[Linux] 系统资源统计工具 dstat

小灰灰 2023-07-13 13:51 31阅读 0赞

1. 默认统计指标

dstat 不带参数，默认参数参考下例，连续打印，1条记录/s  
dstat 3 5 输出时间间隔为3秒钟,总共输出5次(实际会有6行，第一行是针对之前的数据统计),每行记录在3s内会不断刷新

[root@xxx bin]# dstat 3 5
    ----total-cpu-usage---- -dsk/total- -net/total- ---paging-- ---system--
    usr sys idl wai hiq siq| read  writ| recv  send|  in   out | int   csw
      0   0 100   0   0   0|  70B 2299B|   0     0 |   0     0 |  25    12
      0   0 100   0   0   0|   0     0 | 447B  113B|   0     0 |  24    14
      0   0 100   0   0   0|   0     0 | 646B  113B|   0     0 |  27    15
      0   0 100   0   0   0|   0  2731B| 539B  113B|   0     0 |  28    17
      0   0 100   0   0   0|   0     0 | 755B  107B|   0     0 |  27    18
      0   0 100   0   0   0|   0     0 | 554B  107B|   0     0 |  24    14
    [root@xxx bin]#

2.显示其它统计指标

\-l ：显示负载统计  
\-m ：显示内存使用  
\-r ：显示I/O统计  
\-s ：显示交换分区使用  
\-t ：第一列显示当前实际  
–-tcp ：显示常用的TCP统计  
–-socket ：显示网络统计数据  
\-–freespace ：显示当前磁盘空间使用率  
\-–proc-count ：显示正在运行的程序数量

[root@xxx bin]# dstat -m 3 2
    ------memory-usage-----
     used  buff  cach  free
     304M  176M 2213M 12.9G
     304M  176M 2213M 12.9G
     304M  176M 2213M 12.9G
    [root@xxx bin]# dstat -l 3 2
    ---load-avg---
     1m   5m  15m
       0    0    0
       0    0    0
       0    0    0
    [root@xxx bin]# dstat -r 3 2
    --io/total-
     read  writ
    0.00  0.22
       0     0
       0     0
    [root@xxx bin]# dstat -s 3 2
    ----swap---
     used  free
       0    16G
       0    16G
       0    16G
    [root@xxx bin]# dstat -t
    ----system----
      date/time
    12-03 16:19:51
    12-03 16:19:52
    12-03 16:19:53
    12-03 16:19:54
    12-03 16:19:55^C
    [root@xxx bin]# 
    [root@xxx bin]# dstat -t -c -m
    ----system---- ----total-cpu-usage---- ------memory-usage-----
      date/time   |usr sys idl wai hiq siq| used  buff  cach  free
    12-03 16:20:28|  0   0 100   0   0   0| 304M  176M 2213M 12.9G
    12-03 16:20:29|  0   0 100   0   0   0| 304M  176M 2213M 12.9G
    12-03 16:20:30|  0   0 100   0   0   0| 304M  176M 2213M 12.9G^C
    [root@xxx bin]#

3.将结果输写入csv文件

dstat –-output <csv file path> 输出结果写到一个csv格式

[root@xxx bin]# dstat --output test22.csv --tcp
    ----tcp-sockets----
    lis act syn tim clo
     11   8   0   0   0
     11   8   0   0   0
     11   8   0   0   0
     11   8   0   0   0^C
    [root@xxx bin]# cat test22.csv
    "Dstat 0.7.0 CSV output"
    "Author:","Dag Wieers <dag@wieers.com>",,,,"URL:","http://dag.wieers.com/home-made/dstat/"
    "Host:","xxx",,,,"User:","root"
    "Cmdline:","dstat --output test22.csv --tcp",,,,"Date:","12 Mar 2020 16:39:01 CST"
    
    "tcp sockets",,,,
    "lis","act","syn","tim","clo"
    11.0,8.0,0.0,0.0,0.0
    11.0,8.0,0.0,0.0,0.0
    11.0,8.0,0.0,0.0,0.0
    11.0,8.0,0.0,0.0,0.0
    [root@xxx bin]#

*更多选项参考 man dstat*  
\--top-bio  
       show most expensive block I/O process  
\--top-cpu  
       show most expensive CPU process  
\--top-cputime  
       show process using the most CPU time (in ms)  
\--top-cputime-avg  
       show process with the highest average timeslice (in ms)  
\--top-io  
       show most expensive I/O process  
\--top-latency  
       show process with highest total latency (in ms)  
\--top-latency-avg  
       show process with the highest average latency (in ms)  
\--top-mem  
       show process using the most memory  
\--top-oom  
       show process that will be killed by OOM the first

*参考资料：*Linux终端下dstat监控工具 https://linux.cn/article-3215-1.html