深入理解stressapptest

---

??团队博客: 汽车电子社区

---

一、概述

??stressapptest是一款免费的开源软件，支持用于测试Linux系统的稳定性和可靠性。它可以利用多个CPU/Core，甚至可以访问主内存的所有部分来测试系统的稳定性，以此推断系统是否存在内存泄漏或其他性能方面的问题。

??stressapptest的主要目标是发现难以重现的系统稳定性问题和其他体现在Java虚拟机、KVM Hypervisor和Hadoop集群等普通用例中的Linux系统中的性能问题。它还包括多个测试模式和选项以支持不同测试场景，比如随机测试、循环测试、全内存测试、持续测试等。

二、安装

2.1、源码编译安装

// 1.下载源码
git clone https://github.com/stressapptest/stressapptest.git

cd stressapptest
./configure
make
sudo make install

2.2、命令行安装

??Ubuntu下执行如下命令进行安装：

sudo apt-get install stressapptest

2.3、安装确认

??执行如下命令来确认stressapptest 是否安装成功：

stressapptest -h

lyb@ubuntu:~/learn$ stressapptest -h
Stats: SAT revision 1.0.6_autoconf, 64 bit binary
Log: buildd @ lgw01-amd64-022 on Thu Apr  5 10:28:35 UTC 2018 from open source release
Usage: ./sat(32|64) [options]
 -M mbytes        megabytes of ram to test
 -H mbytes        minimum megabytes of hugepages to require
 -s seconds       number of seconds to run
 -m threads       number of memory copy threads to run
 -i threads       number of memory invert threads to run
 -C threads       number of memory CPU stress threads to run
 --findfiles      find locations to do disk IO automatically
 -d device        add a direct write disk thread with block device (or file) 'device'
 -f filename      add a disk thread with tempfile 'filename'
 -l logfile       log output to file 'logfile'
 --max_errors n   exit early after finding 'n' errors
 -v level         verbosity (0-20), default is 8
 -W               Use more CPU-stressful memory copy
 -A               run in degraded mode on incompatible systems
 -p pagesize      size in bytes of memory chunks
 --filesize size  size of disk IO tempfiles
 -n ipaddr        add a network thread connecting to system at 'ipaddr'
 --listen         run a thread to listen for and respond to network threads.
 --no_errors      run without checking for ECC or other errors
 --force_errors   inject false errors to test error handling
 --force_errors_like_crazy   inject a lot of false errors to test error handling
 -F               don't result check each transaction
 --stop_on_errors  Stop after finding the first error.
 --read-block-size     size of block for reading (-d)
 --write-block-size    size of block for writing (-d). If not defined, the size of block for writing will be defined as the size of block for reading
 --segment-size   size of segments to split disk into (-d)
 --cache-size     size of disk cache (-d)
 --blocks-per-segment  number of blocks to read/write per segment per iteration (-d)
 --read-threshold      maximum time (in us) a block read should take (-d)
 --write-threshold     maximum time (in us) a block write should take (-d)
 --random-threads      number of random threads for each disk write thread (-d)
 --destructive    write/wipe disk partition (-d)
 --monitor_mode   only do ECC error polling, no stress load.
 --cc_test        do the cache coherency testing
 --cc_inc_count   number of times to increment the cacheline's member
 --cc_line_count  number of cache line sized datastructures to allocate for the cache coherency threads to operate
 --paddr_base     allocate memory starting from this address
 --pause_delay    delay (in seconds) between power spikes
 --pause_duration duration (in seconds) of each pause
 --local_numa     choose memory regions associated with each CPU to be tested by that CPU
 --remote_numa    choose memory regions not associated with each CPU to be tested by that CPU
 --interleave_size bytes  size in bytes of each channel's data as interleaved between memory channels
 --channel_width bits     width in bits of each memory channel
 --memory_channel u1,u2   defines a comma-separated list of names
                          for dram packages in a memory channel.
                          Use multiple times to define multiple channels.
lyb@ubuntu:~/learn$ 

三、重要参数详解

3.1、查询支持的参数

??执行如下命令来确认stressapptest支持的参数：

stressapptest -h

lyb@ubuntu:~/learn$ stressapptest -h
Stats: SAT revision 1.0.6_autoconf, 64 bit binary
Log: buildd @ lgw01-amd64-022 on Thu Apr  5 10:28:35 UTC 2018 from open source release
Usage: ./sat(32|64) [options]
 -M mbytes        megabytes of ram to test
 -H mbytes        minimum megabytes of hugepages to require
 -s seconds       number of seconds to run
 -m threads       number of memory copy threads to run
 -i threads       number of memory invert threads to run
 -C threads       number of memory CPU stress threads to run
 --findfiles      find locations to do disk IO automatically
 -d device        add a direct write disk thread with block device (or file) 'device'
 -f filename      add a disk thread with tempfile 'filename'
 -l logfile       log output to file 'logfile'
 --max_errors n   exit early after finding 'n' errors
 -v level         verbosity (0-20), default is 8
 -W               Use more CPU-stressful memory copy
 -A               run in degraded mode on incompatible systems
 -p pagesize      size in bytes of memory chunks
 --filesize size  size of disk IO tempfiles
 -n ipaddr        add a network thread connecting to system at 'ipaddr'
 --listen         run a thread to listen for and respond to network threads.
 --no_errors      run without checking for ECC or other errors
 --force_errors   inject false errors to test error handling
 --force_errors_like_crazy   inject a lot of false errors to test error handling
 -F               don't result check each transaction
 --stop_on_errors  Stop after finding the first error.
 --read-block-size     size of block for reading (-d)
 --write-block-size    size of block for writing (-d). If not defined, the size of block for writing will be defined as the size of block for reading
 --segment-size   size of segments to split disk into (-d)
 --cache-size     size of disk cache (-d)
 --blocks-per-segment  number of blocks to read/write per segment per iteration (-d)
 --read-threshold      maximum time (in us) a block read should take (-d)
 --write-threshold     maximum time (in us) a block write should take (-d)
 --random-threads      number of random threads for each disk write thread (-d)
 --destructive    write/wipe disk partition (-d)
 --monitor_mode   only do ECC error polling, no stress load.
 --cc_test        do the cache coherency testing
 --cc_inc_count   number of times to increment the cacheline's member
 --cc_line_count  number of cache line sized datastructures to allocate for the cache coherency threads to operate
 --paddr_base     allocate memory starting from this address
 --pause_delay    delay (in seconds) between power spikes
 --pause_duration duration (in seconds) of each pause
 --local_numa     choose memory regions associated with each CPU to be tested by that CPU
 --remote_numa    choose memory regions not associated with each CPU to be tested by that CPU
 --interleave_size bytes  size in bytes of each channel's data as interleaved between memory channels
 --channel_width bits     width in bits of each memory channel
 --memory_channel u1,u2   defines a comma-separated list of names
                          for dram packages in a memory channel.
                          Use multiple times to define multiple channels.
lyb@ubuntu:~/learn$ 

3.2、参数说明

??选项说明：
????-s或–timeout: 指定测试运行的总时间，以秒为单位。默认值为0（无限制）。
????-M或–vm-bytes: 指定应用程序所需的内存量，以字节、千字节、兆字节或吉字节为单位。例如，“-M 2048m”表示分配2GB的内存。默认值为0（自动计算）。
????-m或–vm-pattern: 指定要使用的内存模式，如线性、随机、步进等。默认值为“linear”（线性）。
????-c或–cpu-cores: 指定要使用的CPU核心数。例如，“-c 4”将在四个核心上运行测试。默认值为0（使用所有可用核心）。
????–affinity: 将应用程序绑定到指定的CPU核心上，以避免跨核心调度带来的额外开销。例如，“–affinity 0,2,4”将绑定应用程序到第1、3和5个CPU核心上。
????–matrix: 启用矩阵测试模式，该模式可以对整个系统进行更全面的测试，并且需要更长时间才能完成。
????–quiet: 禁止输出详细信息和统计数据，只显示结果摘要信息。
????–log-level: 设置日志级别，以控制输出详细程度。可选值为”debug”、”info”、”warning”、”error”和”critical”。
????–log-file: 将日志输出到指定的文件中，而不是标准输出。
????–version: 显示stressapptest版本信息并退出。

四、实例

4.1、随机测试（默认模式）

??随机模式是stressapptest的默认模式，它会随机访问系统内存中的所有部分来测试系统的稳定性。可以通过以下命令启动：

stressapptest

4.2、循环测试

??循环模式是stressapptest的另一种模式，它会在指定的时间内不断迭代相同的内存模式。可以使用以下命令启动：

stressapptest -M 100 -s 128 -C 16 -c 8

4.2、全内存测试

??全内存模式会尝试占用所有可用的空闲内存。可以使用以下命令启动：