Centos 6下通过blktrace监控IO
1. blktrace安装
#yum install blktrace#mount a /sys/kernel/debug -t debugfs2. blktrace使用
1.通过blkiomon，可以告诉你IO size,完成花费时间,吞吐量, 还可以统计出这次关键参数的分布
#blktrace /dev/sde -a issue -a complete -w 3600 -o - | blkiomon -I 10 -h -time: Fri Sep 13 09:50:23 2013
device: 8,64
sizes read (bytes): num 0, min -1, max 0, sum 0, squ 0, avg -nan, var -nan
sizes write (bytes): num 156, min 4096, max 524288, sum 80764928, squ 42331633876992, avg 517723.9, var 3318593440.4
d2c read (usec): num 0, min -1, max 0, sum 0, squ 0, avg -nan, var -nan
d2c write (usec): num 156, min 127, max 84010, sum 6410332, squ 330367299862, avg 41091.9, var 429197174.1
throughput read (bytes/msec): num 0, min -1, max 0, sum 0, squ 0, avg -nan, var -nan
throughput write (bytes/msec): num 156, min 6240, max 189068, sum 3417217, squ 164794315173, avg 21905.2, var 576534399.3
sizes histogram (bytes):
0: 0 1024: 0 2048: 0 4096: 1
8192: 0 16384: 0 32768: 1 65536: 0
131072: 0 262144: 0 524288: 154 1048576: 0
2097152: 0 4194304: 0 8388608: 0 > 8388608: 0
d2c histogram (usec):
0: 0 8: 0 16: 0 32: 0
64: 0 128: 1 256: 0 512: 1
1024: 0 2048: 0 4096: 1 8192: 7
16384: 24 32768: 12 65536: 102 131072: 8
262144: 0 524288: 0 1048576: 0 2097152: 0
4194304: 0 8388608: 0 16777216: 0 33554432: 0
>33554432: 0
bidirectional requests: 0
sizes write(bytes):
num：表示写IO次数
min：最小Write IO 4096Bytes
Max：最大Write IO 524288Bytes
sum：总共Write IO 80764928Bytes
avg：平均IO 517723Bytes
d2c write(usec)
num：表示写IO次数
min：最小IO响应时间127微秒
max：最大io响应时间84010微秒
sum：IO响应总时间6410332微秒
avg：平均IO响应时间41091微秒
d2c histogram (usec)柱状图说明：
响应时间128微秒以内的IO 1次
响应时间512微秒以内的IO 1次
响应时间4096微秒以内的IO 1次
响应时间8192微秒以内的IO 7次
响应时间16384微秒以内的IO 24次
响应时间32768微秒以内的IO 12次
响应时间65536微秒以内的IO 102次
响应时间131072微秒以内的IO 8次
blkparse使用
# blktrace /dev/sde -o - | blkparse -i - 8,64 0 1 0.000000000 2480 Q W 11853824 + 1024 [dd]
8,64 0 2 0.000011603 2480 G W 11853824 + 1024 [dd]
8,64 0 3 0.000016037 2480 P N [dd]
8,64 0 4 0.000092645 2480 Q W 11854848 + 1024 [dd]
8,64 0 5 0.000095456 2480 G W 11854848 + 1024 [dd]
8,64 0 6 0.000102259 2480 I W 11853824 + 1024 [dd]
8,64 0 0 0.000108432 0 m N cfq34A / insert_request
8,64 0 0 0.000110791 0 m N cfq34A / add_to_rr
8,64 0 7 0.000113614 2480 I W 11854848 + 1024 [dd]
8,64 0 0 0.000114744 0 m N cfq34A / insert_request
8,64 0 8 0.000115696 2480 U N [dd] 2
8,64 0 0 0.000118660 0 m N cfq workload slice:60
8,64 0 0 0.000120308 0 m N cfq34A / set_active wl_class:0 wl_type:0
8,64 0 0 0.000123098 0 m N cfq34A / fifo= (null)
8,64 0 0 0.000124405 0 m N cfq34A / dispatch_insert
8,64 0 0 0.000125979 0 m N cfq34A / dispatched a request
8,64 0 0 0.000127375 0 m N cfq34A / activate rq, drv=1
8,64 0 9 0.000127803 2480 D W 11853824 + 1024 [dd]
8,64 0 0 0.000153214 0 m N cfq34A / dispatch_insert
8,64 0 0 0.000154664 0 m N cfq34A / dispatched a request
8,64 0 0 0.000155569 0 m N cfq34A / activate rq, drv=2
8,64 0 10 0.000155730 2480 D W 11854848 + 1024 [dd]
8,64 0 0 0.002036492 0 m N cfq34A / slice expired t=0
8,64 0 0 0.002039138 0 m N / served: vt=444186624 min_vt=444178432
8,64 0 0 0.002041404 0 m N cfq34A / sl_used=2 disp=2 charge=2 iops=1 sect=2048
8,64 0 0 0.002042463 0 m N cfq34A / del_from_rr
8,64 0 0 0.002043051 0 m N / del_from_rr group
8,64 0 11 0.002073197 2436 C W 11853824 + 1024 [0]
8,64 0 0 0.002125838 0 m N cfq34A / complete rqnoidle 0
8,64 0 12 0.003306086 2288 C W 11854848 + 1024 [0]
8,64 0 0 0.003358069 0 m N cfq34A / complete rqnoidle 0
CPU0 (8,64):
Reads Queued: 0, 0KiB Writes Queued: 2, 1,024KiB
Read Dispatches: 0, 0KiB Write Dispatches: 2, 1,024KiB
Reads Requeued: 0 Writes Requeued: 0
Reads Completed: 0, 0KiB Writes Completed: 2, 1,024KiB
Read Merges: 0, 0KiB Write Merges: 0, 0KiB
Read depth: 0 Write depth: 2
IO unplugs: 1 Timer unplugs: 0
Throughput (R/W): 0KiB/s / 341,333KiB/s
Events (8,64): 31 entries
Skips: 6 forward (51 - 62.2%)
blktrace是个很强大的IO监控工具，还往后的工作中会经常用到，需要更加详细的了解应用。