今天晚上发现服务器io有点高,顺带看看哪些进程在读写磁盘。
1. lsof
用 lsof 查看某块磁盘上的读写进程。
gddg:~ # lsof /dev/xvda2 |head COMMAND PID USER FD TYPE DEVICE SIZE/OFF NODE NAME init 1 root cwd DIR 202,2 4096 2 / init 1 root rtd DIR 202,2 4096 2 / init 1 root txt REG 202,2 40784 193218 /sbin/init init 1 root mem REG 202,2 19114 8063 /lib64/libdl-2.11.1.so init 1 root mem REG 202,2 1661454 8057 /lib64/libc-2.11.1.so init 1 root mem REG 202,2 236384 8114 /lib64/libsepol.so.1 init 1 root mem REG 202,2 113904 8115 /lib64/libselinux.so.1 init 1 root mem REG 202,2 149797 8050 /lib64/ld-2.11.1.so kthreadd 2 root cwd DIR 202,2 4096 2 /
然后可以通过 lsof -p $pid 查看详情
gddg:~ # lsof -p 32597 COMMAND PID USER FD TYPE DEVICE SIZE/OFF NODE NAME bash 32597 root cwd DIR 202,2 4096 16097 /root bash 32597 root rtd DIR 202,2 4096 2 / bash 32597 root txt REG 202,2 584016 32203 /bin/bash bash 32597 root mem REG 202,2 293936 8125 /lib64/libncurses.so.5.6 bash 32597 root mem REG 202,2 1661454 8057 /lib64/libc-2.11.1.so bash 32597 root mem REG 202,2 19114 8063 /lib64/libdl-2.11.1.so bash 32597 root mem REG 202,2 263568 8153 /lib64/libreadline.so.5.2 bash 32597 root mem REG 202,2 149797 8050 /lib64/ld-2.11.1.so bash 32597 root mem REG 202,2 217016 16498 /var/run/nscd/passwd bash 32597 root mem REG 202,2 256324 149503 /usr/lib/locale/en_US.utf8/LC_CTYPE bash 32597 root mem REG 202,2 54 149490 /usr/lib/locale/en_US.utf8/LC_NUMERIC bash 32597 root mem REG 202,2 2454 133112 /usr/lib/locale/en_US.utf8/LC_TIME bash 32597 root mem REG 202,2 1163682 149504 /usr/lib/locale/en_US.utf8/LC_COLLATE bash 32597 root mem REG 202,2 286 133111 /usr/lib/locale/en_US.utf8/LC_MONETARY bash 32597 root mem REG 202,2 57 149408 /usr/lib/locale/en_US.utf8/LC_MESSAGES/SYS_LC_MESSAGES bash 32597 root mem REG 202,2 34 149428 /usr/lib/locale/en_US.utf8/LC_PAPER bash 32597 root mem REG 202,2 77 149438 /usr/lib/locale/en_US.utf8/LC_NAME bash 32597 root mem REG 202,2 155 133108 /usr/lib/locale/en_US.utf8/LC_ADDRESS bash 32597 root mem REG 202,2 59 149407 /usr/lib/locale/en_US.utf8/LC_TELEPHONE bash 32597 root mem REG 202,2 23 149429 /usr/lib/locale/en_US.utf8/LC_MEASUREMENT bash 32597 root mem REG 202,2 26050 149293 /usr/lib64/gconv/gconv-modules.cache
2. cat /proc/$pid/io
如果内核版本大于2.6.20,通过cat /proc/pid/io 便可以获取进程的io信息。详细解释
gddg:~ # cat /proc/4140/io rchar: 197448798054 // 读出的总字节数,read()或者pread()中的长度参数总和(pagecache中统计而来,不代表实际磁盘的读入) wchar: 209896059897 // 写入的总字节数,write()或者pwrite()中的长度参数总和 syscr: 6491904 // read()或者pread()总的调用次数 syscw: 13633940 // write()或者pwrite()总的调用次数 read_bytes: 49616125952 // 实际从磁盘中读取的字节总数 write_bytes: 14038130688 // 实际写入到磁盘中的字节总数 cancelled_write_bytes: 2473984 // 由于截断pagecache导致应该发生而没有发生的写入字节数
3. block_dump
通过echo 1 > /proc/sys/vm/block_dump ,来把 block 读写(WRITE/READ/DIRTY)状况 dump 到日志里,通过 dmesg 命令来查看
#!/bin/sh /etc/init.d/syslog stop echo 1 > /proc/sys/vm/block_dump sleep 60 dmesg | awk '/(READ|WRITE|dirtied)/ {process[$1]++} END {for (x in process) \ print process[x],x}' |sort -nr |awk '{print $2 " " $1}' | \ head -n 10
#dmesg | egrep "READ|WRITE|dirtied" | egrep -o '([a-zA-Z]*)' | sort | uniq -c | sort -rn | head echo 0 > /proc/sys/vm/block_dump /etc/init.d/syslog start
输出如下:
pdflush(10423): 4000 Nginx(1167): 179 Nginx(1229): 172 Nginx(1187): 111 Nginx(1243): 105 Nginx(1213): 92 Nginx(1233): 69 Nginx(1157): 61 Nginx(1161): 50 Nginx(1155): 32
或者参考这段代码吧:
http://stackoverflow.com/questions/249570/how-can-i-record-what-process-or-kernel-activity-is-using-the-disk-in-gnu-linux
sudo -s dmesg -c /etc/init.d/klogd stop echo 1 > /proc/sys/vm/block_dump rm /tmp/disklog watch "dmesg -c >> /tmp/disklog" CTRL-C when you're done collecting data echo 0 > /proc/sys/vm/block_dump /etc/init.d/klogd start exit (quit root shell) cat /tmp/disklog | awk -F"[() \t]" '/(READ|WRITE|dirtied)/ {activity[$1]++} END {for (x in activity) print x, activity[x]}'| sort -nr -k2
4. iotop类脚本
python版
#!/usr/bin/python # Monitoring per-process disk I/O activity # written by http://www.vpsee.com import sys, os, time, signal, re class diskIO: def __init__(self, pname=None, pid=None, reads=0, writes=0): self.pname = pname self.pid = pid self.reads = 0 self.writes = 0 def main(): argc = len(sys.argv) if argc != 1: print "usage: ./iotop" sys.exit(0) if os.getuid() != 0: print "must be run as root" sys.exit(0) signal.signal(signal.SIGINT, signal_handler) os.system('echo 1 > /proc/sys/vm/block_dump') print "TASK PID READ WRITE" while True: os.system('dmesg -c > /tmp/diskio.log') l = [] f = open('/tmp/diskio.log', 'r') line = f.readline() while line: m = re.match(\ '^(\S+)\((\d+)\): (READ|WRITE) block (\d+) on (\S+)', line) if m != None: if not l: l.append(diskIO(m.group(1), m.group(2))) line = f.readline() continue found = False for item in l: if item.pid == m.group(2): found = True if m.group(3) == "READ": item.reads = item.reads + 1 elif m.group(3) == "WRITE": item.writes = item.writes + 1 if not found: l.append(diskIO(m.group(1), m.group(2))) line = f.readline() time.sleep(1) for item in l: print "%-10s %10s %10d %10d" % \ (item.pname, item.pid, item.reads, item.writes) def signal_handler(signal, frame): os.system('echo 0 > /proc/sys/vm/block_dump') sys.exit(0) if __name__=="__main__": main()
perl版,强烈推荐打开下面链接
http://www.xaprb.com/blog/2009/08/23/how-to-find-per-process-io-statistics-on-linux/
#!/usr/bin/env perl # This program is part of Aspersa (http://code.google.com/p/aspersa/) =pod =head1 NAME iodump - Compute per-PID I/O stats for Linux when iotop/pidstat/iopp are not available. =head1 SYnopSIS Prepare the system: dmesg -c /etc/init.d/klogd stop echo 1 > /proc/sys/vm/block_dump Start the reporting: while true; do sleep 1; dmesg -c; done | perl iodump CTRL-C Stop the system from dumping these messages: echo 0 > /proc/sys/vm/block_dump /etc/init.d/klogd start =head1 AUTHOR Baron Schwartz =cut use strict; use warnings FATAL => 'all'; use English qw(-no_match_vars); use sigtrap qw(handler finish untrapped normal-signals); my %tasks; my $oktorun = 1; my $line; while ( $oktorun && (defined ($line = <>)) ) { my ( $task, $pid, $activity, $where, $device ); ( $task, $pid, $activity, $where, $device ) = $line =~ m/(\S+)\((\d+)\): (READ|WRITE) block (\d+) on (\S+)/; if ( !$task ) { ( $task, $pid, $activity, $where, $device ) = $line =~ m/(\S+)\((\d+)\): (dirtied) inode \(.*?\) (\d+) on (\S+)/; } if ( $task ) { my $s = $tasks{$pid} ||= { pid => $pid, task => $task }; ++$s->{lc $activity}; ++$s->{activity}; ++$s->{devices}->{$device}; } } printf("%-15s %10s %10s %10s %10s %10s %s\n", qw(TASK PID TOTAL READ WRITE DIRTY DEVICES)); foreach my $task ( reverse sort { $a->{activity} <=> $b->{activity} } values %tasks ) { printf("%-15s %10d %10d %10d %10d %10d %s\n", $task->{task}, $task->{pid}, ($task->{'activity'} || 0), ($task->{'read'} || 0), ($task->{'write'} || 0), ($task->{'dirty'} || 0), join(', ', keys %{$task->{devices}})); } sub finish { my ( $signal ) = @_; if ( $oktorun ) { print STDERR "# Caught SIG$signal.\n"; $oktorun = 0; } else { print STDERR "# Exiting on SIG$signal.\n"; exit(1); } }
以后有时间,得研究下iostat.c
http://code.google.com/p/tester-higkoo/source/browse/trunk/Tools/iostat/iostat.c
参考:
1. http://www.vpsee.com/2009/08/monitor-process-io-activity/
2. http://www.vpsee.com/2009/08/monitor-process-io-activity/
版权声明:本文内容由互联网用户自发贡献,该文观点与技术仅代表作者本人。本站仅提供信息存储空间服务,不拥有所有权,不承担相关法律责任。如发现本站有涉嫌侵权/违法违规的内容, 请发送邮件至 dio@foxmail.com 举报,一经查实,本站将立刻删除。