOME环境变量不能传递给root,
# 而root用户往往没有配置JAVA_HOME且不方便配置,不能找到jstack命令。
# 这时显式指定jstack命令的路径就反而显得更方便了

# -m 选项:执行jstack命令时加上 -m 选项,显示上Native的栈帧,一般应用排查不需要使用
show-busy-java-threads -m
# -F 选项:执行jstack命令时加上 -F 选项(如果直接jstack无响应时,用于强制jstack),一般情况不需要使用
show-busy-java-threads -F
# -l 选项:执行jstack命令时加上 -l 选项,显示上更多相关锁的信息,一般情况不需要使用
# 注意:和 -m -F 选项一起使用时,可能会大大增加jstack操作的耗时
show-busy-java-threads -l

# 帮助信息
$ show-busy-java-threads -h
Usage: show-busy-java-threads [OPTION]... [delay [count]]
Find out the highest cpu consumed threads of java processes,
and print the stack of these threads.

Example:
  show-busy-java-threads       # show busy java threads info
  show-busy-java-threads 1     # update every 1 second, (stop by eg: CTRL+C)
  show-busy-java-threads 3 10  # update every 3 seconds, update 10 times

Output control:
  -p, --pid <java pid(s)>   find out the highest cpu consumed threads from
                            the specified java process.
                            support pid list(eg: 42,47).
                            default from all java process.
  -c, --count <num>         set the thread count to show, default is 5.
                            set count 0 to show all threads.
  -a, --append-file <file>  specifies the file to append output as log.
  -S, --store-dir <dir>     specifies the directory for storing
                            the intermediate files, and keep files.
                            default store intermediate files at tmp dir,
                            and auto remove after run. use this option to keep
                            files so as to review jstack/top/ps output later.
  delay                     the delay between updates in seconds.
  count                     the number of updates.
                            delay/count arguments imitates the style of
                            vmstat command.

jstack control:
  -s, --jstack-path <path>  specifies the path of jstack command.
  -F, --force               set jstack to force a thread dump.
                            use when jstack does not respond (process is hung).
  -m, --mix-native-frames   set jstack to print both java and
                            native frames (mixed mode).
  -l, --lock-info           set jstack with long listing.
                            prints additional information about locks.

CPU usage calculation control:
  -i, --cpu-sample-interval specifies the delay between cpu samples to get
                            thread cpu usage percentage during this interval.
                            default is 0.5 (second).
                            set interval 0 to get the percentage of time spent
                            running during the *entire lifetime* of a process.

Miscellaneous:
  -h, --help                display this help and exit.
  -V, --version             display version information and exit.

示例

$ show-busy-java-threads
[1] Busy(57.0%) thread(23355/0x5b3b) stack of java process(23269) under user(admin):
"pool-1-thread-1" prio=10 tid=0x000000005b5c5000 nid=0x5b3b runnable [0x000000004062c000]
   java.lang.Thread.State: RUNNABLE
    at java.text.DateFormat.format(DateFormat.java:316)
    at com.xxx.foo.services.common.DateFormatUtil.format(DateFormatUtil.java:41)
    at com.xxx.foo.shared.monitor.schedule.AppMonitorDataAvgScheduler.run(AppMonitorDataAvgScheduler.java:127)
    at com.xxx.foo.services.common.utils.AliTimer$2.run(AliTimer.java:128)
    at java.util.concurrent.ThreadPoolExecutor$Worker.runTask(ThreadPoolExecutor.java:886)
    at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:908)
    at java.lang.Thread.run(Thread.java:662)

[2] Busy(26.1%) thread(24018/0x5dd2) stack of java process(23269) under user(admin):
"pool-1-thread-2" prio=10 tid=0x000000005a968800 nid=0x5dd2 runnable [0x00000000420e9000]
   java.lang.Thread.State: RUNNABLE
    at java.util.Arrays.copyOf(Arrays.java:2882)
    at java.lang.AbstractStringBuilder.expandCapacity(AbstractStringBuilder.java:100)
    at java.lang.AbstractStringBuilder.append(AbstractStringBuilder.java:572)
    at java.lang.StringBuffer.append(StringBuffer.java:320)
    - locked <0x00000007908d0030> (a java.lang.StringBuffer)
    at java.text.SimpleDateFormat.format(SimpleDateFormat.java:890)
    at java.text.SimpleDateFormat.format(SimpleDateFormat.java:869)
    at java.text.DateFormat.format(DateFormat.java:316)
    at com.xxx.foo.services.common.DateFormatUtil.format(DateFormatUtil.java:41)
    at com.xxx.foo.shared.monitor.schedule.AppMonitorDataAvgScheduler.run(AppMonitorDataAvgScheduler.java:126)
    at com.xxx.foo.services.common.utils.AliTimer$2.run(AliTimer.java:128)
    at java.util.concurrent.ThreadPoolExecutor$Worker.runTask(ThreadPoolExecutor.java:886)
    at java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:908)
    at java.lang.Thread.run(Thread.java:662)

......

上面的线程栈可以看出,CPU消耗最高的2个线程都在执行java.text.DateFormat.format,业务代码对应的方法是shared.monitor.schedule.AppMonitorDataAvgScheduler.run。可以基本确定:

  • AppMonitorDataAvgScheduler.run调用DateFormat.format次数比较频繁。
  • DateFormat.format比较慢。(这个可以由DateFormat.format的实现确定。)

多执行几次show-busy-java-threads,如果上面情况高概率出现,则可以确定上面的判定。
因为调用越少代码执行越快,则出现在线程栈的概率就越低。
脚本有自动多次执行的功能,指定 重复执行的间隔秒数/重复执行的次数 参数。

脚本

#!/bin/bash
# @Function
# Find out the highest cpu consumed threads of java processes, and print the stack of these threads.
#
# @Usage
#   $ ./show-busy-java-threads
#
# @online-doc https://github.com/oldratlee/useful-scripts/blob/dev-2.x/docs/java.md#-show-busy-java-threads
# @author Jerry Lee (oldratlee at gmail dot com)
# @author superhj1987 (superhj1987 at 126 dot com)
#
# NOTE about Bash Traps and Pitfalls:
#
# 1. DO NOT combine var declaration and assignment which value supplied by subshell in ONE line!
#    for example: readonly var1=$(echo value1)
#                 local var2=$(echo value1)
#
#    Because the combination make exit code of assignment to be always 0,
#      aka. the exit code of command in subshell is discarded.
#      tested on bash 3.2.57/4.2.46
#
#    solution is separation of var declaration and assignment:
#           var1=$(echo value1)
#           readonly var1
#           local var2
#           var2=$(echo value1)

# NOTE: DO NOT declare var PROG as readonly in ONE line!
PROG="$(basename "$0")"
readonly PROG
readonly PROG_VERSION='2.5.0-dev'
# choosing between $0 and BASH_SOURCE
# https://stackoverflow.com/a/35006505/922688
# How can I get the source directory of a Bash script from within the script itself?
# https://stackoverflow.com/questions/59895
# Will $0 always include the path to the script?
# https://unix.stackexchange.com/questions/119929
readonly -a COMMAND_LINE=("${BASH_SOURCE[0]}" "$@")
# Get current user name via whoami command
#   See https://www.lifewire.com/current-linux-user-whoami-command-3867579
# Because if run command by `sudo -u`, env var $USER is not rewritten/correct, just inherited from outside!
#
# NOTE: DO NOT declare var USER as readonly in ONE line!
USER="$(whoami)"
readonly USER

################################################################################
# util functions
################################################################################

# NOTE: $'foo' is the escape sequence syntax of bash
readonly ec=$'\033'      # escape char
readonly eend=$'\033[0m' # escape end
readonly nl=$'\n'        # new line

colorEcho() {
  local color=$1
  shift

  # if stdout is console, turn on color output.
  [ -t 1 ] && echo "${ec}[1;${color}m$*$eend" || echo "$@"
}

colorPrint() {
  local color=$1
  shift

  colorEcho "$color" "$@"
  [[ -n "$append_file" && -w "$append_file" ]] && echo "$@" >>"$append_file"
  [[ -n "$store_dir" && -w "$store_dir" ]] && echo "$@" >>"${store_file_prefix}$PROG"
}

# shellcheck disable=SC2120
normalPrint() {
  echo "$@"
  [[ -n "$append_file" && -w "$append_file" ]] && echo "$@" >>"$append_file"
  [[ -n "$store_dir" && -w "$store_dir" ]] && echo "$@" >>"${store_file_prefix}$PROG"
}

redPrint() {
  colorPrint 31 "$@"
}

greenPrint() {
  colorPrint 32 "$@"
}

yellowPrint() {
  colorPrint 33 "$@"
}

bluePrint() {
  colorPrint 36 "$@"
}

die() {
  redPrint "Error: $*" 1>&2
  exit 1
}

logAndRun() {
  echo "$@"
  echo
  "$@"
}

logAndCat() {
  echo "$@"
  echo
  cat
}

# Bash RegEx to check floating point numbers from user input
# https://stackoverflow.com/questions/13790763
isNonNegativeFloatNumber() {
  [[ "$1" =~ ^[+]?[0-9]+\.?[0-9]*$ ]]
}

isNaturalNumber() {
  [[ "$1" =~ ^[+]?[0-9]+$ ]]
}

isNaturalNumberList() {
  [[ "$1" =~ ^([0-9]+)(,[0-9]+)*$ ]]
}

# print calling(quoted) command line which is able to copy and paste to rerun safely
#
# How to get the complete calling command of a BASH script from inside the script (not just the arguments)
# https://stackoverflow.com/questions/36625593
printCallingCommandLine() {
  local arg isFirst=true
  for arg in "${COMMAND_LINE[@]}"; do
    if $isFirst; then
      isFirst=false
    else
      printf ' '
    fi
    printf '%q' "$arg"
  done
  echo
}

usage() {
  local -r exit_code="${1:-0}"
  (($# > 0)) && shift
  # shellcheck disable=SC2015
  [ "$exit_code" != 0 ] && local -r out=/dev/stderr || local -r out=/dev/stdout

  (($# > 0)) && colorEcho 31 "$*$nl" >$out

  cat >$out <<EOF
Usage: ${PROG} [OPTION]... [delay [count]]
Find out the highest cpu consumed threads of java processes,
and print the stack of these threads.

Example:
  ${PROG}       # show busy java threads info
  ${PROG} 1     # update every 1 second, (stop by eg: CTRL+C)
  ${PROG} 3 10  # update every 3 seconds, update 10 times

Output control:
  -p, --pid <java pid(s)>   find out the highest cpu consumed threads from
                            the specified java process.
                            support pid list(eg: 42,47).
                            default from all java process.
  -c, --count <num>         set the thread count to show, default is 5.
                            set count 0 to show all threads.
  -a, --append-file <file>  specifies the file to append output as log.
  -S, --store-dir <dir>     specifies the directory for storing
                            the intermediate files, and keep files.
                            default store intermediate files at tmp dir,
                            and auto remove after run. use this option to keep
                            files so as to review jstack/top/ps output later.
  delay                     the delay between updates in seconds.
  count                     the number of updates.
                            delay/count arguments imitates the style of
                            vmstat command.

jstack control:
  -s, --jstack-path <path>  specifies the path of jstack command.
  -F, --force               set jstack to force a thread dump.
                            use when jstack does not respond (process is hung).
  -m, --mix-native-frames   set jstack to print both java and
                            native frames (mixed mode).
  -l, --lock-info           set jstack with long listing.
                            prints additional information about locks.

CPU usage calculation control:
  -i, --cpu-sample-interval specifies the delay between cpu samples to get
                            thread cpu usage percentage during this interval.
                            default is 0.5 (second).
                            set interval 0 to get the percentage of time spent
                            running during the *entire lifetime* of a process.

Miscellaneous:
  -h, --help                display this help and exit.
  -V, --version             display version information and exit.
EOF

  exit "$exit_code"
}

progVersion() {
  echo "$PROG $PROG_VERSION"
  exit
}

################################################################################
# Check os support
################################################################################

uname | grep '^Linux' -q || die "$PROG only support Linux, not support $(uname) yet!"

################################################################################
# parse options
################################################################################

# DO NOT declare var ARGS as readonly in ONE line!
ARGS=$(
  getopt -n "$PROG" -a -o c:p:a:s:S:i:Pd:FmlhV \
    -l count:,pid:,append-file:,jstack-path:,store-dir:,cpu-sample-interval:,use-ps,top-delay:,force,mix-native-frames,lock-info,help,version \
    -- "$@"
) || {
  echo
  usage 1
}
readonly ARGS
eval set -- "${ARGS}"

count=5
cpu_sample_interval=0.5

while true; do
  case "$1" in
  -c | --count)
    count="$2"
    shift 2
    ;;
  -p | --pid)
    pid_list="$2"
    shift 2
    ;;
  -a | --append-file)
    append_file="$2"
    shift 2
    ;;
  -s | --jstack-path)
    jstack_path="$2"
    shift 2
    ;;
  -S | --store-dir)
    store_dir="$2"
    shift 2
    ;;
  # support the legacy option name -P,--use-ps for compatibility
  -P | --use-ps)
    cpu_sample_interval=0
    shift
    ;;
  # support the legacy option name -d,--top-delay for compatibility
  -i | --cpu-sample-interval | -d | --top-delay)
    cpu_sample_interval="$2"
    shift 2
    ;;
  -F | --force)
    force=-F
    shift
    ;;
  -m | --mix-native-frames)
    mix_native_frames=-m
    shift
    ;;
  -l | --lock-info)
    lock_info=-l
    shift
    ;;
  -h | --help)
    usage
    ;;
  -V | --version)
    progVersion
    ;;
  --)
    shift
    break
    ;;
  esac
done

readonly count cpu_sample_interval force mix_native_frames lock_info
readonly update_delay=${1:-0}
isNonNegativeFloatNumber "$update_delay" || die "update delay($update_delay) is not a non-negative float number!"

[ -z "$1" ] && update_count=1 || update_count=${2:-0}
isNaturalNumber "$update_count" || die "update count($update_count) is not a natural number!"
readonly update_count

if [ -n "$pid_list" ]; then
  pid_list="${pid_list//[[:space:]]/}" # delete white space
  isNaturalNumberList "$pid_list" || die "pid(s)($pid_list) is illegal! example: 42 or 42,99,67"
fi
readonly pid_list

# check the directory of append-file(-a) mode, create if not exist.
if [ -n "$append_file" ]; then
  if [ -e "$append_file" ]; then
    [ -f "$append_file" ] || die "$append_file(specified by option -a, for storing run output files) exists but is not a file!"
    [ -w "$append_file" ] || die "file $append_file(specified by option -a, for storing run output files) exists but is not writable!"
  else
    append_file_dir="$(dirname "$append_file")"
    if [ -e "$append_file_dir" ]; then
      [ -d "$append_file_dir" ] || die "directory $append_file_dir(specified by option -a, for storing run output files) exists but is not a directory!"
      [ -w "$append_file_dir" ] || die "directory $append_file_dir(specified by option -a, for storing run output files) exists but is not writable!"
    else
      mkdir -p "$append_file_dir" || die "fail to create directory $append_file_dir(specified by option -a, for storing run output files)!"
    fi
  fi
fi
readonly append_file

# check store directory(-S) mode, create directory if not exist.
if [ -n "$store_dir" ]; then
  if [ -e "$store_dir" ]; then
    [ -d "$store_dir" ] || die "$store_dir(specified by option -S, for storing output files) exists but is not a directory!"
    [ -w "$store_dir" ] || die "directory $store_dir(specified by option -S, for storing output files) exists but is not writable!"
  else
    mkdir -p "$store_dir" || die "fail to create directory $store_dir(specified by option -S, for storing output files)!"
  fi
fi
readonly store_dir

isNonNegativeFloatNumber "$cpu_sample_interval" || die "cpu sample interval($cpu_sample_interval) is not a non-negative float number!"

################################################################################
# search/check the existence of jstack command
#
# search order/priority:
#    1. from -s option
#    2. from under env var JAVA_HOME
#    3. from under env var PATH
################################################################################

if [ -n "$jstack_path" ]; then
  # 1. check jstack_path set by -s option
  [ -f "$jstack_path" ] || die "$jstack_path (set by -s option) is NOT found!"
  [ -x "$jstack_path" ] || die "$jstack_path (set by -s option) is NOT executable!"
elif [ -n "$JAVA_HOME" ]; then
  # 2. search jstack under JAVA_HOME
  if [ -f "$JAVA_HOME/bin/jstack" ]; then
    [ -x "$JAVA_HOME/bin/jstack" ] || die "found \$JAVA_HOME/bin/jstack($JAVA_HOME/bin/jstack) is NOT executable!${nl}Use -s option set jstack path manually."
    jstack_path="$JAVA_HOME/bin/jstack"
  elif [ -f "$JAVA_HOME/../bin/jstack" ]; then
    [ -x "$JAVA_HOME/../bin/jstack" ] || die "found \$JAVA_HOME/../bin/jstack($JAVA_HOME/../bin/jstack) is NOT executable!${nl}Use -s option set jstack path manually."
    jstack_path="$JAVA_HOME/../bin/jstack"
  fi
elif command -v jstack &>/dev/null; then
  # 3. search jstack under PATH
  jstack_path="$(command -v jstack)"
  [ -x "$jstack_path" ] || die "found $jstack_path from PATH is NOT executable!${nl}Use -s option set jstack path manually."
else
  die "jstack NOT found by JAVA_HOME(${JAVA_HOME:-not set}) setting and PATH!${nl}Use -s option set jstack path manually."
fi
readonly jstack_path

################################################################################
# biz logic
################################################################################

# NOTE: DO NOT declare var run_timestamp as readonly in ONE line!
run_timestamp="$(date "+%Y-%m-%d_%H:%M:%S.%N")"
readonly run_timestamp
readonly uuid="${PROG}_${run_timestamp}_${$}_${RANDOM}"

readonly tmp_store_dir="/tmp/${uuid}"
if [ -n "$store_dir" ]; then
  readonly store_file_prefix="$store_dir/${run_timestamp}_"
else
  readonly store_file_prefix="$tmp_store_dir/${run_timestamp}_"
fi
mkdir -p "$tmp_store_dir"

cleanupWhenExit() {
  rm -rf "$tmp_store_dir" &>/dev/null
}
trap cleanupWhenExit EXIT

headInfo() {
  colorEcho "0;34;42" ================================================================================
  echo "$(date "+%Y-%m-%d %H:%M:%S.%N") [$((update_round_num + 1))/$update_count]: $(printCallingCommandLine)"
  colorEcho "0;34;42" ================================================================================
  echo
}

if [ -n "${pid_list}" ]; then
  readonly ps_process_select_options="-p $pid_list"
else
  readonly ps_process_select_options="-C java -C jsvc"
fi

__die_when_no_java_process_found() {
  if [ -n "${pid_list}" ]; then
    die "process($pid_list) is not running, or not java process!"
  else
    die 'No java process found!'
  fi
}

# output field: pid, thread id(lwp), pcpu, user
#   order by pcpu(percentage of cpu usage)
#
# NOTE:
# use ps command to find busy thread(cpu usage)
# cpu usage of ps command is expressed as
# the percentage of time spent running during the *entire lifetime* of a process,
# this is not ideal in general.
findBusyJavaThreadsByPs() {
  # 1. sort by %cpu by ps option `--sort -pcpu`
  #    unfortunately, ps from `procps-ng 3.3.12`, `--sort` does not work properly with other options,
  #    use
  #       ps <other options>
  #    combined
  #       sort -k3,3nr
  #    instead of
  #       ps <other options> --sort -pcpu
  # 2. use wide output(unlimited width) by ps option `-ww`
  #    avoid trunk user column to username_fo+ or $uid alike

  # shellcheck disable=SC2206
  local -a ps_cmd_line=(ps $ps_process_select_options -wwLo 'pid,lwp,pcpu,user' --no-headers)
  # DO NOT combine var ps_out declaration and assignment in ONE line!
  local ps_out
  ps_out="$("${ps_cmd_line[@]}" | sort -k3,3nr)"
  [ -n "$ps_out" ] || __die_when_no_java_process_found

  if [ -n "$store_dir" ]; then
    echo "$ps_out" | logAndCat "${ps_cmd_line[*]} | sort -k3,3nr" >"${store_file_prefix}$((update_round_num + 1))_ps"
  fi

  if ((count > 0)); then
    echo "$ps_out" | head -n "${count}"
  else
    echo "$ps_out"
  fi
}

# top with output field: thread id, %cpu
__top_threadId_cpu() {
  # DO NOT combine var java_pid_list declaration and assignment in ONE line!
  local java_pid_list
  # shellcheck disable=SC2086
  java_pid_list="$(ps $ps_process_select_options -o pid --no-headers)"
  [ -n "$java_pid_list" ] || __die_when_no_java_process_found
  # shellcheck disable=SC2086
  java_pid_list="$(echo $java_pid_list | tr ' ' ,)" # join with ,

  # 1. sort by %cpu by top option `-o %CPU`
  #    unfortunately, top version 3.2 does not support -o option(supports from top version 3.3+),
  #    use
  #       HOME="$tmp_store_dir" top -H -b -n 1
  #    combined
  #       sort
  #    instead of
  #       HOME="$tmp_store_dir" top -H -b -n 1 -o '%CPU'
  # 2. change HOME env var when run top,
  #    so as to prevent top command output format being change by .toprc user config file unexpectedly
  # 3. use option `-d 0.5`(update interval 0.5 second) and `-n 2`(update 2 times),
  #    and use second time update data to get cpu percentage of thread in 0.5 second interval
  # 4. top v3.3, there is 1 black line between 2 update;
  #    but top v3.2, there is 2 blank lines between 2 update!
  local -a top_cmd_line=(top -H -b -d "$cpu_sample_interval" -n 2 -p "$java_pid_list")
  # DO NOT combine var ps_out declaration and assignment in ONE line!
  local top_out
  top_out=$(HOME="$tmp_store_dir" "${top_cmd_line[@]}")
  if [ -n "$store_dir" ]; then
    echo "$top_out" | logAndCat "${top_cmd_line[@]}" >"${store_file_prefix}$((update_round_num + 1))_top"
  fi

  # DO NOT combine var result_threads_top_info declaration and assignment in ONE line!
  local result_threads_top_info
  result_threads_top_info=$(
    echo "$top_out" | awk '{
            # from text line to empty line, increase block index
            if (previousLine && !$0) blockIndex++
            # only print 4th text block(blockIndex == 3), aka. process info of second top update
            if (blockIndex == 3 && $1 ~ /^[0-9]+$/)
                print $1, $9  # $1 is thread id field, $9 is %cpu field
            previousLine = $0
        }'
  )
  [ -n "$result_threads_top_info" ] || __die_when_no_java_process_found

  echo "$result_threads_top_info" | sort -k2,2nr
}

__complete_pid_user_by_ps() {
  # ps output field: pid, thread id(lwp), user
  # shellcheck disable=SC2206
  local -a ps_cmd_line=(ps $ps_process_select_options -wwLo 'pid,lwp,user' --no-headers)
  # DO NOT combine var ps_out declaration and assignment in ONE line!
  local ps_out
  ps_out="$("${ps_cmd_line[@]}")"
  if [ -n "$store_dir" ]; then
    echo "$ps_out" | logAndCat "${ps_cmd_line[@]}" >"${store_file_prefix}$((update_round_num + 1))_ps"
  fi

  local idx=0 threadId pcpu output_fields
  while read -r threadId pcpu; do
    ((count <= 0 || idx < count)) || break

    # output field: pid, threadId, pcpu, user
    output_fields="$(echo "$ps_out" |
      awk -v "threadId=$threadId" -v "pcpu=$pcpu" '$2==threadId {
                print $1, threadId, pcpu, $3; exit
            }')"
    if [ -n "$output_fields" ]; then
      ((idx++))
      echo "$output_fields"
    fi
  done
}

# output format is same as function findBusyJavaThreadsByPs
findBusyJavaThreadsByTop() {
  __top_threadId_cpu | __complete_pid_user_by_ps
}

printStackOfThreads() {
  local idx=0 pid threadId pcpu user threadId0x
  while read -r pid threadId pcpu user; do
    threadId0x="0x$(printf %x "${threadId}")"

    ((idx++))
    local jstackFile="${store_file_prefix}$((update_round_num + 1))_jstack_${pid}"
    [ -f "${jstackFile}" ] || {
      # shellcheck disable=SC2206
      local -a jstack_cmd_line=("$jstack_path" ${force} $mix_native_frames $lock_info ${pid})
      if [ "${user}" == "${USER}" ]; then
        # run without sudo, when java process user is current user
        logAndRun "${jstack_cmd_line[@]}" >"${jstackFile}"
      elif [ $UID == 0 ]; then
        # if java process user is not current user, must run jstack with sudo
        logAndRun sudo -u "${user}" "${jstack_cmd_line[@]}" >"${jstackFile}"
      else
        # current user is not root user, so can not run with sudo; print error message and rerun suggestion
        redPrint "[$idx] Fail to jstack busy(${pcpu}%) thread(${threadId}/${threadId0x}) stack of java process(${pid}) under user(${user})."
        redPrint "User of java process($user) is not current user($USER), need sudo to rerun:"
        yellowPrint "    sudo $(printCallingCommandLine)"
        normalPrint
        continue
      fi || {
        redPrint "[$idx] Fail to jstack busy(${pcpu}%) thread(${threadId}/${threadId0x}) stack of java process(${pid}) under user(${user})."
        normalPrint
        rm "${jstackFile}" &>/dev/null
        continue
      }
    }

    bluePrint "[$idx] Busy(${pcpu}%) thread(${threadId}/${threadId0x}) stack of java process(${pid}) under user(${user}):"

    if [ -n "$mix_native_frames" ]; then
      local sed_script="/--------------- $threadId ---------------/,/^---------------/ {
                /--------------- $threadId ---------------/b # skip first separator line
                /^---------------/d # delete second separator line
                p
            }"
    elif [ -n "$force" ]; then
      local sed_script="/^Thread ${threadId}:/,/^$/ {
                /^$/d; p # delete end separator line
            }"
    else
      local sed_script="/ nid=${threadId0x} /,/^$/ {
                /^$/d; p # delete end separator line
            }"
    fi
    {
      sed "$sed_script" -n "${jstackFile}"
      echo
    } | tee ${append_file:+-a "$append_file"} ${store_dir:+-a "${store_file_prefix}$PROG"}
  done
}

################################################################################
# Main
################################################################################

main() {
  local update_round_num
  # if update_count <= 0, infinite loop till user interrupted (eg: CTRL+C)
  for ((update_round_num = 0; update_count <= 0 || update_round_num < update_count; ++update_round_num)); do
    ((update_round_num > 0)) && sleep "$update_delay"

    [[ -n "$append_file" || -n "$store_dir" ]] && headInfo |
      tee ${append_file:+-a "$append_file"} ${store_dir:+-a "${store_file_prefix}$PROG"} >/dev/null
    ((update_count != 1)) && headInfo

    if [ "$cpu_sample_interval" == 0 ]; then
      findBusyJavaThreadsByPs
    else
      findBusyJavaThreadsByTop
    fi | printStackOfThreads
  done
}

main

测试

public class CPULoad {
    public static void main(String[] args) {
        // 创建并启动一个线程来占用 CPU 资源
        Thread cpuLoadThread = new Thread(() -> {
            while (true) {
                // 在这里添加一些耗时的计算任务,以让 CPU 跑满
                // 这里简单地使用一个空循环来示例,但您可以根据需要执行实际的计算任务
                for (int i = 0; i < Integer.MAX_VALUE; i++) {
                    // 空循环
                }
            }
        });
        cpuLoadThread.start();

        // 主线程休眠一段时间,让 CPU 被占用
        try {
            Thread.sleep(5000); // 这里休眠 5 秒,您可以根据需要调整时间
        } catch (InterruptedException e) {
            e.printStackTrace();
        }

        // 停止 CPU 占用线程
        cpuLoadThread.interrupt();
    }
}

安装java

yum install -y java java-devel

编译运行

javac CPULoad.java
java CPULoad

使用脚本

[root@hecs-76151 ~]# ./cpu.sh -p 2146218
[1] Busy(98.0%) thread(2146230/0x20bfb6) stack of java process(2146218) under user(root):
"Thread-0" #8 prio=5 os_prio=0 tid=0x00007fc3c420c000 nid=0x20bfb6 runnable [0x00007fc3ad62a000]
   java.lang.Thread.State: RUNNABLE
        at CPULoad.lambda$main$0(CPULoad.java:8)
        at CPULoad$$Lambda$1/471910020.run(Unknown Source)
        at java.lang.Thread.run(Thread.java:750)

[2] Busy(0.0%) thread(2146218/0x20bfaa) stack of java process(2146218) under user(root):

[3] Busy(0.0%) thread(2146219/0x20bfab) stack of java process(2146218) under user(root):
"DestroyJavaVM" #9 prio=5 os_prio=0 tid=0x00007fc3c404b800 nid=0x20bfab waiting on condition [0x0000000000000000]
   java.lang.Thread.State: RUNNABLE

[4] Busy(0.0%) thread(2146220/0x20bfac) stack of java process(2146218) under user(root):
"GC task thread#0 (ParallelGC)" os_prio=0 tid=0x00007fc3c405e000 nid=0x20bfac runnable

[5] Busy(0.0%) thread(2146221/0x20bfad) stack of java process(2146218) under user(root):
"GC task thread#1 (ParallelGC)" os_prio=0 tid=0x00007fc3c4060000 nid=0x20bfad runnable

./cpu.sh: line 634: PROG: readonly variable
./cpu.sh: line 636: PROG_VERSION: readonly variable
./cpu.sh: line 643: COMMAND_LINE: readonly variable
./cpu.sh: line 649: USER: readonly variable
./cpu.sh: line 657: ec: readonly variable
./cpu.sh: line 658: eend: readonly variable
./cpu.sh: line 659: nl: readonly variable
./cpu.sh: line 829: ARGS: readonly variable
./cpu.sh: line 833: count: readonly variable
./cpu.sh: line 834: cpu_sample_interval: readonly variable
./cpu.sh: line 843: pid_list: readonly variable
./cpu.sh: line 846: update_delay: readonly variable
./cpu.sh: line 849: update_count: readonly variable
./cpu.sh: line 854: pid_list: readonly variable
./cpu.sh: line 923: run_timestamp: readonly variable
./cpu.sh: line 925: uuid: readonly variable
./cpu.sh: line 927: tmp_store_dir: readonly variable
./cpu.sh: line 931: store_file_prefix: readonly variable
./cpu.sh: line 948: ps_process_select_options: readonly variable
[1] Busy(99.9%) thread(2146230/0x20bfb6) stack of java process(2146218) under user(root):
"Thread-0" #8 prio=5 os_prio=0 tid=0x00007fc3c420c000 nid=0x20bfb6 runnable [0x00007fc3ad62a000]
   java.lang.Thread.State: RUNNABLE
        at CPULoad.lambda$main$0(CPULoad.java:8)
        at CPULoad$$Lambda$1/471910020.run(Unknown Source)
        at java.lang.Thread.run(Thread.java:750)

[2] Busy(0.0%) thread(2146218/0x20bfaa) stack of java process(2146218) under user(root):

[3] Busy(0.0%) thread(2146219/0x20bfab) stack of java process(2146218) under user(root):
"DestroyJavaVM" #9 prio=5 os_prio=0 tid=0x00007fc3c404b800 nid=0x20bfab waiting on condition [0x0000000000000000]
   java.lang.Thread.State: RUNNABLE

[4] Busy(0.0%) thread(2146220/0x20bfac) stack of java process(2146218) under user(root):
"GC task thread#0 (ParallelGC)" os_prio=0 tid=0x00007fc3c405e000 nid=0x20bfac runnable

[5] Busy(0.0%) thread(2146221/0x20bfad) stack of java process(2146218) under user(root):
"GC task thread#1 (ParallelGC)" os_prio=0 tid=0x00007fc3c4060000 nid=0x20bfad runnable

[root@hecs-76151 ~]# ./cpu.sh
[1] Busy(99.9%) thread(2146230/0x20bfb6) stack of java process(2146218) under user(root):
"Thread-0" #8 prio=5 os_prio=0 tid=0x00007fc3c420c000 nid=0x20bfb6 runnable [0x00007fc3ad62a000]
   java.lang.Thread.State: RUNNABLE
        at CPULoad.lambda$main$0(CPULoad.java:8)
        at CPULoad$$Lambda$1/471910020.run(Unknown Source)
        at java.lang.Thread.run(Thread.java:750)

[2] Busy(0.0%) thread(1155/0x483) stack of java process(681) under user(root):
"GC task thread#0 (ParallelGC)" os_prio=0 tid=0x00007fed0401f800 nid=0x483 runnable

[3] Busy(0.0%) thread(1157/0x485) stack of java process(681) under user(root):
"GC task thread#1 (ParallelGC)" os_prio=0 tid=0x00007fed04021800 nid=0x485 runnable

[4] Busy(0.0%) thread(1158/0x486) stack of java process(681) under user(root):

[5] Busy(0.0%) thread(1159/0x487) stack of java process(681) under user(root):

./cpu.sh: line 634: PROG: readonly variable
./cpu.sh: line 636: PROG_VERSION: readonly variable
./cpu.sh: line 643: COMMAND_LINE: readonly variable
./cpu.sh: line 649: USER: readonly variable
./cpu.sh: line 657: ec: readonly variable
./cpu.sh: line 658: eend: readonly variable
./cpu.sh: line 659: nl: readonly variable
./cpu.sh: line 829: ARGS: readonly variable
./cpu.sh: line 833: count: readonly variable
./cpu.sh: line 834: cpu_sample_interval: readonly variable
./cpu.sh: line 894: update_delay: readonly variable
./cpu.sh: line 897: update_count: readonly variable
./cpu.sh: line 973: run_timestamp: readonly variable
./cpu.sh: line 975: uuid: readonly variable
./cpu.sh: line 977: tmp_store_dir: readonly variable
./cpu.sh: line 981: store_file_prefix: readonly variable
./cpu.sh: line 1000: ps_process_select_options: readonly variable
[1] Busy(98.0%) thread(2146230/0x20bfb6) stack of java process(2146218) under user(root):
"Thread-0" #8 prio=5 os_prio=0 tid=0x00007fc3c420c000 nid=0x20bfb6 runnable [0x00007fc3ad62a000]
   java.lang.Thread.State: RUNNABLE
        at CPULoad.lambda$main$0(CPULoad.java:8)
        at CPULoad$$Lambda$1/471910020.run(Unknown Source)
        at java.lang.Thread.run(Thread.java:750)

[2] Busy(0.0%) thread(1155/0x483) stack of java process(681) under user(root):
"GC task thread#0 (ParallelGC)" os_prio=0 tid=0x00007fed0401f800 nid=0x483 runnable

[3] Busy(0.0%) thread(1157/0x485) stack of java process(681) under user(root):
"GC task thread#1 (ParallelGC)" os_prio=0 tid=0x00007fed04021800 nid=0x485 runnable

[4] Busy(0.0%) thread(1158/0x486) stack of java process(681) under user(root):

[5] Busy(0.0%) thread(1159/0x487) stack of java process(681) under user(root):